A complex electromechanical system health assessment method and system considering interpretability

Abstract

The present invention relates to a complex electromechanical system health assessment method and system considering interpretability. The method comprises: constructing a complex electromechanical system health assessment model based on a confidence rule base; defining search intensity, rule activation factor and interpretable distribution constraint respectively according to expert knowledge; selecting activated parameters in the complex electromechanical system health assessment model according to the rule activation factor; establishing an objective function according to the interpretable distribution constraint and the activated parameters; optimizing the activated parameters according to the search intensity, the interpretable distribution constraint and the objective function to obtain an optimized complex electromechanical system health assessment model; inputting complex electromechanical system data into the optimized complex electromechanical system health assessment model for evaluation to obtain a complex electromechanical system health assessment state. The present invention improves the evaluation accuracy while ensuring the interpretability of the model, making the evaluation result more reliable.

Description

Technical Field

[0001] The present invention relates to the field of complex electromechanical system health assessment, and in particular to a complex electromechanical system health assessment method and system taking interpretability into consideration. Background Art

[0002] In the process of modern industrialization in my country, the supporting role of complex electromechanical systems represented by high-end CNC machine tools and aircraft engines in economic development has become increasingly prominent, and their reliability and safety have received widespread attention. Health assessment methods can identify the weak links and hidden dangers of complex systems, providing a basis for improving the reliability of the system. However, unreliable and black-box assessment processes may lead to potential risks, such as unreasonable maintenance decisions and delayed alarm information. Therefore, how to effectively integrate multi-source information such as expert knowledge and monitoring data, and accurately and reliably assess the health status of complex electromechanical systems in an interpretable way has become an urgent problem to be solved in various fields.

[0003] Existing research mainly uses three types of health assessment models: 1) Data-driven models based on a large number of samples, such as support vector machines (SVM) and artificial neural networks (ANN). Due to the opaque modeling process of this type of model, the rationality of its results is difficult to convince people; 2) White box models built based on mechanism information. This type of model can provide a transparent modeling process and interpretable results. However, due to the challenges of extracting precise mathematical expressions from complex systems, the establishment and application of white box models become difficult to achieve. 3) Gray box models based on limited expert knowledge and data samples. This type of model can achieve good modeling accuracy and interpretability to a certain extent. The confidence rule base (BRB) is a typical gray box model that can effectively handle uncertainty and realize health assessment of complex electromechanical systems.

[0004] Expert knowledge is an important source of interpretability for BRB-based health assessment models. However, there are three problems in the model optimization process, which may undermine the interpretability of the model and lead to unreliable output results. 1) Expert knowledge is not effectively utilized in the optimization process; 2) The optimized rules may be inconsistent with common sense; 3) Some rules are over-optimized, which will change the initial judgment of experts. Therefore, in order to accurately and reliably implement health assessment of complex electromechanical systems, the above three aspects must be fully considered. Summary of the invention

[0005] The purpose of the present invention is to provide a complex electromechanical system health assessment method and system taking interpretability into consideration, thereby improving the assessment accuracy while ensuring the interpretability of the model and making the assessment results more reliable.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A complex mechatronic system health assessment method considering interpretability, including:

[0008] Constructing a health assessment model for complex electromechanical systems based on a trust rule base;

[0009] Based on expert knowledge, the search intensity, rule activation factor and interpretable distribution constraints are defined respectively;

[0010] According to the rule activation factor, selecting the activated parameters in the complex electromechanical system health assessment model;

[0011] Establishing an objective function according to the interpretable distribution constraints and the activated parameters;

[0012] Optimizing the activated parameters according to the search intensity, the interpretable distribution constraints and the objective function to obtain an optimized complex electromechanical system health assessment model;

[0013] The complex electromechanical system data is input into the optimized complex electromechanical system health assessment model for assessment to obtain the complex electromechanical system health assessment status.

