Secondary equipment state evaluation method based on evidence theory

Abstract

The invention discloses a secondary equipment state evaluation method based on an evidence theory. According to the method, the application advantages of the evidence theory in comprehensive evaluation of secondary equipment are considered and analyzed. Firstly, equipment health states are divided; and secondly, determining an information fusion model according to the index characteristics of each level, fusing evidences by adopting an ER rule and an ER algorithm for different index levels, and finally obtaining a comprehensive evaluation model of the equipment state. Thirdly, determining a model input quantity, acquiring evidence weight by adopting an improved G1 method, acquiring confidence distribution of evidence by adopting a model based on a ridge-shaped membership function, and acquiring evidence reliability by adopting an evidence distance; and finally, determining the final state of the equipment according to the degree of support of the equipment to each state obtained by the evaluation model and the maximum membership principle. The method is suitable for comprehensively evaluating the secondary equipment, and the health state of the secondary equipment can be more accurately evaluated, so that the method can be used for guiding the analysis design, operation maintenance and other work of the secondary equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power system secondary equipment evaluation technology, and particularly relates to a secondary equipment state evaluation method based on evidence theory. BACKGROUND

[0002] The rapid development of smart grid brings more opportunities and challenges to secondary equipment, and puts forward higher requirements for the safe and stable operation of secondary equipment itself. However, with the continuous increase of the number of secondary equipment, the burden of operation and maintenance personnel is increasing, and the problems of fault maintenance and periodic maintenance mode are increasingly prominent: fault maintenance is a passive working mode, time is tight, task is heavy, and it is unpredictable; and periodic maintenance has the problems of "under-inspection" and "over-inspection". Compared with the above, as a preventive maintenance strategy, state maintenance can reduce the operation and maintenance pressure and improve economic benefits according to the actual operation condition and health state of the equipment to determine the maintenance and repair time.

[0003] As the basis of state maintenance, equipment state evaluation can provide basis and guidance for it. In the field of state evaluation, the method based on evidence theory is widely used because it has many advantages such as no need for prior probability, strong decision processing ability, good algorithm convergence and strong anti-interference. On the one hand, from the research application field, evidence theory is less used in the specific field of secondary equipment state evaluation of power system, the reason of evidence conflict in secondary equipment state evaluation is unknown, and there is a lack of suitable evaluation model; on the other hand, from the angle of information processing, the generation of basic probability distribution, the measurement of evidence conflict information and the fusion of conflict evidence are still the research hotspots in the field of evidence theory, which need to be further studied. Therefore, considering the above problems, it is of great significance to establish a secondary equipment state evaluation method based on evidence theory. SUMMARY

[0004] The purpose of the present application is to provide a secondary equipment state evaluation method based on evidence theory, which considers the application advantages of evidence theory in the field of state evaluation, establishes a secondary equipment state evaluation model based on evidence theory, and can be used to accurately evaluate the actual state of secondary equipment, and has strong practical value.

[0005] The purpose of the present application is achieved by the following technical solutions:

[0006] A secondary equipment state evaluation method based on evidence theory, the method comprises:

[0007] Step 1, referring to national standards and related literature, determining the secondary equipment state evaluation index system, and dividing the health state of secondary equipment;

[0008] Step 2, according to the conflict causes of each level index, determine the information fusion method, and build a secondary equipment comprehensive evaluation model;

[0009] Step 3, determine the input quantity of the model, that is, the weight of evidence, the confidence distribution and the reliability;

[0010] Step 4, according to the support degree of the equipment to each state obtained by the evaluation model, determine the final state of the equipment.

[0011] From the above technical solution provided by the present application, it can be seen that the above method is simple to calculate, can accurately evaluate the health state of the secondary equipment, and has strong practical value. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0013] Figure 1 The secondary equipment general evaluation index system framework provided for the embodiments of the present application;

[0014] Figure 2 The secondary equipment state comprehensive evaluation flowchart provided for the embodiments of the present application;

[0015] Figure 3 The membership degree image corresponding to the deterioration degree for different states provided for the embodiments of the present application; DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0017] The embodiments of the present application will be further described in detail below in combination with the drawings. The method comprises:

[0018] Step 1, refer to national standards and related literature to determine the secondary equipment state evaluation index system, and divide the health state of the secondary equipment;

[0019] In the step 1, the specific process is:

[0020] Referring to the relevant standards of the power industry, the health state level of secondary equipment is generally divided into normal, attention, abnormal, and serious states, wherein the specific meaning and recommended processing method of each state are shown in Table 1.

