Research on CBM of the Intelligent Substation SCADA System

transcript

energies

Article

Research on CBM of the Intelligent SubstationSCADA System

Jyh-Cherng Gu 1, Chun-Hung Liu 1 , Kai-Ying Chou 1 and Ming-Ta Yang 2,*1 Department of Electrical Engineering, National Taiwan University of Science and Technology, 43, Sec. 3,

Keelung Road, Taipei 10607, Taiwan; jcgu@mail.ntust.edu.tw (J.-C.G.);d10207102@mail.ntust.edu.tw (C.-H.L.); m10407127@mail.ntust.edu.tw (K.-Y.C.)

2 Department of Electrical Engineering, St. John’s University, 499, Sec. 4, Tam King Road, Tamsui District,New Taipei City 25135, Taiwan

* Correspondence: mtyang@mail.sju.edu.tw

Received: 22 August 2019; Accepted: 14 October 2019; Published: 15 October 2019��

Abstract: An equipment status management and maintenance platform of an intelligent substationmonitoring and control system is built in the Tai-Tam substation of the Taipower company. The real-time operating status of the equipment, such as the server, supervisory control and data acquisition(SCADA) human–machine interface (HMI) software, switches, intelligent electronic devices (IEDs),merging units (MUs), as well as the entire SCADA system, are evaluated comprehensively. First,the status information of all equipment is collected, and the theory of relative deterioration degree(RDD) and fuzzy theory (FT) are applied to calculate the fuzzy evaluation matrix of the equipmentinfluencing factors. Then, the subjective analytic hierarchy process (AHP) and the objective entropymethod for weighting are combined to calculate the comprehensive weights of the equipmentinfluencing factors. Finally, the result of the equipment status evaluation is obtained using the fuzzycomprehensive evaluation (FCE) method and is presented at the equipment status management andmaintenance platform. Such equipment status evaluation results can be used by the inspection andmaintenance personnel to determine the priority for equipment maintenance and repair. The result ofthis study may serve as a valuable reference to utility companies when making maintenance plans.

Keywords: IEC 61850; intelligent substation; SCADA; condition-based maintenance (CBM); fuzzycomprehensive evaluation (FCE); comprehensive weighting method

1. Introduction

Condition-based maintenance (CBM) is a preventive maintenance (PM) policy, which usesinformation and communication technologies to monitor equipment status and to inspect and test thedegree of interior deterioration of equipment. Based on equipment reliability, and with the equipmentoperating status and other importance factors taken into consideration, CBM is more target-specific,rational, and scientific when making equipment inspection, maintenance, and repair decisions [1,2].

Maity et al. point out that time-based maintenance (TBM) may easily lead to over-maintenance orunder-maintenance of electrical equipment. If equipment status information can be taken real-time andon-line by sensors and be provided to back-end platforms, data can be analyzed and actual equipmentoperating condition can be assessed. Thus, a maintenance plan based on the deterioration degreeof individual equipment can be made, and proper maintenance can be achieved [3]. Unnecessarymaintenance is one of the major causes of power grid failures. To avoid this from happening, Ohlen builta power system monitor and control platform for on-site measurement and on-line real-time monitoringin order to provide data needed by CBM. The goal is to achieve best balance between operation andmaintenance based on field equipment status information [4]. The working environment, operation

Energies 2019, 12, 3892; doi:10.3390/en12203892 www.mdpi.com/journal/energies

Energies 2019, 12, 3892 2 of 22

data, inspection and maintenance records, and conditions of internal parts of power transformersare included in a multistage hierarchical assessment index system. The relative deterioration degree(RDD) method is used to assess the degree of similarity between actual equipment operating statusand fault condition, and the analytic hierarchy process (AHP) method and the fuzzy comprehensiveevaluation (FCE) method are used to evaluate the health status of power transformers. The result mayserve as the basis of preventive maintenance practices [5]. A circuit breaker (CB) maintenance researchfocused on reliability is proposed which integrates intelligent electronic devices (IEDs) in an intelligentmaintenance model. IEDs are used to build a CB status monitoring system and to collect relevantmaintenance information. The fuzzy set theory, the AHP method, and the Dempster–Shafer evidencetheory are applied to analyze and evaluate CB status. In addition, important influencing factors ofCBs are collected, and the Delphi method is applied to calculate the weights of the importance factorsand to analyze and evaluate the importance of CBs. Finally, reliability-focused maintenance decisionanalysis is performed based on the result of CB status and importance evaluation [6]. The multistageFCE method is applied to evaluate the operating status of the protection system of a power grid.A multistage status index model is built based on equipment status data, actual system operating status,and the experience of operation personnel and experts. The output of the model is the comprehensiveevaluation result. Finally, a maintenance strategy, including status evaluation result, maintenancegrades, optimal maintenance time, and maintenance suggestions, is proposed and can be adopted bythe inspection and maintenance personnel to perform CBM on power grid protection systems [7,8].Qian et al. proposed a CBM approach for wind turbines based on long short-term memory (LSTM)algorithms to improve defect detection from supervisory control and data acquisition (SCADA).LSTM algorithms have the capability of capturing long-term dependencies hidden within a sequenceof measurements, which can be exploited to increase the prediction accuracy for CBM [9]. Someresearch works presented the discrete Markov chain model as a simplified probabilistic model fordamages in wind turbine blades. The classic Bayesian pre-posterior decision theory is applied for thedecision-making of the CBM strategy [10–12]. Tian et al. presented a transformer assessing model forCBM by employing a Cauchy membership function for fuzzy grade division, and then a fuzzy evidencefusion method was represented to handle the fuzzy evidences fusion processes. This approach canrecommend the condition-based maintenance of power transformer [13].

An intelligent substation is an important link in the realization of a smart grid and is responsiblefor electric power delivery, power dispatch, power flow, and equipment monitoring and control. As theintelligent substation evolves along with the announcement of IEC 61,850 communication standards,more equipment can be easily integrated into power monitoring and control systems, and innovativeand intelligent CBM strategies can be proposed for power equipment maintenance and replacement.There are many studies on equipment maintenance and management on the substation primary side,but few on the reliability evaluation of intelligent substation SCADA systems. The goal of this study isto improve the existing maintenance methods of substation SCADA system equipment, and to proposea condition-based equipment inspection and maintenance strategy. In this study, substation equipmentstatus information is collected following the IEC 61,850 communication protocol, equipment operatingstatus is analyzed through status inspection and testing platform in real time, and a SCADA systemmaintenance and management platform applying intelligent inspection and maintenance strategy isrealized. The result of this study may serve as a valuable reference to the inspection and maintenancepersonnel and the operation and management personnel of a power utility when making a maintenanceplan, and may help to improve the stability and reliability of the utility power supply.

The objective of this paper is to propose the CBM evaluation model of a smart substation controlsystem through a SCADA platform where the real-time health condition of substation equipment can bemonitored. The result of the evaluation can be categorized into four degrees: Good Condition, AttentionRequired, Critical Condition, as well as Immediate Inspection and Maintenance. The maintenance staff

can analyze the equipment condition and make the maintenance plan in priority order.

Energies 2019, 12, 3892 3 of 22

In this paper, an equipment status evaluation method is proposed, and a reliability evaluationmodel of an integrated intelligent substation SCADA system consisting of a server, SCADAhuman–machine interface (HMI) software, switches, IEDs, and merging units (MUs) is built inthis study. First, the relative deterioration degree (RDD) method is applied to process the relevantparameters of performance-influencing factors. Then the comprehensive weighting method is appliedto analyze the degree of importance of equipment. Finally, the fuzzy comprehensive evaluation (FCE)method is applied to perform comprehensive equipment status evaluation. The flowchart of theproposal is shown in Figure 1. The evaluation result serves as a useful guide for system maintenanceand management personnel, equipment manufacturers, and power equipment repair crews whenmaintenance is made. Eventually, an intelligent substation SCADA system example platform is builtin the Tai-Tam substation of the Taipower company. Two operation cases of this platform have beenapplied for validation.

Energies 2019, 12, x FOR PEER REVIEW 3 of 22

Maintenance. The maintenance staff can analyze the equipment condition and make the maintenance plan in priority order.

