Chemical equipment state evaluation method and system based on expert evaluation information

Abstract

The invention discloses a chemical equipment state evaluation method and system based on expert evaluation information, and relates to the technical field of equipment detection, and the method comprises the steps: obtaining expert language evaluation information, analyzing the expert language evaluation information, and generating an ordered language power set; carrying out uncertainty measurement and fusion on the ordered language power set; generating a corresponding fuzzy evaluation matrix based on the fusion result; and on the basis of the fuzzy evaluation matrix, a good and bad solution distance method is adopted to calculate and obtain a state evaluation result of the sample. According to the method, accurate extraction and effective fusion of expert evaluation information are realized, and evaluation is automatically given.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment detection, and more particularly to a method and system for evaluating the state of chemical equipment based on expert evaluation information. Background Art

[0002] Due to the particularity of the chemical industry, once a chemical equipment fails, it often causes serious economic losses. At the same time, it is very likely to cause the leakage of toxic and harmful gases or even explosion accidents, resulting in casualties. With the rapid development of artificial intelligence technology, deep learning algorithms have been widely used in the state monitoring, detection and fault diagnosis of chemical equipment. However, deep learning algorithms represented by neural networks have the problem that their performance is highly dependent on training samples, resulting in low fault diagnosis ability in the case of serious imbalance between the categories of training samples in the actual field, especially for unknown faults that have not been trained, the recognition ability of the system is very limited. Therefore, it is very necessary to retain the expert evaluation method with stronger generalization ability and higher-dimensional thinking ability as an auxiliary.

[0003] The traditional expert evaluation method uses hierarchical scoring or degree words to manually evaluate the state of each feature of the current chemical equipment, and the mechanism of fuzzy evaluation is not introduced in the expert evaluation process. If an expert encounters a situation where a certain feature can only be clearly defined in a certain fuzzy interval rather than a specific value or degree during the scoring process, it is possible to give a wrong evaluation because a specific and definite degree must be given, resulting in a deviation in the final overall evaluation result.

[0004] Therefore, how to provide a method and system for evaluating the state of chemical equipment based on expert evaluation information, realizing the accurate extraction and effective fusion of expert evaluation information, and automatically giving an evaluation is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a method and system for evaluating the state of chemical equipment based on expert evaluation information. Based on the ordered language power set, the language comprehensive entropy is used to measure and fuse the uncertainty of expert evaluation information, and the automatic evaluation of the state of chemical equipment is realized through the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) algorithm.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A method for evaluating the state of chemical equipment based on expert evaluation information, comprising:

[0007] Obtaining expert language evaluation information, analyzing the expert language evaluation information, and generating an ordered language power set;

[0008] Performing uncertainty measurement and fusion on the ordered language power set;

[0009] Generate a corresponding fuzzy evaluation matrix based on the fusion result;

[0010] Based on the fuzzy evaluation matrix, calculate the state evaluation result of the sample by using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS).

[0011] Preferably, analyze the expert linguistic evaluation information to generate an ordered linguistic power set, including:

[0012] Establish a corresponding identification framework and a corresponding evaluation index system for the expert linguistic evaluation information;

[0013] Generate an ordered linguistic power set \(P\) under the ordered linguistic identification framework Ω The formula is as follows:

[0014]

[0015] where \(\{\nu i ,\nu j \}\ o =\{\nu i ,\nu i+1 ,\cdots,\nu j-1 ,\nu j \}(1\leq i\leq j\leq n)\) represents that the elements in the subset are continuous;

[0016] The numerical form of the identification framework \(\Omega = \{-\chi,-\chi + 1,\cdots,0,\cdots,\chi - 1,\chi\}\) is based on the ordered linguistic set, and the number of corresponding power sets is \((\chi + 1)(2\chi + 1)\).

[0017] Preferably, perform uncertainty measurement and fusion on the ordered linguistic power set, including:

[0018] Set corresponding coefficients for the entropy \(H L \) of the uncertainty of the linguistic degree word itself and the entropy \(H E \) of the evaluation uncertainty to form a corresponding comprehensive entropy \(H\);

[0019] Calculate the evaluation information matrix given by each expert for \(q\) features of \(m\) samples;

[0020] Calculate the weight of each expert based on the comprehensive entropy \(H\);

[0021] Perform weighted fusion calculation on the corresponding items in the evaluation information matrices of all experts.