[0021] Table 1: State classification of secondary equipment and recommended processing method

[0022]

[0023]

[0024] Note: "standard limit value / attention value / warning value" is defined as a relative degree of deterioration of 0.5;

[0025] According to the hardware architecture design of secondary equipment, monitoring information, and the like, the running index and real-time running index are comprehensively counted to form a general evaluation index system framework of secondary equipment, as shown in Table 1. Figure 1 Further, according to the difference in hardware structure and function implementation of the specific evaluated secondary equipment type, other performance class indexes and self-checking information are selected to form the corresponding equipment state evaluation index system. Table 2 is the state evaluation index system of wideband measurement equipment.

[0026] Table 2: State evaluation index system of wideband measurement equipment

[0027]

[0028] Step 2: According to the reasons for the conflict of each level index, the information fusion method is determined, and a comprehensive evaluation model of secondary equipment is constructed;

[0029] In the step 2, as shown in Table 1, the specific process is as follows: Figure 2

[0030] In the comprehensive evaluation of the equipment health state, the reasons for the conflict of evidence are mainly four types:

[0031] ① Local information cannot reflect the overall state;

[0032] ② The independence of the functions described by each index is strong;

[0033] ③ The reliability of the sensor / information source is different;

[0034] ④ There is an error in the index state determination;

[0035] ​For the third level indicators, the main causes of conflict are ①③④, for the second, first and target level indicators, the main causes of conflict are ①②, that is, unlike other levels of indicators, for the third level indicators, the reliability needs to be considered. Therefore, for the third level indicators, the ER rule is used to fuse the evidence, and for the second, first and target levels, the ER algorithm is used to fuse the evidence, and finally the comprehensive evaluation model of the equipment health state is obtained.

[0036] The specific content of the ER algorithm is as follows:

[0037] For the evaluation of the health state of the equipment, it is assumed that the identification framework Θ is composed of N evaluation levels H n (n=1,2,···,N), that is, Θ={H1, H2, …, H N}, and P(Θ) is the power set of Θ. Then, the corresponding evidence e i formed by the ith evaluation index can be expressed in the form of confidence distribution as follows:

[0038] e i ={(H n ,p n,i ),n=1,2,…,N;(Θ,p Θ,i )} (1)

[0039] Where p n,i represents the confidence of the ith index corresponding evidence e i with respect to the nth evaluation level, and satisfies 0≤p n,i ≤1, p Θ,i is the confidence of the evidence e i with respect to the identification framework, which represents global ignorance.

[0040] Assume that the normalized weight of the evidence e i is w i (i=1,2,···,L), which satisfies 0≤w i ≤1, and The basic probability mass of the evidence e i is expressed as follows:

[0041]

[0042] For any two independent evidences e i and e j , the combined probability mass is determined by equations (3)-(6):

[0043] m n,e(2) =K0[m n,i m n,j +m n,i (m Θ,j +m P(Θ),j)+m n,j (m Θ,i +m P(Θ),i )] (3)

[0044] m Θ,e(2) =K0(m Θ,i m Θ,j +m Θ,i m P(Θ),j +m P(Θ),i m Θ,j ) (4)

[0045] m P(Θ),e(2) =K0m P(Θ),i m P(Θ),j (5)

[0046]

[0047] Where m n,e(2) 、m Θ,e(2) and m P(Θ),e(2) Respectively represent evidence e i and e j After combining, the evaluation level is assigned to H n , the joint probability mass of the identification framework and the power set of the identification framework; K0 is the normalization coefficient, which ensures that the sum of the probability masses after combination is 1.