In this paper, an equipment status evaluation method is proposed, and a reliability evaluation model of an integrated intelligent substation SCADA system consisting of a server, SCADA human–machine interface (HMI) software, switches, IEDs, and merging units (MUs) is built in this study. First, the relative deterioration degree (RDD) method is applied to process the relevant parameters of performance-influencing factors. Then the comprehensive weighting method is applied to analyze the degree of importance of equipment. Finally, the fuzzy comprehensive evaluation (FCE) method is applied to perform comprehensive equipment status evaluation. The flowchart of the proposal is shown in Figure 1. The evaluation result serves as a useful guide for system maintenance and management personnel, equipment manufacturers, and power equipment repair crews when maintenance is made. Eventually, an intelligent substation SCADA system example platform is built in the Tai-Tam substation of the Taipower company. Two operation cases of this platform have been applied for validation.

Figure 1. The flowchart of the FCE method.

2. Architecture of the Proposed SCADA System of an Intelligent Substation

The example configuration of the proposed intelligent substation SCADA system is shown in Figure 2. The SCADA system of the control center is at the station level, which integrates the important information of all equipment, issues controlling commands, and performs remote monitoring. The bay level is composed mainly of IEDs, which receive the power parameters and status information of the equipment, perform protection functions such as overcurrent protection and voltage differential protection, control the tripping and reclosing of CBs, and upload all status messages and event records to the SCADA system. The merging units (MUs) are the core devices at the process level, which convert analog voltage/current signals to digital signals and transmit the

Figure 1. The flowchart of the FCE method.

2. Architecture of the Proposed SCADA System of an Intelligent Substation

The example configuration of the proposed intelligent substation SCADA system is shown inFigure 2. The SCADA system of the control center is at the station level, which integrates the importantinformation of all equipment, issues controlling commands, and performs remote monitoring. The baylevel is composed mainly of IEDs, which receive the power parameters and status information ofthe equipment, perform protection functions such as overcurrent protection and voltage differentialprotection, control the tripping and reclosing of CBs, and upload all status messages and event recordsto the SCADA system. The merging units (MUs) are the core devices at the process level, which convertanalog voltage/current signals to digital signals and transmit the signals to bay level equipment through

Energies 2019, 12, 3892 4 of 22

optical fibers. This study focuses on the status evaluation of the server, SCADA human–machineinterface (HMI) software, switches, IEDs, and MUs.

signals to bay level equipment through optical fibers. This study focuses on the status evaluation of the server, SCADA human–machine interface (HMI) software, switches, IEDs, and MUs.

Figure 2. The example configuration of the proposed intelligent substation SCADA system.

2.1. Server

The server is the hardware equipment of the intelligent substation SCADA system which is capable of running various application software. The server manages the substation automation system efficiently and ensures the power quality provided by the power grid is high. The maintenance and management personnel can monitor the substation operating status, look up both real-time and historical alarms, and print out event records on-line merely through the server of the SCADA system.

2.2. SCADA HMI Software

The SCADA HMI software is a key element in the realization of intelligent substation automatic control and logical analysis. The main functions of the SCADA HMI software include equipment data collecting, system status monitoring and testing, power flow calculation, short circuit analysis, power grid energy dispatch and management, sending control commands, and security control. Statistical data show that 65% of computer system failures are caused by HMI software malfunction; HMI software abnormality may lead to system crash and even casualties and great economic loss [14].

2.3. Switch

Switches are very important in communication and equipment messages exchange. There are many real-time messages inside the substation, such as the manufacturing message specification

(MMS) message from IEDs to the SCADA system, generic object-oriented substation events (GOOSE) messages among IEDs, sampling values (SVs) of power parameter messages from MUs to IEDs, and the SNTP/IRIG-B/IEEE 1588 time synchronizing messages for equipment timing accuracy adjustment. Switches can also implement virtual local area network (VLAN), reduce network traffic jam caused by a large amount of data flowing through one single path, increase performance efficiency, and improve information security [15]. A switch must communicate and transmit data in real time and with accuracy, be resistant to electromagnetic interference (EMI), avoid long delay time which may expose the system in dangerous conditions, tolerate ambient temperature variation, and not malfunction or fail due to high temperature.

2.4. IEDs

Figure 2. The example configuration of the proposed intelligent substation SCADA system.

2.1. Server

The server is the hardware equipment of the intelligent substation SCADA system which iscapable of running various application software. The server manages the substation automationsystem efficiently and ensures the power quality provided by the power grid is high. The maintenanceand management personnel can monitor the substation operating status, look up both real-time andhistorical alarms, and print out event records on-line merely through the server of the SCADA system.

2.2. SCADA HMI Software

The SCADA HMI software is a key element in the realization of intelligent substation automaticcontrol and logical analysis. The main functions of the SCADA HMI software include equipment datacollecting, system status monitoring and testing, power flow calculation, short circuit analysis, powergrid energy dispatch and management, sending control commands, and security control. Statisticaldata show that 65% of computer system failures are caused by HMI software malfunction; HMIsoftware abnormality may lead to system crash and even casualties and great economic loss [14].

2.3. Switch

Switches are very important in communication and equipment messages exchange. There aremany real-time messages inside the substation, such as the manufacturing message specification(MMS) message from IEDs to the SCADA system, generic object-oriented substation events (GOOSE)messages among IEDs, sampling values (SVs) of power parameter messages from MUs to IEDs, and theSNTP/IRIG-B/IEEE 1588 time synchronizing messages for equipment timing accuracy adjustment.Switches can also implement virtual local area network (VLAN), reduce network traffic jam caused bya large amount of data flowing through one single path, increase performance efficiency, and improveinformation security [15]. A switch must communicate and transmit data in real time and with accuracy,be resistant to electromagnetic interference (EMI), avoid long delay time which may expose the systemin dangerous conditions, tolerate ambient temperature variation, and not malfunction or fail due tohigh temperature.

Energies 2019, 12, 3892 5 of 22

2.4. IEDs

IEDs are core protection and control devices in a power system. Parameters are set and protectionlogics are programmed differently depending on the needs and operating conditions of different systems.IEDs receive various electrical data from potential transformers (PTs), current transformers (CTs), MUs,and other equipment. IEDs send dispatch control commands and transmit power operating data tothe SCADA system at the station level, or other related equipment through network communication,thus effectively reducing power outages and harm caused by power system abnormal operation.

2.5. MUs

According to the IEC 60044-7 and IEC 60044-8 standards published by the InternationalElectrotechnical Commission (IEC), an MU may be defined preliminarily as “an electronic transformerwhich replaces a conventional electromagnetic equipment, converts analog voltage/current signals todigital signals to avoid EMI to conventional analog SVs and transmit the digital signals to bay levelequipment through optical fibers, thus improves the accuracies of power parameters effectively” [16].

3. Fuzzy Comprehensive Evaluation Method for Equipment Maintenance Strategy

3.1. Establishment of the Evaluation Factors of an Intelligent Substation SCADA System

When evaluating the status of an intelligent substation SCADA system, equipment influencingthe normal operation of the SCADA system must be inspected and tested. Because equipment mayhave many attributes and characteristics, the SCADA system is inevitably affected by many uncertainfactors and by many different types of equipment [7].

An intelligent substation SCADA system consists of a lot of equipment, and each performanceindex of individual equipment reflects the performance status of individual equipment and the systemas a whole. Server (u1), SCADA HMI software (u2), Switch (u3), IED (u4), and MU (u5) are analyzedin this study. Equipment availability (u11), communication port failure rate (u12), CPU usage rate(u13), memory usage rate (u14), and transmission load rate (u15) are included in the server reliabilityevaluation model [17,18]. Software availability (u21), data receiving rate (u22), data access time (u23),and failure repair time (u24) are included in the SCADA HMI software reliability evaluation model [19].Equipment availability (u31), communication port failure rate (u32), packet lost rate (u33), transmissionload rate (u34), and CPU usage rate (u35) are included in the switch reliability evaluation model [16].Equipment availability (u41), operation failure rate (u42), communication failure rate (u43), operatingenvironment (u44), and CPU usage rate (u45) are included in the IED reliability evaluation model [20].Equipment availability (u51), SV packet lost rate (u52), SV code error rate (u53), transmission time(u54), time synchronization accuracy (u55), and CPU usage rate (u56) are included in the MU reliabilityevaluation model [20].