[0022] Preferably, based on the fusion result, generate a corresponding fuzzy evaluation matrix, including:

[0023] Calculate the comprehensive entropy \(H\) for the fusion result and obtain a corresponding comprehensive entropy matrix;

[0024] Calculate the weight corresponding to each feature and generate the corresponding evaluation matrix;

[0025] Generate a weighted average matrix based on the corresponding evaluation matrix.

[0026] Preferably, the weighted average matrix is expressed as:

[0027]

[0028] T ij represents the evaluation matrix corresponding to the i-th feature of sample j; represents the weight corresponding to each feature.

[0029] Preferably, based on the fuzzy evaluation matrix, the state evaluation result of the sample is calculated by using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), including:

[0030] Calculate the corresponding optimal solution Q + and the worst solution Q - :

[0031]

[0032] wherein, represents the weighted average matrix; T i + 、T i - respectively represent the maximum and minimum values of the samples in the i-th feature;

[0033] Calculate the distance value d(Q j ,Q + ) between each element in the weighted average matrix and the optimal solution and the distance value d(Q j ,Q - ) between each element in the weighted average matrix and the worst solution:

[0034]

[0035] wherein, q represents the total number of features;

[0036] According to the distance value d(Q j ,Q + ) between each element in the weighted average matrix and the optimal solution and the distance value d(Q j ,Q - ), calculate the relative similarity of the corresponding sample j:

[0037]

[0038] Relative similarity S jThe larger it is, the smaller the distance between sample j and the optimal solution, and the better the corresponding sample j is.

[0039] Preferably, according to the relative similarity S j sort the samples, and obtain the state evaluation results of the corresponding samples.

[0040] Preferably, a chemical equipment state evaluation system based on expert evaluation information includes:

[0041] An information analysis module, configured to obtain expert language evaluation information, analyze the expert language evaluation information, and generate an ordered language power set;

[0042] A fusion module, configured to perform uncertainty measurement and fusion on the ordered language power set;

[0043] A matrix generation module, configured to generate a corresponding fuzzy evaluation matrix based on the fusion result;

[0044] A state evaluation module, configured to calculate the state evaluation results of samples based on the fuzzy evaluation matrix by using the method of distance between the best and the worst solutions.

[0045] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a method and system for evaluating the state of chemical equipment based on expert evaluation information, including: obtaining expert language evaluation information, analyzing the expert language evaluation information, and generating an ordered language power set; performing uncertainty measurement and fusion on the ordered language power set; generating a corresponding fuzzy evaluation matrix based on the fusion result; calculating the state evaluation results of samples based on the fuzzy evaluation matrix by using the method of distance between the best and the worst solutions. The present invention adopts a more reasonable ordered language power set for the ordered symmetric language evaluation information given by experts, excluding evaluation situations that are impossible to occur in reality. In terms of measuring the uncertainty of language evaluation information, the present invention comprehensively considers the uncertainty of the ordered language evaluation degree words themselves and the uncertainty of language evaluation information, so that the measurement accuracy is higher. By introducing the method of distance between the best and the worst solutions, the present invention realizes a simpler framework from language evaluation information to finally giving state ranking. Description of the Drawings

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0047] Figure 1 It is a schematic diagram of the working principle of the state evaluation algorithm provided by the embodiment of the present invention.

[0048] Figure 2 Schematic diagram of the identification framework provided by the embodiment of the present invention.

[0049] Figure 3 Schematic diagram of the working process of the evaluation information uncertainty measurement method provided by the embodiment of the present invention. Detailed implementation manners

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0051] The embodiment of the present invention discloses a chemical equipment status evaluation method based on expert evaluation information, including:

[0052] Obtain expert linguistic evaluation information, analyze the expert linguistic evaluation information, and generate an ordered linguistic power set;

[0053] Perform uncertainty measurement and fusion on the ordered linguistic power set;

[0054] Generate a corresponding fuzzy evaluation matrix based on the fusion result;

[0055] Based on the fuzzy evaluation matrix, calculate the status evaluation result of the sample by using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS).