[0014] Optionally, the building of a complex electromechanical system health assessment model based on a confidence rule base specifically includes:

[0015] Building a health assessment model for complex electromechanical systems based on a trust rule base:

[0016] with a rule weightθ k andattributeweightsδ i (k=1,2,...,L; i=1,2,...,T k )in, Represents the monitoring index of complex electromechanical systems, represents the reference value of the i-th indicator, θ k is the weight of the kth rule, which reflects the importance of this rule relative to other rules; i (i=1,2,...,T k ) represents the weight of the i-th indicator, which is used to characterize its importance relative to other indicators; L represents the number of rules, β nk (n=1,2,...,N) represents the reference level D n confidence level.

[0017] Optionally, the search strength is defined as:

[0018]

[0019] Among them, P(Ω) represents the search intensity, which is used to describe the sampling probability in the optimization process; P(Ω) satisfies the two properties of regularity and unimodal distribution. represents the credibility of expert knowledge, and σ is the covariance matrix, which is used to reflect and adjust the credibility of expert knowledge.

[0020] Optionally, the interpretable distribution constraint is defined as:

[0021] β k ~C k ,(k=1,...,L)

[0022]

[0023] Among them, C k represents the explainability distribution, β k represents the confidence of the kth rule relative to a certain health level, N represents the number of health levels, and L represents the total number of rules.

[0024] Optionally, the rule activation factor is defined as:

[0025]

[0026] Among them, W k Represents the vector constructed by the activation weight of the kth rule, assuming that the size of the dataset is P, W k Represented as W k =(w 1 ,...,w p ,...,w P ), k=1,...,L; p=1,...,P.

[0027] Optionally, selecting the activated parameters in the complex electromechanical system health assessment model according to the rule activation factor specifically includes:

[0028] activating the rule according to the rule activating factor;

[0029] When the activation rules are determined, the corresponding parameters are selected as optimization vectors;

[0030] The optimization vector is expressed as in, and Indicates the activated parameter.

[0031] Optionally, the objective function is:

[0032]

[0033] in, represents the output error of the model, It is Interpretability constraints on the confidence distribution in the rules.

[0034] Optionally, optimizing the activated parameters according to the search intensity, the interpretable distribution constraint and the objective function to obtain an optimized complex electromechanical system health assessment model specifically includes:

[0035] The activated parameters are optimized using a differential evolution algorithm according to the search intensity, the explainable distribution constraints and the objective function to obtain an optimized complex electromechanical system health assessment model.

[0036] A complex mechatronic system health assessment system considering interpretability, including:

[0037] Complex electromechanical system health assessment model building module, used to build a complex electromechanical system health assessment model based on a confidence rule base;

[0038] A definition module is used to define the search intensity, rule activation factor and interpretable distribution constraints respectively based on expert knowledge;

[0039] An activated parameter determination module, used to select activated parameters in the complex electromechanical system health assessment model according to the rule activation factor;

[0040] An objective function establishment module, used for establishing an objective function according to the interpretable distribution constraint and the activated parameters;

[0041] An optimization module, used for optimizing the activated parameters according to the search intensity, the interpretable distribution constraint and the objective function to obtain an optimized complex electromechanical system health assessment model;

[0042] The health status assessment module is used to input the complex electromechanical system data into the optimized complex electromechanical system health assessment model for assessment to obtain the health status of the complex electromechanical system.

[0043] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0044] The present invention firstly constructs a complex electromechanical system health assessment model based on a confidence rule base, which can effectively combine expert knowledge and monitoring data; then, in order to ensure the interpretability of the model during the optimization process, the search intensity, rule activation factor and interpretable distribution constraints are defined respectively according to expert knowledge; finally, an optimization model considering interpretability is constructed, and the model is optimized using monitoring data, which ensures the interpretability of the model while improving the evaluation accuracy, making the evaluation results more reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A flow chart of a complex electromechanical system health assessment method considering interpretability for the present invention;

[0047] Figure 2 A structural diagram of a complex electromechanical system health assessment system considering interpretability for the present invention;

[0048] Figure 3 It is the original vibration data and characteristic diagram;

[0049] Figure 4 Schematic diagram of evaluation results and confidence distribution;

[0050] Figure 5 This is a schematic diagram for comparing rule weights;

[0051] Figure 6 This is a comparison chart of confidence distribution. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] The purpose of the present invention is to provide a complex electromechanical system health assessment method and system taking interpretability into consideration, thereby improving the assessment accuracy while ensuring the interpretability of the model and making the assessment results more reliable.