Different comments are needed to describe different equipment status, and five ratings ofinfluencing factor status are chosen according to the equipment operating status: Excellent, Good,Average, Poor, and Worst. The corresponding numerical values are 100, 75, 50, 25, and 0, respectively.The comment set V of intelligent substation SCADA system is shown below.

V = { Excellent, Good, Average, Poor, Worst } = { 100, 75, 50, 25, 0 }

3.2. Establishment of the Equipment Status Evaluation Matrix

The difference between current equipment status and its failure status is represented by thenumerical value of the relative deterioration degree (RDD), with values ranging from 0 to 1. Two typesof indices are defined in Equations (1) and (2), and either may be chosen depending on the indexcharacteristic [21]. The larger-is-better type of index may be applied to equipment availability withvalues closer to 1, implying less degree of deterioration. The less-is-better type of index may be appliedto equipment failure rate with values closer to 1, implying a higher degree of deterioration and moreurgency for maintenance.

Energies 2019, 12, 3892 6 of 22

The larger-is-better type:

xi =Xmxu −XminXint −Xmin

The less-is-better type:

xi =Xmxu −XintXmax −Xint

where xi is the RDD of the i-th evaluation factor, Xint is the initial test value of the factor, Xmin is theminimum limit of the factor, Xmax is the maximum limit of the factor, and Xmxu is the measured valueof the factor.

The equipment influencing factors of the intelligent substation SCADA system are evaluated bytrapezoidal and triangular membership functions to determine their degrees of membership in theevaluation comment set V, as shown in Figure 3. The membership functions of the linguistic variables“Excellent”, “Good”, “Average”, “Poor”, and “Worst” are defined in Equations (3)–(7), with theparameters defined and evenly distributed as α = 0.1, β = 0.3, γ = 0.5, δ = 0.7, and ε = 0.9, respectively.

intiX XxX X

−=−

The less-is-better type:

mxu int

intiX XxX X

where xi is the RDD of the i-th evaluation factor, Xint is the initial test value of the factor, Xmin is the minimum limit of the factor, Xmax is the maximum limit of the factor, and Xmxu is the measured value of the factor.

The equipment influencing factors of the intelligent substation SCADA system are evaluated by trapezoidal and triangular membership functions to determine their degrees of membership in the evaluation comment set V, as shown in Figure 3. The membership functions of the linguistic variables “Excellent”, “Good”, “Average”, “Poor”, and “Worst” are defined in Equations (3), (4), (5), (6), and (7), with the parameters defined and evenly distributed as α = 0.1, β = 0.3, γ = 0.5, δ = 0.7, and ε = 0.9, respectively.

Figure 3. The membership functions of the linguistic variables.

1 ( )( ) ( )

0 other

αβ α ββ α

≤ −= < ≤ −

( ) ( )( x)( ) ( )

0 other

α α ββ αγ β γγ β

− < ≤ −

−= < ≤ −

( ) ( )( )( ) ( )

0 other

xm x x

β β γγ βδ γ δδ γ

− < ≤ −

−= < ≤ −

Figure 3. Figure 3. The membership functions of the linguistic variables.

m1(x) =

1, x ≤ α(β−x)(β−α) , α < x ≤ β

0, other

m2(x) =

(x−α)(β−α) , α < x ≤ β(γ−x)(γ−β) , β < x ≤ γ

0, other

m3(x) =

(x−β)(γ−β) , β < x ≤ γ(δ−x)(δ−γ) , γ < x ≤ δ

0, other

m4(x) =

(x−γ)(δ−γ) , γ < x ≤ δ(ε−x)(ε−δ) , δ < x ≤ ε

0, other

m5(x) =

(x−δ)(ε−δ) , δ < x ≤ ε

1, ε < x

0, other

Energies 2019, 12, 3892 7 of 22

where mk(x) is the membership function of the k-th status comment, and x is the RDD of the evaluationfactor.

The evaluation results of the evaluation factors are put together to form the fuzzy evaluationmatrix R, as shown in Equation (8):

r11 . . . r1 j...

. . ....

ri1 · · · ri j

i = 1, 2, · · ·, m; j = 1, 2, · · ·, n (8)

where i is the number of evaluation factors, j is the number of status comments, and rij is the membershipdegree of the j-th comment of the i-th evaluation factor.

3.3. Calculation of Equipment Weights

The importance of individual equipment in an intelligent substation SCADA system is differentdepending on its functionality. Therefore, it is scientifically crucial in status evaluation to assign properweight to equipment according to its degree of importance. Subjective weighting relies heavily onhuman thinking pattern and is not very scientific, while objective weighting depends on parametervariation. A comprehensive weighting method is proposed in this study, which combines the subjectiveanalytic hierarchy process (AHP) method and the objective entropy weighting method. Becausethe proposed comprehensive weighting method inherits the field experiences of experts and is incompliance with parameter authenticity, it can be applied to evaluate equipment status effectively.

3.3.1. Calculation of the Subjective Weights

The AHP method resolves a complex problem to a configuration of concise factor levels. Evaluationtables are formed to assess the importance degree of each factor and to serve as reference during theweighting process [22,23]. Researchers, manufacturers, and utility personnel are consulted on theimportance degrees of the equipment and influencing factors of the intelligent substation SCADAsystem. The geometric means of their opinions are calculated to obtain the subjective weights. The stepsof calculations are as follows.

Step 1: Construct a hierarchy structure model: Group the factors in the problem according totheir interrelationship into three layers—ultimate target layer, evaluation item layer, and index layer.Sublayers may be formed to simplify the computation if there are too many factors in a layer.

Step 2: Construct a comparison judgement matrix: The judgement matrix A is formed by pairwisecomparisons between factors of the same level to evaluate their relative degree of importance, as shownin Equation (9). The definitions and explanations of the evaluation scales are shown in Table 1.

a11 . . . a1 j

.... . .

...ai1 · · · ai j

i = 1, 2, · · ·, m; j = 1, 2, · · ·, n (9)

Table 1. The definition and explanation of the evaluation scales.

Evaluation Scale Definition Explanation

1 Equally important The two factors are equally important.3 Slightly more important One factor is only slightly more important than the other.5 Significantly more important One factor is significantly more important than the other.7 Highly more important One factor is highly more important than the other.9 Absolutely more important One factor is absolutely more important than the other.

2,4,6,8 In-between scales The degree of relative importance lies between the twoadjacent scales.

Energies 2019, 12, 3892 8 of 22

Step 3: Calculate the weights: After the judgement matrix A is derived from the opinions ofexperts, numerical analysis is applied to calculate the maximum eigenvalue and its correspondingeigenvector, as shown in Equation (10). The relative importance or weight distribution of the factorsmay be sorted by analyzing the eigenvector. The relationship between the eigenvector and the weightis shown in Equation (11).

A · E =m∑

n∑j=1

ai j · ei = λmax · E (10)

m∑i=1

where Wa is the subjective weight, aij is an element of the judgment matrix A, λmax is the maximumeigenvalue of A, and ei is an element of eigenvector E corresponding to λmax.

The purpose of a consistancy test is to check if the respondents’ answers to the questionnairecomply with transitivity, which is the basis to verify the validity of the questionnaire to avoid thewrong decision of evaluation. The consistanct test is the important factor to verify the accuracy of thesubjective weighting.

The validity of the questionnaire is based on the Consistency Index (CI), as shown in Equation(12). The tolerance value is CI ≤ 0.1. In addition, to further test whether the hierarchical structureof the judgment matrix complies with the consistency standard, the consistency ratio (CR) is tested.In Equation (13), the tolerance value CR ≤ 0.1. If CI and CR fail the test, the steps of analytic hierarchyprocess must be re-executed.