[0048] The corresponding combined confidence can be expressed as:

[0049]

[0050]

[0051] Furthermore, for L independent pieces of evidence {e1, e2, ..., e L}, repeat equations (3) to (8) and iterate them to get the final evaluation result of the health status:

[0052] e(L)={(H n ,p n,e(L) ),n=1,2,…,N;(Θ,p Θ,e(L) )} (9)

[0053] Formulas (3) to (8) can be equivalently described as:

[0054]

[0055] The ER rule further incorporates the reliability of evidence based on the ER algorithm, taking into account both the inherent properties of evidence (reliability) and the subjective concepts of the evaluator (weight). The specific content is as follows:

[0056] For the device health state assessment, suppose the recognition framework consists of N assessment levels H n (n = 1, 2,..., N), i.e., Θ = {H1, H2,..., H n}. Then, the corresponding evidence e i formed by the ith assessment indicator can be expressed in the form of a belief distribution as shown in equation (1). Suppose the weight and reliability of the evidence e i are w i and r i , respectively, then the belief distribution of the evidence weighted with the reliability is:

[0057]

[0058] where,

[0059]

[0060] In the equation, p is the mixed probability mass assigned to the level H n in the ith indicator; P(Θ) represents the power set of the recognition framework, is the mixed probability mass assigned to the power set in the ith indicator; φ represents the empty set; c rw,i = 1 / (1 + w i - r i ) represents the normalization coefficient; m n,i is the basic probability mass assigned to the level H n in the ith indicator, and m n,i = w i p n,i .

[0061] For any two independent evidences e i and e j , suppose their belief distribution forms can be expressed by equation (15), then their joint support p n to the level H θ,e(2) is:

[0062]

[0063] In the equation, p is the unnormalized combined probability mass assigned to H i after combining the evidences e j and e n ; A, B, and E all represent subsets of Θ; represents the unnormalized combined probability mass assigned to the assessment level E after combining the evidences e i and e j .

[0064] Without loss of generality, for L independent pieces of evidence {e1, e2, ..., e L}, they are for level H n Joint support of p n,e(L) By continuously iterating formulas (19) to (22), we can obtain:

[0065]

[0066] Where, k = 3, 4, ···, L; Assign the first k pieces of evidence to level H after combination n The unnormalized probability mass of n,e(k-1) The first (k-1) pieces of evidence are combined and assigned to level H n The normalized probability mass of the first k and (k-1) pieces of evidence respectively reflects the probability of the first k and (k-1) pieces of evidence for the level H n The joint support level of m n,e(1) =m n,1 、m P(Θ),e(1) =m P(Θ),1 , where the unreliability of the first indicator m P(Θ),1 =1-r1.

[0067] Through iteration, the final comprehensive health evaluation result is:

[0068] M={(H n ,p n,e(L) ),n∈[1,N];(Θ,p Θ,e(L) )} (twenty three)

[0069] Where p n,e(L) and p Θ,e(L) The results after the fusion of L pieces of evidence are respectively n The joint confidence (support) with the entire set.

[0070] Step 3: Determine the model inputs, i.e., weight of evidence, confidence distribution, and reliability;

[0071] In step 3, the specific process is:

[0072] For model input, ER rules require the weight, confidence distribution, and reliability of evidence, while ER algorithms require the confidence distribution and weight of evidence. The confidence distribution of evidence corresponding to the indicators at the second and first level can be obtained from the fusion results at the third level. Therefore, it is necessary to determine the weight, confidence distribution, and reliability of evidence.

[0073] (1) The specific method for obtaining the weight of evidence is as follows:

[0074] Firstly, the normal weight value of the index is obtained based on the sequence relationship analysis method. Considering that in the actual evaluation process, the contribution of the index with small weight but serious degradation may be "submerged" and cannot be truly reflected on the device state, the degradation degree is introduced to dynamically correct the normal weight value, and the variable weight value is obtained as the maximum weight of the index.

[0075] Among them, the degradation degree reflects the degree of degradation of each index, and its value is in [0, 1], which can intuitively reflect the degree of current index close to failure. In the device evaluation index, according to the index type, it can be divided into qualitative index and quantitative index, and the quantitative index is divided into two-way degradation index and one-way degradation index according to its characteristics, as follows:

[0076] 1) Two-way degradation index

[0077] The two-way degradation index has upper and lower limit values. When the index value is within the good upper and lower limit value range, it is considered that the index has no degradation. The part exceeding the good upper and lower limit value is quantitatively processed according to the deviation degree, and the calculation formula is:

[0078]

[0079] In the formula, x is the measured value of the index; β and α are the upper and lower limit values allowed by the index; β1 and α1 are the upper and lower limit values within the good range.

[0080] 2) One-way degradation index

[0081] The one-way degradation index has an upper limit value or a lower limit value, which can be divided into two types: the smaller the better and the larger the better.