[0056] Specifically, the embodiment of the present invention first fuses the obtained linguistic evaluation information of multiple experts, secondly, generates a corresponding evaluation fuzzy matrix, and finally calculates the final status ranking result based on the evaluation matrix. The working principle of the corresponding status evaluation algorithm is as Figure 1 shown.

[0057] Step 1: Obtain expert linguistic evaluation information. By analyzing the expert linguistic evaluation information, establish a corresponding identification framework and a corresponding evaluation index system, as Figure 2 shown.

[0058] From Figure 2In this embodiment of the present invention, evaluation degree words such as "very good", "good", and "medium" are converted into corresponding indexes and values, so as to prepare for subsequent calculation and analysis. Due to the particularity of ordered evaluation language, jumpy language evaluation is impossible. For example, the situation of "{very poor, medium} with a probability of 0.3" is impossible. Only "{very poor, poor, medium} with a probability of 0.3" is possible. Because if both "very poor" and "medium" occur, then "poor" within the range between the two will surely occur. From the above analysis, the ordered power set P under the corresponding ordered language identification framework is obtained. Ω Form.

[0059] Specifically, analyzing the expert language evaluation information to generate an ordered language power set, including:

[0060] Establishing a corresponding identification framework and a corresponding evaluation index system for the expert language evaluation information;

[0061] Generating an ordered language power set P under the ordered language identification framework Ω The formula is as follows:

[0062]

[0063] Wherein, {ν i , ν j} o = {ν i , ν i+1 ,..., ν j-1 , ν j} (1 ≤ i ≤ j ≤ n) represents that the elements in the subset are continuous; and there are no jumpy elements, and the number of corresponding power sets is

[0064] Since the identification framework numerical form Ω = {-χ, -χ + 1,..., 0,..., χ - 1, χ} of this embodiment of the present invention is based on an ordered language set, the number of corresponding power sets is (χ + 1)(2χ + 1).

[0065] For ordered language degree measurement, it is often calculated based on the deviation degree between the degree word and the maximum degree word. The traditional measurement formula is as follows:

[0066]

[0067] Wherein, δ ν represents a certain language degree, the corresponding value of δ ν is ν, and χ is the value corresponding to the maximum language degree.

[0068] In order to more comprehensively measure the linguistic evaluation information, an embodiment of the present invention proposes a measurement method that comprehensively considers uncertainty, and the corresponding schematic diagram of the comprehensive entropy work is as shown in Figure 3 shown.

[0069] Specifically, the uncertainty measurement and fusion of the ordered linguistic power set include:[[]]

[0070] Entropy H of the uncertainty of the linguistic degree word itself L and entropy H of the evaluation uncertainty E are respectively set with corresponding coefficients to form the corresponding comprehensive entropy H;

[0071] Calculate the evaluation information matrix given by each expert for q features of m samples;

[0072] Based on the comprehensive entropy H, calculate the weight of each expert;

[0073] Perform weighted fusion calculation on the corresponding items in the evaluation information matrices of all experts.[[]]

[0074] In addition to measuring the uncertainty of the linguistic degree word itself, the comprehensive entropy of the embodiment of the present invention also measures the uncertainty of the evaluation. The embodiment of the present invention sets corresponding coefficients for the entropy H L of the uncertainty of the linguistic degree word itself and the entropy H E of the evaluation uncertainty to form the corresponding comprehensive entropy H, and the corresponding formula is as follows:

[0075] H = λH L + (1 - λ)H E (3)

[0076]

[0077] where λ is an adjustable parameter, |Ω| represents the cardinality of the set Ω, and m(A) represents the basic probability distribution value corresponding to A; A i , B i are subsets belonging to the ordered linguistic power set P Ω ;

[0078] Suppose there are n experts, and the evaluation matrix given by each expert for q features of m samples is as follows:

[0079]

[0080] where ψ ij is the evaluation information given by the corresponding expert for the i-th feature of sample j. For example: ψ ij is ({good, medium}, 0.5).

[0081] Next, based on the comprehensive entropy H proposed in the embodiments of the present invention, the weight of each expert is calculated:

[0082]

[0083] In the formula, τ k represents the unnormalized expert weight;

[0084]

[0085] Finally, weighted fusion calculation is performed on the corresponding items in the evaluation information matrix of all experts:

[0086]

[0087] In the formula, σ k represents the weight coefficient, (Tx k ) ij represents the evaluation matrix given by the k-th expert, and the suffix ij represents the j-th sample of the i-th feature, as shown in formula (6).