CI =λmax − n

n− 1(12)

CR =CIRI

CI: Consistency Index (CI) of judgement matrixRI: Random Index (RI) of judgement matrixn: Matrix order

The random index value (RI) increases as the matrix order increases. For the 1–10 order pairingmatrix, the RI values are shown in Table 2.

Table 2. Order of AHP method and relative random index values.

Order (n) 1 2 3 4 5 6 7 8 9 10

RI 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49

If CR ≤ 0.1, the calculated weight value can be accepted. Otherwise, if the CR value is larger,the judgment matrix filled in by the respondent may be inconsistent due to the complexity of theproblem and the diversity of subjective cognition.

3.3.2. Calculation of the Objective Weights

The concept of entropy is employed by the entropy weighting method in information theory toassess the degree of decision information the value of a parameter of an evaluation item can carrywhen a target is evaluated, and the relative weight of an index is determined based on the variation ordegree of difference of a parameter. In general, an evaluation factor with higher entropy is assigned aless weight. The steps of the calculation are as follows [24].

Energies 2019, 12, 3892 9 of 22

Step 1: Standardize the parameters: Parameter values of different orders of magnitude ofevaluation items are converted to relative quantities by the standardization rule shown in Equation (14).A standardization matrix Z = [ zij ]m×n can be built for a system with m evaluation index data and nsemantic variables.

zi j =ri j − rmin

rmax − rmin(14)

where rij is the original value of membership degree, rmax is the maximum value and rmin is theminimum value of the evaluation factor, respectively.

Step 2: Calculate the message entropy: By the definition of message entropy, the standardizationmatrix Z is used to calculate the value of entropy Hi of the evaluation factors, as shown inEquations (15) and (16):

Hi = −1

n∑j=1

fi j · ln fi j) i= 1, 2, · ··, m (15)

fi j =zi j + 1

n∑j=1

(zi j + 1)(16)

where Hi is the entropy of the i-th evaluation factor, zij is the standardized index of the j-th semanticvariable of the i-th evaluation factor, m is the number of the evaluation factors, and n is the number ofthe semantic variables.

Step 3: Calculate the weights: After the values of the message entropies are obtained as describedabove, the degrees of difference of all the evaluation factors are known and the corresponding weightsare calculated according to the entropy values, as shown in Equation (17):

We =1−Hi

m−m∑

where We is the objective weight.

3.3.3. Calculation of the Comprehensive Weights

The subjective weights of the AHP method and the objective weights of the entropy weightingmethod are integrated in the comprehensive weighting method. The comprehensive weighting methodis applied in this study to obtain the equipment weight. The mathematical model of the comprehensiveweighting method is shown in Equation (18):

wi =Wai ·Wei

m∑i=1

Wai ·Wei

where wi, Wai, and Wei are the comprehensive weight, subjective weight, and objective weight of thei-th evaluation factor, respectively.

3.4. Establishment of the Equipment FCE Matrix

Te fuzzy linear transform principle and the maximum membership principle are applied in theFCE method. The relationships among the evaluation problem and the relevant factors are considered,and rational comprehensive evaluations are made [7,25]. The steps of evaluation are as follows.

Only one single value of equipment evaluation with equal importance degree can be obtained bythe fuzzy evaluation matrix. To effectively improve the accuracy of the evaluation matrix, the conceptof weight is introduced into the evaluation matrix, and the mathematical representation of the FCEmatrix is shown in Equation (19):

Energies 2019, 12, 3892 10 of 22

Bi = wi ·Ri = [bi1, bi2, bi3, · · · , bin] i = 1, 2, · · ·, m (19)

where bij is the weighted membership degree of the j-th comment of the i-th evaluation factor.The FCE method is applied to integrate the evaluation matrix of each factor, assign the weight of

relative importance, and calculate the overall evaluation result of each factor on the final evaluationtarget. The mathematical representation is shown in Equation (20):

S = w · BT = w · [B1, B2, · · ·, Bn]T (20)

where S is the membership degree of the n evaluation factors.

3.5. Numerical Calculations of the Evaluation

The result of the FCE matrix calculation is the membership degrees of evaluation ratings ofthe evaluation target. Because the analyzing personnel cannot determine the real-time status of theevaluation target quickly enough, a numerical value is assigned to each evaluation rating in order tosimplify the analytical evaluation result. The assessment result of the FCE matrix, Ti and T, are shownin Equations (21) and (22), respectively. The equipment status ratings and the corresponding numericalevaluation ranges are summarized in Table 3.

Ti = Bi · vT = [bi1, bi2, · · · , bin][100, 75, 50, 25, 0]T (21)

T = S · vT = [S1, S2, · · · , Sn][100, 75, 50, 25, 0]T (22)

Table 3. Equipment status ratings and the corresponding numerical evaluation ranges.

Numerical Evaluation Equipment Status

90~100 In Good Condition80~90 Attention Required70~80 In Critical Condition0~70 Immediate Inspection and Maintenance Required

4. Case Study

An intelligent substation SCADA system example platform is built in the Tai-Tam substation of theTaipower company. The hardware configuration of this platform is shown in Figure 4. Two operationcases of this platform have been explored and analyzed to determine the priority for equipmentmaintenance and repair. For case 1, the actual measured values of equipment influencing factors andthe corresponding fuzzified RDDs are shown in Tables 4–8. The membership degrees of the equipmentinfluencing factors are shown in Tables 9–13.

80~90 Attention Required 70~80 In Critical Condition 0~70 Immediate Inspection and Maintenance Required

4. Case Study

An intelligent substation SCADA system example platform is built in the Tai-Tam substation of the Taipower company. The hardware configuration of this platform is shown in Figure 4. Two operation cases of this platform have been explored and analyzed to determine the priority for equipment maintenance and repair. For case 1, the actual measured values of equipment influencing factors and the corresponding fuzzified RDDs are shown in Tables 4–8. The membership degrees of the equipment influencing factors are shown in Tables 9–13.

Figure 4. The hardware configuration of the example platform.

Table 4. Server status data and status evaluation.

Influencing Factor Index Type Measured Value RDD/Status Membership u11 Equipment availability Larger-is-better 0.9664 0.8188/Excellent

u12 Communication port failure rate (%) Less-is-better 0 0/Excellent u13 CPU usage rate (%) Less-is-better 32 0/Excellent

u14 Memory usage rate (%) Less-is-better 63 0.0750/Excellent u15 Transmission load rate (%) Less-is-better 11 0/Excellent

Table 5. SCADA human–machine interface (HMI) software status data and status evaluation.

Influencing Factor Index Type Measured Value RDD/Status Membership u21 Software availability Larger-is-better 0.9998 0.9990/Excellent

u22 Data receiving rate (%) Larger-is-better 100 1.0000/Excellent u23 Data access time (s) Less-is-better 10 0.1667/Excellent

u24 Failure repair time (hr) Less-is-better 24 0/Excellent

Table 6. Switches status data.

Influencing Factor Measured Values

A B C D E F u31 Equipment availability 0.9498 0.9872 0.9901 0.9901 0.9872 0.9872

u32 Communication port failure rate (%) 0 0 0 0 0 0 u33 Packet lost rate (%) 0 0 0 0 0 0 u34 CPU usage rate (%) 18 18 24 23 9 9

u35 Transmission load rate (%) 33 21 26 19 16 15

Influencing Factor RDD/Status Membership

A B C D E F

u31 Equipment availability 0.5845/Excell

ent 0.8941/Excell

ent 0.8955/

Excellent 0.8955/Excell

ent 0.8941/Excell

Figure 4. The hardware configuration of the example platform.

Energies 2019, 12, 3892 11 of 22

Table 4. Server status data and status evaluation.

Influencing Factor Index Type Measured Value RDD/Status Membership

u11 Equipment availability Larger-is-better 0.9664 0.8188/Excellentu12 Communication port failure rate (%) Less-is-better 0 0/Excellent

u13 CPU usage rate (%) Less-is-better 32 0/Excellentu14 Memory usage rate (%) Less-is-better 63 0.0750/Excellent

u15 Transmission load rate (%) Less-is-better 11 0/Excellent

Table 5. SCADA human–machine interface (HMI) software status data and status evaluation.

u21 Software availability Larger-is-better 0.9998 0.9990/Excellentu22 Data receiving rate (%) Larger-is-better 100 1.0000/Excellent

u23 Data access time (s) Less-is-better 10 0.1667/Excellentu24 Failure repair time (hr) Less-is-better 24 0/Excellent

Table 6. Switches status data.