[0082] ① The smaller the better index: the smaller the value, the better the state of the object, such as utilization rate, temperature, etc., and the calculation formula is:

[0083]

[0084] ② The larger the better index: the larger the value, the better the state of the object, and the calculation formula is:

[0085]

[0086] Therefore, the calculation formula of the variable weight value is:

[0087]

[0088] In the formula, A it is the variable weight value corresponding to the tth state characteristic parameter in the ith subsystem, and m is the number of state parameters under the ith subsystem; w is (s = 1, 2, 3, …, m) is the normal weight value corresponding to the sth state characteristic parameter in the ith subsystem, and d is(s=1, 2, 3, ..., m) represents the degradation value of the sth state characteristic parameter under the i-th subsystem; T is the variable weight coefficient, which is 3 in this patent.

[0089] For the bottom-level indicators (i.e., third-level indicators), the degradation degree can be obtained from the monitoring data and the corresponding standards. For the first and second-level indicators, the maximum degradation degree among the bottom-level indicators is taken as the degradation value of the upper-level indicator, that is:

[0090] d i =max(d is ) (28)

[0091] Where, d i is the degradation value of the i-th subsystem. For example, for secondary equipment, the degradation value of the CPU board is the maximum value among the MCU junction temperature, CPU load rate and memory self-test abnormality.

[0092] (2) The specific method for obtaining the confidence distribution of evidence is as follows:

[0093] This patent adopts a method based on fuzzy semantic value to obtain the confidence distribution of evidence. This method adopts a method based on membership function to solve the confidence of the input quantitative information relative to the fuzzy reference value.

[0094] For secondary equipment status evaluation, the ridge distribution is selected as the membership function, and the specific function expression is as follows:

[0095]

[0096] Where h1(d), h2(d), h3(d), and h4(d) are the ridge-shaped membership functions corresponding to the “normal”, “caution”, “abnormal”, and “serious” situations, respectively, when the indicator degradation degree is d.

[0097] like Figure 3 As shown in , when two ridge-shaped membership functions intersect, the part falling in the intersection area will have two membership degrees that are not 0. In this case, it is reasonable to believe that the degradation degree corresponds to one of the two ridge-shaped membership function categories, and the intersection of different ridge-shaped membership functions is regarded as a new membership function to represent a composite category. Figure 3 In, h 12 Represents the membership function corresponding to the composite category consisting of the "normal" and "caution" states. It means that the state is one of "normal" and "caution", but it is not certain which state it is.

[0098] Obtain the degree of membership of the evidence for each state as the confidence level. The specific rules are as follows:

[0099] 1) When a sample intersects with the ridge fuzzy number representation model of a single subset proposition (such as normal {H1}), the ordinate of the intersection point is the confidence of the sample in the single subset proposition.

[0100] 2) When a sample intersects the ridge-shaped fuzzy number representation model of a multi-subset proposition (e.g., {H1, H2}), it will inevitably intersect with the ridge-shaped fuzzy number model of the single-subset proposition. The high point of the intersection is the sample's confidence in the single-subset proposition, and the low point is the sample's confidence in the multi-subset proposition.

[0101] 3) If the sum of the confidences for each state is greater than 1, it is normalized; if the sum is less than 1, the redundant confidences are allocated to the entire set, that is, to the unknown.

[0102] (3) The specific method for obtaining the reliability of evidence is as follows:

[0103] 1) Calculate the support S between evidences ij

[0104] S ij =1-d JU (e i ,e j ) (30)

[0105] Where S ij Representative evidence i For e j Support degree; d JU (e i ,e j ) indicates evidence e i Information provided for evidence j degree of conflict.

[0106] 2) Calculate the total support of the evidence

[0107] Assuming the number of collected evidence is L, the support matrix Sup can be obtained as follows:

[0108]

[0109] The smaller the conflict between two pieces of evidence, the higher the support level. i The total support level S i for:

[0110]

[0111] 3) Calculate the reliability of evidence

[0112] Reliability reflects the credibility of a piece of evidence. It is generally believed that the higher the degree to which a piece of evidence is supported by other evidence, the higher the credibility of the evidence. Therefore, after normalizing the support, we can get the reliability, thereby defining the evidence e i The reliability r i for:

[0113]

[0114] The reliability of the three-level indicators can be obtained through formula (8) and then used as the input of the evidence reasoning rules.