[0054] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] Figure 1 The flowchart of the complex electromechanical system health assessment method considering interpretability of the present invention is shown in FIG. Figure 1 As shown in Figure 1, a complex electromechanical system health assessment method considering interpretability includes:

[0056] Step 101: Construct a complex electromechanical system health assessment model based on a confidence rule base, including:

[0057] Building a health assessment model for complex electromechanical systems based on a trust rule base:

[0058]

[0059] with a rule weightθ k and attribute weightsδ i (k=1,2,...,L; i=1,2,...,T k ) (1)

[0061] in, Represents the monitoring index of complex electromechanical systems, represents the reference value of the i-th indicator, θ k is the weight of the kth rule, which reflects the importance of this rule relative to other rules; i (i=1,2,...,T k ) represents the weight of the i-th indicator, which is used to characterize its importance relative to other indicators; L represents the number of rules, β nk (n=1,2,...,N) represents the reference level D n confidence level.

[0062] The reasoning steps of the complex electromechanical system health assessment model based on the confidence rule library are:

[0063] Step 1: Convert quantitative and qualitative information into a confidence distribution:

[0064] S(x i )={(A i,j ,a i,j ),i=1,...,T k ; j = 1, ..., J i} (2)

[0065] Among them, A i,j Represents input x i The jth reference value of a. i,j Indicates A i,j degree of matching.

[0066] Step 2: Calculate the rule activation weight as follows:

[0067]

[0068] Among them, θ k ∈[0,1] is the weight of the kth rule. is x i The normalized weight of . L is the number of rules. Indicates that the input x in the kth rulei Relative to its jth reference value A i,j degree of matching.

[0069] Step 3: Generate the final confidence using the evidence reasoning algorithm as follows:

[0070]

[0071] Step 4: The final distributed result is expressed as:

[0072] S(X)={(D n ,β n ); n=1,...,N} (5)

[0073] Where X represents the actual input vector. S(X) represents the health status, and its utility can be calculated by the following formula:

[0074]

[0075] Step 102: Define the search intensity, rule activation factor and interpretable distribution constraints respectively based on expert knowledge.

[0076] In the complex electromechanical system health assessment method considering interpretability of the present invention, in order to effectively utilize expert knowledge in the model optimization process, the search intensity is defined as:

[0077]

[0078] In the above formula, P(Ω) represents the search intensity, which is used to describe the sampling probability during the optimization process. P(Ω) satisfies the two properties of regularity and unimodal distribution. Represents the credibility of expert knowledge. σ is the covariance matrix, which can be used to reflect and adjust the credibility of expert knowledge. Expert knowledge accumulated in long-term practice is the basis of model interpretability. The above search intensity ensures the characteristics of local optimization and can effectively improve the modeling accuracy of the BRB-based health assessment model under the interpretability requirement.

[0079] In the complex electromechanical system health assessment method considering interpretability of the present invention, in order to ensure the consistency of rules and common sense, the interpretable distribution constraint is defined as:

[0080] β k ~C k ,(k=1,...,L) (8)

[0081] In the field of health assessment, conflicting health states cannot obtain high confidence at the same time. Therefore, the interpretable distribution of the rule conclusion part can be specifically as follows:

[0082]

[0083] The complex electromechanical system health assessment method considering interpretability of the present invention comprises the following steps of calculating and applying the rule activation factor:

[0084] Generally speaking, the expert knowledge accumulated in long-term practice contains information about all working states of the real system, while the training data set is limited and can only contain information about some working states. That is, the initial BRB-based model is established based on considering all working states of the system, but the optimization process can only adjust the parameters related to some working states reflected in the data set. Therefore, it can be determined that only the parameters of the activation rules can be fine-tuned through the training data set. Over-optimization refers to the process of optimizing non-activated parameters, which may change the initial judgment information of the experts and lead to the loss of interpretability.