Influencing Factor Measured Values

A B C D E F

u31 Equipment availability 0.9498 0.9872 0.9901 0.9901 0.9872 0.9872u32 Communication port failure rate (%) 0 0 0 0 0 0

u33 Packet lost rate (%) 0 0 0 0 0 0u34 CPU usage rate (%) 18 18 24 23 9 9

u35 Transmission load rate (%) 33 21 26 19 16 15

Influencing Factor RDD/Status Membership

A B C D E F

u31 Equipment availability 0.5845/Excellent

0.8941/Excellent

0.8955/Excellent

0.8941/Excellent

u32 Communication port failure rate (%) 0/Excellent 0/Excellent 0/Excellent 0/Excellent 0/Excellent 0/Excellent

u33 Packet lost rate (%) 0/Excellent 0/Excellent 0/Excellent 0/Excellent 0/Excellent 0/Excellent

u34 CPU usage rate (%) 0.1368/Excellent

0.1368/Excellent

0.2/Excellent

0.1895/Excellent

0.0421/Excellent

u35 Transmission load rate (%) 0/Excellent 0/Excellent 0/Excellent 0/Excellent 0/Excellent 0/Excellent

A: PT-G7828_A, B: PT-G7828_B, C: PT-G503_A, D: PT-G503_B, E: PT-7728_A, F: PT-7728_B.

Table 7. Intelligent electronic device (IED) status data and status evaluation.

A B A B

u41 Equipmentavailability Larger-is-better 0.9950 0.9950 0.8977/Excellent 0.8977/Excellent

u42 Operation failurerate (%) Less-is-better 0 0 0/Excellent 0/Excellent

u43 Communicationfailure rate (%) Less-is-better 0 0 0/Excellent 0/Excellent

u44 Operatingenvironment (◦C) Less-is-better 27 27 0/Excellent 0/Excellent

u45 CPU usage rate (%) Less-is-better 21 19 0.1684/Excellent 0.1474/Excellent

Energies 2019, 12, 3892 12 of 22

Table 8. Merging unit (MU) status data and status evaluation.

u51 Equipment availability Larger-is-better 0.9967 0.8985/Excellentu52 SV packet lost rate (%) Less-is-better 0 0/Excellentu53 SV code error rate (%) Less-is-better 0 0/Excellentu54 Transmission time (µs) Less-is-better 1 0/Excellent

u55 Time synchronization accuracy (µs) Less-is-better 0.1 0.1000/Excellentu56 CPU usage rate (%) Less-is-better 20 0.1000/Excellent

Table 9. Membership degrees of server influencing factors (Ru1).

IndexMembership Degree

R1 R2 R3 R4 R5

u11 0.5938 0.4062 0 0 0u12 1.0000 0 0 0 0u13 1.0000 0 0 0 0u14 1.0000 0 0 0 0u15 1.0000 0 0 0 0

Table 10. Membership degrees of SCADA HMI software influencing factors (Ru2).

R1 R2 R3 R4 R5

u21 1.0000 0 0 0 0u22 1.0000 0 0 0 0u23 0.6667 0.3333 0 0 0u24 1.0000 0 0 0 0

Table 11. Membership degrees of switch influencing factors (Ru3).

R1 R2 R3 R4 R5

u311 0.9706 0.0294 0 0 0u312 1.0000 0 0 0 0u313 1.0000 0 0 0 0u314 0.8158 0.1842 0 0 0u315 1.0000 0 0 0 0

u321 0.9706 0.0294 0 0 0u322 1.0000 0 0 0 0u323 1.0000 0 0 0 0u324 0.8158 0.1842 0 0 0u325 1.0000 0 0 0 0

u331 0.9774 0.0226 0 0 0u332 1.0000 0 0 0 0u333 1.0000 0 0 0 0u334 0.5000 0.500 0 0 0u335 1.0000 0 0 0 0

u341 0.9774 0.0226 0 0 0u342 1.0000 0 0 0 0u343 1.0000 0 0 0 0u344 0.5526 0.4474 0 0 0u345 1.0000 0 0 0 0

Energies 2019, 12, 3892 13 of 22

Table 11. Cont.

R1 R2 R3 R4 R5

u351 0 0.9319 0.0681 0 0u352 1.0000 0 0 0 0u353 1.0000 0 0 0 0u354 1.0000 0 0 0 0u355 1.0000 0 0 0 0

u361 0.9706 0.0294 0 0 0u362 1.0000 0 0 0 0u363 1.0000 0 0 0 0u364 1.0000 0 0 0 0u365 1.0000 0 0 0 0

Table 12. Membership degrees of IED influencing factors (Ru4).

R1 R2 R3 R4 R5

u411 0.9887 0.1113 0 0 0u412 1.0000 0 0 0 0u413 1.0000 0 0 0 0u414 1.0000 0 0 0 0u415 0.6579 0.3421 0 0 0

u421 0.9887 0.1113 0 0 0u422 1.0000 0 0 0 0u423 1.0000 0 0 0 0u424 1.0000 0 0 0 0u425 0.7632 0.2368 0 0 0

Table 13. Membership degrees of MU influencing factors (Ru5).

R1 R2 R3 R4 R5

u51 0.9925 0.0075 0 0 0u52 1.0000 0 0 0 0u53 1.0000 0 0 0 0u54 1.0000 0 0 0 0u55 1.0000 0 0 0 0u56 1.0000 0 0 0 0

The analytic hierarchy process (AHP) method is applied in the intelligent substation SCADAsystem analysis to calculate the equipment subjective weights, as shown in Table 14. The maximumeigenvalue and the corresponding eigenvector are λmax = 5.0438 and E = [0.7149 0.2241 0.3732 0.51700.1794], respectively. The consistency index and the consistency ratio are CI = 0.0110, n = 5, RI = 1.12,and CR = 0.0098, respectively. Since CR 5 0.1, the weights are acceptable.

The subjective, objective, and comprehensive weights of the server status evaluation are shown inTable 15. The AHP method is applied and the maximum eigenvalue and the corresponding eigenvectorare λmax = 5.0164 and E = [0.2240 0.3262 0.1863 0.0854 0.1781], respectively. The consistency index andthe consistency ratio are CI = 0.0041, n = 5, RI = 1.12, and CR = 0.0037, respectively. Since CR 5 0.1,the weights are acceptable.

The subjective, objective, and comprehensive weights of the SCADA HMI software statusevaluation are shown in Table 16. The AHP method is applied and the maximum eigenvalue andthe corresponding eigenvector are λmax = 4.0803 and E = [0.8001 0.4425 0.2237 0.3376], respectively.

Energies 2019, 12, 3892 14 of 22

The consistency index and the consistency ratio are CI = 0.0268, n = 4, RI = 0.9, and CR = 0.0297,respectively. Since CR 5 0.1, the weights are acceptable.

The subjective, objective, and comprehensive weights of the switch status evaluation are shownin Tables 17–20. The AHP method is applied and the maximum eigenvalue and the correspondingeigenvector are λmax = 5.0494 and E = [0.4118 0.7113 0.4775 0.2052 0.2331], respectively. The consistencyindex and the consistency ratio are CI = 0.0124, n = 5, RI = 1.12, and CR = 0.0110, respectively. SinceCR 5 0.1, the weights are acceptable.

The subjective, objective, and comprehensive weights of the IED status evaluation are shown inTables 21 and 22. The AHP method is applied and the maximum eigenvalue and the correspondingeigenvector are λmax = 5.0096 and E = [0.2105 0.6576 0.6058 0.2915 0.2671], respectively. The consistencyindex and the consistency ratio are CI = 0.0024, n = 5, RI = 1.12, and CR = 0.0021, respectively. SinceCR 5 0.1, the weights are acceptable.