[0115] Step 4: Determine the final state of the device based on the degree of support for each state obtained by the evaluation model.

[0116] In step 4, the specific process is:

[0117] The state evaluation model is used to obtain the support of the device for each state, including the support of uncertain states such as {normal, attention}. Pignisitic probability transformation is further used to evenly distribute the confidence of multiple subsets to the included single subsets. For example, the state evaluation result of the broadband measurement device is:

[0118] Table 3 Status evaluation results of broadband measurement equipment

[0119]

[0120] Following the "maximum membership principle", that is, if the difference between the maximum support value and the second largest support value of the final evaluation state exceeds the set threshold, the state with the largest support value is selected as the final evaluation result. The threshold value of this patent is 0.01.

[0121] It should be noted that the contents not described in detail in the embodiments of the present invention belong to the prior art known to those skilled in the art.

[0122] Actual examples show that considering the non-independence of some signals in the monitoring system can make the quantitative calculation of substation monitoring system reliability more accurate, and can therefore be used to guide the analysis, design, operation and maintenance of substation monitoring systems.

[0123] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A secondary equipment status evaluation method based on evidence theory, characterized in that: The method comprises: Step 1: Refer to national standards and relevant literature to determine the secondary equipment status evaluation index system and classify the secondary equipment health status; Step 2: According to the causes of conflicts among indicators at each level, determine the information fusion method and build a comprehensive evaluation model for secondary equipment; Step 3: Determine the model inputs, i.e., weight of evidence, confidence distribution, and reliability; Step 4: Determine the final state of the device based on the degree of support for each state obtained by the evaluation model.

2. The secondary equipment status evaluation method according to claim 1, characterized in that: In step 1, the process of determining the secondary equipment status evaluation index system and classifying the secondary equipment health status is specifically as follows: According to relevant standards of the power industry, the health status of secondary equipment is generally divided into: normal, caution, abnormal, and serious status. At the same time, based on the hardware architecture design and monitoring information of secondary equipment, comprehensive statistical operation indicators and real-time operation indicators are compiled to form a general evaluation index system framework for secondary equipment. Furthermore, based on the differences in hardware structure and functional implementation of the specific secondary equipment type being evaluated, other performance indicators and self-test information are selected to form a corresponding equipment status evaluation index system.

3. The secondary equipment status evaluation method according to claim 1, characterized in that: In step 2, the process of determining the information fusion method and constructing the comprehensive evaluation model for secondary equipment is specifically as follows: In the comprehensive evaluation of equipment health status, there are four main reasons for evidence conflicts: ① Local information cannot reflect the overall status; ② The functions described by each indicator are highly independent; ③ Different reliability of sensors / information sources; ④ There are errors in the determination of indicator status; For third-level indicators, the main causes of conflict are ①, ③, and ④. For second-level, first-level, and target-level indicators, the main causes of conflict are ① and ②. Unlike indicators at other levels, third-level indicators require consideration of their reliability. Therefore, this patent uses the ER rule to fuse evidence for the third-level indicator layer, and the ER algorithm to fuse evidence for the second-level, first-level, and target layers, ultimately resulting in a comprehensive evaluation model for equipment health status.

4. The secondary equipment status evaluation method according to claim 1, characterized in that: In step 3, the process of determining the model input, namely, the weight of evidence, confidence distribution and reliability, is specifically as follows: First, based on the membership function method, the degree of membership of each piece of evidence to each state is obtained, which is used as the confidence of the state to form a confidence distribution form; secondly, based on the confidence distribution form of each piece of evidence, the reliability of the evidence is obtained; then the weight of the evidence is obtained; finally, the mixed probability mass assigned to each indicator at each level is calculated, thereby obtaining the weighted confidence distribution form of evidence with reliability.

5. The secondary equipment status evaluation method according to claim 1, characterized in that: In step 4, the process of determining the final state of the device based on the support degree of the device for each state obtained by the evaluation model is specifically as follows: The support degree of the equipment for each state is obtained through the state evaluation model, and the Pignisitic probability transformation is used to evenly distribute the credibility of multiple subsets to the included single subsets. Furthermore, the "maximum membership principle" is followed, that is, if the difference between the maximum support value and the second largest support value of the final evaluation state exceeds the set threshold, the state with the largest support value is selected as the final evaluation result.

Images