[0088] Specifically, based on the fusion result, a corresponding fuzzy evaluation matrix is generated, including:

[0089] Calculate the comprehensive entropy H of the fusion result and obtain the corresponding comprehensive entropy matrix;

[0090] Calculate the weight corresponding to each feature and generate the corresponding evaluation matrix;

[0091] Based on the corresponding evaluation matrix, generate a weighted average matrix.

[0092] In a specific embodiment of the present invention, in step two, a fuzzy evaluation matrix is generated. First, calculate the comprehensive entropy H of L ij obtained by formula (10) and obtain the corresponding comprehensive entropy matrix T F :

[0093]

[0094] Among them

[0095] Then, calculate the weight corresponding to each feature:

[0096]

[0097] Among them,

[0098] Then, generate the corresponding evaluation matrix T:

[0099]

[0100] Among them, φ(δ ν ) represents the value corresponding to δ ν . For example, if δ1 corresponds to "good", then φ(δ1) = 1.

[0101] Finally, the weighted average matrix is calculated as follows:

[0102] T ij represents the evaluation matrix corresponding to the i-th feature of sample j; represents the weight corresponding to each feature.

[0103] Specifically, based on the fuzzy evaluation matrix, the state evaluation result of the sample is calculated by using the method of distance between ideal solution and negative-ideal solution, including:

[0104] Step 3: Calculate the sorting result. Based on the weighted average matrix, the corresponding ideal solution Q + and the negative-ideal solution Q - are calculated as follows:

[0105]

[0106] Among them, represents the weighted average matrix; T i + , T i - respectively represent the maximum and minimum values of the samples in the i-th feature;

[0107] The distance value d(Q j , Q + ) between each element in the weighted average matrix and the ideal solution and the distance value d(Q j , Q - ) between each element and the negative-ideal solution are calculated as follows:

[0108]

[0109] Among them, q represents the total number of features;

[0110] According to the distance value d(Q j , Q + ) between each element in the weighted average matrix and the ideal solution and the distance value d(Q j , Q - ) between each element and the negative-ideal solution, the relative similarity of the corresponding sample j is calculated:

[0111]

[0112] Because the relative similarity S jThe larger it is, the smaller the distance between sample j and the optimal solution, that is, the more optimal the corresponding sample j is. Therefore, according to the relative similarity, the samples can be sorted, and finally the state evaluation results of the corresponding samples can be obtained.

[0113] In a specific embodiment of the present invention, 2 samples (two chemical equipment), 2 experts, and 3 features (temperature, pressure, vibration) are selected;

[0114] Among them, m(good)=xx represents the probability of "evaluated as good", as shown in Table 1 and Table 2 specifically:

[0115] Table 1 Evaluation Information of Expert 1 Tx1

[0116]

[0117] Table 2 Evaluation Information of Expert 2 Tx2 is as follows

[0118]

[0119] Specifically, the calculation method of the standardized expert weight σ k is as shown in formula (7), and the expert weights are obtained: [0.56357025, 0.43642975];

[0120] Comprehensive entropy feature matrix T F The calculation method is as shown in formula (10),

[0121] T F =[[0.63559934, 0.48508562]

[0122] [0.6233005, 0.56220033]

[0123] [0.51347174, 0.6145719]];

[0124] Feature weight Corresponding to formula (11),

[0125]

[0126] Evaluation matrix T ij Corresponding to formula (12),

[0127] T ij =[[0.23731628, 0.18111837]

[0128] [0.17148514, 0.154675]

[0129] [0.18048615, 0.21602302]];

[0130] Relative similarity S j Corresponding to formula (16), S j = [0.67260679, 0.32739321];

[0131] Therefore, the similarity of sample device 1 is higher than that of device 2, which means the state of device 1 is better than that of device 2.

[0132] An embodiment of the present invention discloses a chemical equipment state evaluation system based on expert evaluation information, including:

[0133] An information analysis module, configured to obtain expert language evaluation information, analyze the expert language evaluation information, and generate an ordered language power set;

[0134] A fusion module, configured to perform uncertainty measurement and fusion on the ordered language power set;

[0135] A matrix generation module, configured to generate a corresponding fuzzy evaluation matrix based on the fusion result;

[0136] A state evaluation module, configured to calculate the state evaluation result of the sample by using the method for calculating the distance between the ideal solution and the negative ideal solution based on the fuzzy evaluation matrix.