The subjective, objective, and comprehensive weights of the MU status evaluation are shown inTable 23. The AHP method is applied and the maximum eigenvalue and the corresponding eigenvectorare λmax = 6.0210 and E = [0.1981 0.5283 0.4586 0.4479 0.4720 0.2187], respectively. The consistencyindex and the consistency ratio are CI = 0.0042, n = 6, RI = 1.24, and CR = 0.0034, respectively. SinceCR 5 0.1, the weights are acceptable.

Table 14. The subjective weights of the intelligent substation SCADA system influencing factors.

Influencing Factor u1 u2 u3 u4 u5 Subjective Weight (w)

u1 1.0000 2.8914 2.4721 1.4612 3.1468 0.3559u2 0.3459 1.0000 0.5109 0.4391 1.3457 0.1116u3 0.4045 1.9573 1.0000 0.6371 2.5196 0.1858u4 0.6844 2.2774 1.5696 1.0000 2.8143 0.2574u5 0.3178 0.7431 0.3969 0.3553 1.0000 0.0893

Table 15. The weights of the server influencing factors.

InfluencingFactor u11 u12 u13 u14 u15

SubjectiveWeight

IndexEntropy

ObjectiveWeight

ComprehensiveWeight (w1)

u11 1.0000 0.6177 1.4225 2.5189 1.2244 0.2240 h11 0.4197 0.1267 0.1435u12 1.6189 1.0000 1.6861 4.0145 1.6329 0.3262 h12 0 0.2183 0.3600u13 0.7030 0.5931 1.0000 2.1227 1.2139 0.1863 h13 0 0.2183 0.2056u14 0.3970 0.2491 0.4711 1.0000 0.4764 0.0854 h14 0 0.2183 0.0943u15 0.8167 0.6124 0.8238 2.0992 1.0000 0.1781 h15 0 0.2183 0.1966

Table 16. The weights of the SCADA HMI software influencing factors.

InfluencingFactor u21 u22 u23 u24

SubjectiveWeight

IndexEntropy

ObjectiveWeight

u21 1.0000 2.4662 3.5569 1.7100 0.4435 h21 0 0.2774 0.4664u22 0.4055 1.0000 2.4662 1.4422 0.2453 h22 0 0.2774 0.2580u23 0.2811 0.4055 1.0000 0.8434 0.1240 h23 0.3955 0.1677 0.0788u24 0.5848 0.6934 1.1857 1.0000 0.1872 h24 0 0.2774 0.1969

Table 17. The subjective weights of the switch influencing factors.

SubjectiveWeight

u31 1.0000 0.6177 0.7003 2.5189 1.6166 0.2020u32 1.6189 1.0000 1.4763 2.8257 3.9857 0.3489u33 1.4280 0.6774 1.0000 2.1227 1.8372 0.2342u34 0.3970 0.3539 0.4711 1.0000 0.8181 0.1006u35 0.6186 0.2509 0.5443 1.2224 1.0000 0.1143

Energies 2019, 12, 3892 15 of 22

Table 18. The objective and comprehensive weights of the influencing factors of Switch A and Switch B.

InfluencingFactor

Switch A Switch B

Index Entropy ObjectiveWeight

ComprehensiveWeight (w31) Index Entropy Objective

WeightComprehensive

Weight (w32)

u31 h311 0.0825 0.1986 0.1944 h321 0.0825 0.1986 0.1944u32 h312 0 0.2164 0.3659 h322 0 0.2164 0.3659u33 h313 0 0.2164 0.2456 h323 0 0.2164 0.2456u34 h314 0.2968 0.1522 0.0742 h324 0.2968 0.1522 0.0742u35 h315 0 0.2164 0.1199 h325 0 0.2164 0.1199

Table 19. The objective and comprehensive weights of the influencing factors of Switch C and Switch D.

InfluencingFactor

Switch C Switch D

WeightComprehensive

Weight (w34)

u31 h331 0.0672 0.2072 0.2005 h341 0.0672 0.2070 0.1997u32 h332 0 0.2210 0.3695 h342 0 0.2219 0.3698u33 h333 0 0.2210 0.2480 h343 0 0.2219 0.2482u34 h334 0.4307 0.1265 0.0610 h344 0.4272 0.1271 0.0611u35 h335 0 0.2210 0.1210 h345 0 0.2219 0.1212

Table 20. The objective and comprehensive weights of the influencing factors of Switch E and Switch F.

InfluencingFactor

Switch E Switch F

WeightComprehensive

Weight (w36)

u31 h351 0.1546 0.1745 0.1763 h361 0.0825 0.1866 0.1855u32 h352 0 0.2064 0.3601 h362 0 0.2034 0.3548u33 h353 0 0.2064 0.2417 h363 0 0.2034 0.2382u34 h354 0 0.2064 0.1038 h364 0 0.2034 0.1023u35 h355 0 0.2064 0.1180 h365 0 0.2034 0.1162

Table 21. The subjective weights of IED influencing factors.

SubjectiveWeight

u41 1.0000 0.2818 0.3555 0.8011 0.7864 0.1036u42 3.5491 1.0000 1.0975 2.1824 2.2039 0.3236u43 2.8129 0.9112 1.0000 2.0794 2.3647 0.2981u44 1.2483 0.4582 0.4809 1.0000 1.1728 0.1434u45 1.2716 0.4537 0.4229 0.8527 1.0000 0.1314

Table 22. The objective and comprehensive weights of the influencing factors of IED A and IED B.

InfluencingFactor

IED A IED B

WeightComprehensive

Weight (w42)

u41 h411 0.0385 0.2170 0.1083 h421 0.0385 0.2081 0.1047u42 h412 0 0.2192 0.3419 h422 0 0.2164 0.3401u43 h413 0 0.2192 0.3149 h423 0 0.2164 0.3133u44 h414 0 0.2192 0.1515 h424 0 0.2164 0.1507u45 h415 0.3992 0.1317 0.0834 h425 0.3401 0.1428 0.0911

Energies 2019, 12, 3892 16 of 22

Table 23. The weights of the MU influencing factors.

InfluencingFactor u51 u52 u53 u54 u55 u56

SubjectiveWeight

IndexEntropy

ObjectiveWeight

u51 1.0000 0.4137 0.3544 0.4311 0.3810 1.0917 0.0853 h51 0.0275 0.1628 0.0831u52 2.4173 1.0000 1.3197 1.2097 1.0674 2.3916 0.2274 h52 0 0.1674 0.2279u53 2.8214 0.7577 1.0000 0.9714 0.9913 2.0147 0.1974 h53 0 0.1674 0.1978u54 2.3198 0.8267 1.0294 1.0000 0.9688 1.9358 0.1928 h54 0 0.1674 0.1932u55 2.6249 0.9369 1.0088 1.0322 1.0000 1.9961 0.2031 h55 0 0.1674 0.2036u56 0.9160 0.4181 0.4964 0.5166 0.5010 1.0000 0.0941 h56 0 0.1674 0.0943

As a demonstration, the procedures and the numerical calculations of the FCE matrix for case 1 inthis study are described in detail as follows.

(1) Server FCE matrix B1 and its numerical value T1:

B1 = w1 ·Ru1

= [0.1435, 0.3600, 0.2056, 0.0943, 0.1966]

0.59381.00001.00001.00001.0000

0.40620000

0 00 00 00 00 0

= [0.9417, 0.0583 , 0, 0, 0]

T1 = B1 ·VT

= [0.9417, 0.0583, 0, 0, 0]

1007550250

= 98.54

(2) SCADA HMI software fuzzy evaluation matrix B2 and its numerical value T2:

B2 = w2 ·Ru2

= [0.4664, 0.2580, 0.0788, 0.1969]

1.00001.00000.66671.0000

0.33330

0 00 00 00 0

= [0.9738, 0.0262, 0, 0, 0]

T2 = B2 ·VT = 99.35

(3) Switches fuzzy evaluation matrix B3:

The PT-G7828_A fuzzy evaluation matrix B31 and its numerical value T31:

B31 = w31 ·Ru31

= [0.1944, 0.3659, 0.2456, 0.0742, 0.1199]

0.97061.00001.00000.81581.0000

0.029400

0.18420

0 00 00 00 00 0

= [0.9806, 0.0194, 0, 0, 0]

Energies 2019, 12, 3892 17 of 22

T31 = B31 ·VT = 99.52

The PT-G503_A fuzzy evaluation matrix B33 and its numerical value T33:

B33 = w33 ·Ru33

= [0.2005, 0.3695, 0.2480, 0.0610, 0.1210]

0.97741.00001.00000.50001.0000

0.022600

0.5000

0 00 00 00 00 0

= [0.9651, 0.0349, 0, 0, 0]

T33 = B33 ·VT = 99.13

In order to improve the data transmission reliability, parallel redundancy protocol (PRP)and high-availability seamless redundancy (HSR) are introduced in the network communicationconfiguration, where PRP is followed using switches PT-G7828_A and PT-G7828_B, while HSR isachieved using PT-G503_A, PT-G503_B, PT-7728_A, and PT-7728_B. The six switches are representedby one equivalent switch to simplify the calculation with fuzzy evaluation matrix B3 and its numericalvalue T3:

B3 = 0.5[0.5(B31 + B32) + 0.25(B33 + B34 + B35 + B36)]

= [0.9806, 0.0194, 0, 0, 0]

T3 = B3 ·VT = 99.52

(4) IEDs fuzzy evaluation matrix R4:

The IED_A fuzzy evaluation matrix B41 and its numerical value T41:

B41 = w41 ·Ru41

= [0.1083, 0.3419, 0.3149, 0.1515, 0.0834]

0.98871.00001.00001.00000.6579

0.0113000

0.3421

0 00 00 00 00 0

= [0.9702, 0.0298, 0, 0, 0]

T41 = B41 ·VT = 99.25

B42 = w42 ·Ru42

= [0.1047, 0.3401, 0.3133, 0.1507, 0.0911]

0.98871.00001.00001.00000.7632

0.0113000

0.2368

0 00 00 00 00 0

= [0.9772, 0.0228, 0, 0, 0]

T42 = B42 ·VT = 99.43

Energies 2019, 12, 3892 18 of 22

The equivalent IED fuzzy evaluation matrix B4 and its numerical value T4:

B4 = 0.5(B41 + B42) = [0.9737, 0.0263, 0, 0, 0]

T4 = B4·V T = 99.34

(5) The MU fuzzy evaluation matrix R5 and its numerical value T5:

B5 = w5 ·Ru5

= [0.0831, 0.2279, 0.1978, 0.1932, 0.2036, 0.0943]

0.99251.00001.00001.00001.0000

0.00750000

0 00 00 00 00 0

= [0.9994, 0.0006, 0, 0, 0]

T5 = B5 ·VT = 99.99

(6) The fuzzy evaluation matrix S and its numerical value T:

S = w · BT

= [0.3559, 0.1116, 0.1858, 0.2574, 0.0893]

0.94170.97380.98060.97370.9994

0.05830.02620.01940.02630.0006

0 00 00 00 00 0

= [0.9659, 0.0341, 0, 0, 0]

T = S · vT= 99.15

The equipment FCE result of case 1 is shown in Table 24. All of the equipment is newly purchasedin case 1. The status evaluation of the intelligent substation SCADA system of case 1 is 99.15 or “inGood Condition” and no inspection or maintenance schedule is required. Taipower company Tai-Tamsubstation CBM platform of the intelligent substation SCADA system is shown in Figure 5.

Table 24. The equipment FCE result of case 1 and 2.

Equipment Evaluation Result

Case 1 Case 2

Server 98.54 79.46SCADA HMI Software 99.35 55.74

Switch_A 99.52 22.64Switch_B 99.52 22.64Switch_C 99.13 7.74Switch_D 99.20 22.68Switch_E 99.86 22.64Switch_F 99.86 22.64

IED_A 99.25 88.89IED_B 99.43 88.89

MU 99.99 87.72SCADA System 99.15 69.07

The system in case 2 is old and has run for years, with all of the switches functioning in abnormalstatus, as shown in Table 25. The status evaluation of the SCADA system of case 2 is 69.07 or “Immediate

Energies 2019, 12, 3892 19 of 22

Inspection and Maintenance Required”, as shown in Table 24. Although the SCADA system has a highreliability network configuration, its reliability is reduced drastically due to the serious deterioration ofthe switch, and immediate maintenance of all switches is required. The recommended maintenanceorder for the switches is Switch_C→ Switch_A, Switch_B, Switch_E, and Switch_F→ Switch_D. If thesystem is still in critical condition after the communication problem is resolved, the SCADA HMIsoftware should be maintained too.

Table 25. Case 2 measured value of influencing factor of equipment.

Server SCADAHMI Switch IED MU

IF *1 MV IF MV IFMV *2

IF MVA B C D E F A B

u11 0.9025 u21 0.9991 u31 0.9498 0.9498 0.9610 0.9610 0.9498 0.9498 u41 0.9851 0.9851 u51 0.9900u12 0 u22 80 u32 3 3 4 3 3 3 u42 1 1 u52 0u13 32 u23 20 u33 2 2 3 2 2 2 u43 0 0 u53 0u14 80 u24 120 u34 90 90 100 90 90 90 u44 50 50 u54 5u15 11 u35 100 100 100 100 100 100 u45 25 25 u55 0.6

u56 20*1 IF: Influencing factor, *2 MV: Measured value.

Table 24. The equipment FCE result of case 1 and 2.

Equipment Evaluation result

Case 1 Case 2 Server 98.54 79.46

SCADA HMI Software 99.35 55.74 Switch_A 99.52 22.64 Switch_B 99.52 22.64 Switch_C 99.13 7.74 Switch_D 99.20 22.68 Switch_E 99.86 22.64 Switch_F 99.86 22.64

IED_A 99.25 88.89 IED_B 99.43 88.89

MU 99.99 87.72 SCADA System 99.15 69.07

Table 25. Case 2 measured value of influencing factor of equipment.

Server SCADA HMI Switch IED MU

IF*1 MV IF MV IF MV*2

IF MV A B C D E F A B

u11 0.9025

u21 0.9991

u31 0.9498

0.9498

0.9610

0.9498

u41 0.9851

0.9851

u51 0.9900

u12 0 u22 80 u32 3 3 4 3 3 3 u42 1 1 u52 0 u13 32 u23 20 u33 2 2 3 2 2 2 u43 0 0 u53 0 u14 80 u24 120 u34 90 90 100 90 90 90 u44 50 50 u54 5 u15 11 u35 100 100 100 100 100 100 u45 25 25 u55 0.6

u56 20 ＊1 IF: Influencing factor, ＊2 MV: Measured value.

Figure 5. The CBM platform of the intelligent substation SCADA system.

5. Conclusion

Figure 5. The CBM platform of the intelligent substation SCADA system.

5. Conclusions

Smart substations play a vital role in power systems. However, the installation of a large number ofsecondary equipment makes the traditional correct maintenance and time base maintenance unable tomeet the system requirements. It is imperative to develop a new equipment maintenance managementstrategy. In recent years, equipment maintenance strategies have been further developed and optimizedby monitoring equipment abnormalities and considering the importance of equipment. Implementinga CBM system for a smart substation can effectively avoid over-maintenance or lack of maintenance ofequipment, reducing unnecessary power outage tests, as well as the maintenance workload and cost.It can significantly improve the system operational reliability and economic efficiency.

In contrast to a conventional maintenance strategy, a condition-based maintenance andmanagement strategy is proposed in this study, where equipment status information is gatheredthrough the IEC 61,850 communication protocol. The relative deterioration degree theory (RDD)and fuzzy theory (FT) are used to evaluate the condition of the equipment, combining the subjective

Energies 2019, 12, 3892 20 of 22

analytic hierarchy process (AHP) method with the objective entropy weighting method to analyzethe important factor of equipment. The fuzzy comprehensive evaluation (FCE) method is applied toevaluate the equipment of an intelligent substation SCADA system based on the equipment conditionand their importance. The evaluated equipment includes the server, SCADA HMI software, switch,IED, and MU.