[0137] In this specification, each embodiment is described in a progressive manner. The key points of each embodiment are the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For the relevant parts, please refer to the description of the method part.

[0138] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for evaluating the state of chemical equipment based on expert evaluation information, characterized in that Including: Obtain expert language evaluation information, analyze the expert language evaluation information, and generate an ordered language power set; Perform uncertainty measurement and fusion on the ordered language power set; Generate a corresponding fuzzy evaluation matrix based on the fusion result; Based on the fuzzy evaluation matrix, calculate the state evaluation result of the sample by using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS); 2. The chemical equipment status evaluation method based on expert evaluation information according to claim 1, characterized in that, Analyze the expert language evaluation information to generate an ordered language power set, including: Establish a corresponding identification framework and a corresponding evaluation index system for the expert language evaluation information; Generate the ordered language power set P under the ordered language recognition framework Ω The formula is as follows: Among them, {ν i , ν j} o = {ν i , ν i+1 ,..., ν j-1 , ν j}(1 ≤ i ≤ j ≤ n) means that the elements in the subset are all consecutive; If the numerical form of the identification framework Ω = {−χ, −χ + 1,..., 0,..., χ − 1, χ} is based on an ordered language set, then the number of corresponding power sets is (χ + 1)(2χ + 1).

3. The chemical equipment status evaluation method based on expert evaluation information according to claim 1, characterized in that Perform uncertainty measurement and fusion on the ordered language power set, including: Entropy H of the uncertainty of the language degree word itself L and entropy H of the evaluation uncertainty E respectively set corresponding coefficients to form the corresponding comprehensive entropy H; Calculate the evaluation information matrix given by each expert for q features of m samples; Calculate the weight of each expert based on the comprehensive entropy H; Perform weighted fusion calculation on the corresponding items in the evaluation information matrices of all experts.

4. The chemical equipment status evaluation method based on expert evaluation information according to claim 3, characterized in that, Generate a corresponding fuzzy evaluation matrix based on the fusion result, including: Calculate the comprehensive entropy H for the fusion result and obtain a corresponding comprehensive entropy matrix; Calculate the weight corresponding to each feature and generate a corresponding evaluation matrix; Generate a weighted average matrix based on the corresponding evaluation matrix.

5. The chemical equipment status evaluation method based on expert evaluation information according to claim 4, characterized in that The weighted average matrix is expressed as: T ij represents the evaluation matrix corresponding to the i-th feature of sample j; represents the weight corresponding to each feature.

6. The chemical equipment status evaluation method based on expert evaluation information according to claim 1, characterized in that Based on the fuzzy evaluation matrix, calculate the state evaluation result of the sample by using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), including: The corresponding optimal solution Q is calculated based on the weighted average matrix + and the worst solution Q - : Among them, represents the weighted average matrix; T i + and T i - respectively represent the maximum and minimum values of the sample appearance in the i-th feature; Calculate the distance values d(Q j , Q + ) from each element in the weighted average matrix to the optimal solution and the distance value d(Q j , Q - ) from the worst solution: where q represents the total number of features; According to the distance values d(Q j , Q + ) between each element in the weighted average matrix and the optimal solution, and the distance value d(Q j , Q - ) between the worst solution, calculate the relative similarity of the corresponding sample j: Relative similarity S j The larger it is, the smaller the distance between sample j and the optimal solution, and the better the corresponding sample j is.

7. The chemical equipment status evaluation method based on expert evaluation information according to claim 6, characterized in that According to the size of the relative similarity S j sort the samples to obtain the status evaluation results of the corresponding samples.

8. A chemical equipment status evaluation system based on expert evaluation information, characterized in that, Including: An information analysis module, configured to obtain expert language evaluation information, analyze the expert language evaluation information, and generate an ordered language power set; A fusion module, configured to perform uncertainty measurement and fusion on the ordered language power set; A matrix generation module, configured to generate a corresponding fuzzy evaluation matrix based on the fusion result; A state evaluation module, configured to calculate the state evaluation result of the sample by using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) based on the fuzzy evaluation matrix.

Images