Each equipment in the substation plays a different role according to its function. Assigningthem reasonable weights by the scientific approach is an important process for state evaluation.The comprehensive weighting method proposed in this paper that combines the subjective analytichierarchy process (AHP) method and the objective entropy weighting method can inherit the advantagesof two kinds of weighting methods. This approach not only presents the subjective professionalexperience of decision makers, but also considers the truth of the objective facts. It can accuratelyevaluate equipment status in the smart substation control system.

The result shows that the equipment status maintenance and management platform developed inthis study can diagnose the equipment operating status in real time. The inspection and maintenancepersonnel can analyze the overall equipment condition by knowing which one of the four status ratings,i.e., "in Good Condition", "Attention Required", "in Critical Condition", and "Immediate Inspectionand Maintenance Required", the equipment status has, and can determine the priority for equipmentmaintenance. The evaluation result can serve as a valuable reference to utility companies when makingmaintenance plans.

Author Contributions: Conceptualization, J.-C.G. and C.-H.L.; methodology, J.-C.G. and C.-H.L.; software,C.-H.L.; validation, K.-Y.C. and M.-T.Y.; formal analysis, M.-T.Y.; investigation, K.-Y.C.; resources, C.-H.L.; datacuration, M.-T.Y.; writing—original draft preparation, K.-Y.C. and C.-H.L.; writing—review and editing, M.-T.Y.;supervision, J.-C.G.; project administration, J.-C.G.

Funding: This research received no external funding.

Conflicts of Interest: The authors declare no conflicts of interest.

Nomenclature

AHP Analytic Hierarchy ProcessCB Circuit BreakerCBM Condition-based MaintenanceCT Current TransformerEMI Electromagnetic InterferenceFCE Fuzzy Comprehensive EvaluationFT Fuzzy TheoryGOOSE Generic Object-Oriented Substation EventsHMI Human-machine InterfaceHSR High-Availability Seamless RedundancyIEC International Electrotechnical CommissionIED Intelligent Electronic DeviceMMS Manufacturing Message SpecificationMU Merging UnitPM Preventive MaintenancePRP Parallel Redundancy ProtocolPT Potential TransformerRDD Relative Deterioration DegreeSCADA Supervisory Control and Data AcquisitionSV Sampling ValueTBM Time-Based MaintenanceVLAN Virtual Local Area Network

Energies 2019, 12, 3892 21 of 22

References

1. Orlowska, T.; Balzer, G.; Halfmann, M. Lifecycle management of circuit breaker by applicationof reliability centered maintenance. CIGRE 2000 Sess. 2000, 13–103. Available online:https://www.semanticscholar.org/paper/Life-cycle-management-of-circuit-breakers-by-of-Orlowska-Balzer/8e7501a7e6561108aa232a8f062fef369651d829 (accessed on 14 October 2019).

2. Ye, Y.B.; Xie, M.; Huang, T.G. Research on condition-based maintenance in relay protection. In Proceedingsof the 2017 4th International Conferences on Information Science and Control Engineering, Changsha, China,21–23 July 2017; pp. 1627–1631.

3. Maity, A.; Zaremby, D.; McMullen, D. Automated scheduling using condition based maintenance.In Proceedings of the 2011 IEEE Conferences on Prognostics and Health Management, Montreal, QC,Canada, 20–23 June 2011; pp. 1–8.

4. On-Line Diagnostics for Condition Based and Reliability Centered Maintenance. Available online: http://www.stri.se/wwwpublic/GRIDTECH-On-line-Diagnostics.pdf (accessed on 1 March 2017).

5. Liao, R.J.; Wang, Q.; Luo, S.J. Condition assessment model for power transformer in service based on fuzzysynthetic evaluation. Autom. Electr. Power Syst. 2008, 32, 70–75.

6. Lin, P.C.; Gu, J.C.; Yang, M.T. Intelligent maintenance model for condition assessment of circuit breakersusing fuzzy set theory and evidential reasoning. IET Gener. Transm. Distrib. 2014, 8, 1244–1253. [CrossRef]

7. Li, Z.; Dai, Z.; Jiao, Y. Research on condition-based maintenance strategy of protection systems based onmultistage fuzzy comprehensive evaluation. In Proceedings of the 2014 11th International Conferences onFuzzy Systems and Knowledge Discovery, Xiamen, China, 19–21 August 2014; pp. 268–272.

8. Zhao, D.; Liang, W.; Zhang, X. Relay protection condition assessment based on variable weight fuzzysynthetic evaluation. In Proceedings of the 2014 IEEE Conferences on Technologies for Sustainability,Portland, OR, USA, 24–26 July 2014; pp. 115–120.

9. Qian, P.; Tian, X.; Kanfoud, J.; Lee, J.L.Y.; Gan, T. A novel condition monitoring method of wind turbinesbased on long short-term memory neural network. Energies 2019, 12, 3411. [CrossRef]

10. Yang, Y.; Sørensen, J.D. Cost-optimal maintenance planning for defects on wind turbine blades. Energies2019, 12, 998. [CrossRef]

11. Asgarpour, M.; Sørensen, J.D. Bayesian based diagnostic model for condition based maintenance of offshorewind farms. Energies 2018, 11, 300. [CrossRef]

12. Nielsen, J.S.; Sørensen, J.D. Bayesian estimation of remaining useful life for wind turbine blades. Energies2017, 10, 664. [CrossRef]

13. Tian, F.; Jing, Z.; Zhao, H.; Zhang, E.; Liu, J. A synthetic condition assessment model for power transformersusing the fuzzy evidence fusion method. Energies 2019, 12, 857. [CrossRef]

14. Zeng, F.F.; Sun, J.H.; Wu, X. Software reliability modeling, measuring and forecasting in computer supervisorysystem. Chin. Comput. Eng. Applications 2002, 38, 95–97.

15. Wang, C. Research on Reliability of Digital Substation Relay Protection System. Master’s Thesis, NationalChekiang University, Hangzhou, China, 2013.

16. Pascal, J.; Zurfluh, F. Sensor interface suitable for IEC 60044-7 and 8 metering accuracy class 0.1 and protectionclass 5P. In Proceedings of the 2013 IEEE International Symposium on Industrial Electronics, Taipei, Taiwan,28–31 May 2013; pp. 1–4.

17. Zhao, D.L.; Liu, S.L. Failure model analysis of communication system in intelligent substation. J. Jiyuan Vocat.Tech. Coll. 2014, 13, 9–12.

18. Qiu, J.; Wang, H.F.; Chen, Z.G. Effectiveness evaluation model of smart substation automation system.Autom. Electr. Power Syst. 2013, 37, 87–94.

19. Schneidewind, N.F. Software reliability model with optimal selection of failure data. IEEE Trans. Soft. Eng.1993, 19, 1095–1104. [CrossRef]

20. Peng, Z.F. Research of Secondary Equipment Performance Evaluation Methods in Smart Substation.Master’s Thesis, North China Electric Power University, Beijing, China, 2014.

21. Liao, R.; Zheng, H.; Grzybowski, S. An integrated decision-making model for condition assessment of powertransformers using fuzzy approach and evidential reasoning. IEEE Trans. Power Deliv. 2011, 26, 1111–1118.[CrossRef]

22. Satty, T.L. What is the analytic hierarchy process? Math. Model. Decis. Support 1988, 48, 109–121.

Energies 2019, 12, 3892 22 of 22

23. Satty, T.L. The seven pillars of the analytic hierarchy process. Mult. Criteria Decis. Mak. New Millenn. 2001, 7,15–37.

24. Shannon, C.E. A mathematical theory of communication. ACM SIGMOBILE Mob. Comput. Commun. Rev.2001, 5, 3–55. [CrossRef]

25. Deng, C.; Liu, J.; Liu, Y. A fuzzy comprehensive evaluation for metropolitan power grid risk assessment.In Proceedings of the 2016 International Conferences on Smart Grid and Clean Energy Technologies, Chengdu,China, 19–22 October 2016; pp. 1–5.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open accessarticle distributed under the terms and conditions of the Creative Commons Attribution(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Research on CBM of the Intelligent Substation SCADA System

Documents