[0085] In order to effectively optimize the model parameters, it is necessary to distinguish the activated rules from all rules. The rule activation factor ω is used to mark the activated rules. If ω k =1, it means that the kth rule is activated. Therefore, the activation factor can be defined as:

[0086]

[0087] Among them, W k W represents the vector constructed by the activation weights of the kth rule. Assume that the size of the dataset is P. k It can be expressed as

[0088] W k =(w 1 ,...,w p ,...,w P ),k=1,...,L; p=1,...,P(11)

[0089] Step 103: Selecting activated parameters in the complex electromechanical system health assessment model according to the rule activation factor, specifically including:

[0090] activating the rule according to the rule activating factor;

[0091] When the activation rules are determined, the corresponding parameters are selected as optimization vectors;

[0092] The optimization vector is expressed as:

[0093]

[0094] in, and Indicates the activated parameter.

[0095] Step 104: Establish an objective function according to the interpretable distribution constraints and the activated parameters.

[0096] Consider the interpretability of the objective function. In order to achieve a reasonable optimization process, the objective function needs to be corrected. The corrected objective function is:

[0097]

[0098] in, represents the output error of the model, It is Interpretability constraints on the confidence distribution in the rules.

[0099] Step 105: Optimizing the activated parameters according to the search intensity, the interpretable distribution constraint and the objective function to obtain an optimized complex electromechanical system health assessment model specifically includes:

[0100] The activated parameters are optimized using a differential evolution algorithm according to the search intensity, the explainable distribution constraints and the objective function to obtain an optimized complex electromechanical system health assessment model.

[0101] Taking the differential evolution algorithm (DE) as an example, the search intensity is introduced in the initial operation, and the interpretability constraint is introduced in the newly added control operation. The modified DE algorithm is represented by P-DE-I, and its steps are as follows:

[0102] Step 1 (initialization operation): Generate an initial population based on the search intensity.

[0103]

[0104] in represents the initial parameter vector. The covariance matrix is ​​σ. λ represents the population size. P(·) represents the search intensity under normal distribution.

[0105] Step 2 (Mutation Operation): Generate a new solution using mutation operation, expressed as:

[0106]

[0107] Among them, p 1 ≠p 2 ≠p 3 ≠k. F∈[0,1] is the scaling factor. is the best individual of the gth generation. μ represents the center of the population, which is calculated by the following formula:

[0108]

[0109] Step 3 (crossover operation): Introduce a crossover operation to increase population diversity, expressed as

[0110]

[0111] Among them, Cr∈[0,1] represents the crossover rate. is the updated population.

[0112] Step 4 (control operation): Resample the error confidence distribution in the k-th solution of the g-th generation until all confidence distributions satisfy the distribution constraints, as described below:

[0113]

[0114] in, represents the kth solution of the gth generation, which may contain incorrect belief distribution. is a newly generated confidence distribution that satisfies the constraints.

[0115] Step 5 (Projection Operation): To satisfy the equality constraint, a projection operation is used to transform the equality constraint into an equality constraint in a hyperplane:

[0116]

[0117] The projection operation is as follows:

[0118]

[0119] Step 6 (selection operation): Perform the selection operation and update the best individual and population as follows:

[0120]

[0121] Step 7 (Termination Criteria): Go to step 2 until the maximum number of iterations is reached.

[0122] Step 106: Input the complex electromechanical system data into the optimized complex electromechanical system health assessment model for assessment to obtain the complex electromechanical system health assessment status.

[0123] The present invention also provides a complex electromechanical system health assessment system considering interpretability, such as Figure 2 As shown, the system comprises:

[0124] The complex electromechanical system health assessment model building module 201 is used to build a complex electromechanical system health assessment model based on a confidence rule base.

[0125] The definition module 202 is used to define the search intensity, the rule activation factor and the interpretable distribution constraint respectively according to the expert knowledge.

[0126] The activated parameter determination module 203 is used to select activated parameters in the complex electromechanical system health assessment model according to the rule activation factor.

[0127] The objective function establishing module 204 is used to establish an objective function according to the interpretable distribution constraints and the activated parameters.

[0128] The optimization module 205 is used to optimize the activated parameters according to the search intensity, the interpretable distribution constraints and the objective function to obtain an optimized complex electromechanical system health assessment model.

[0129] The health status assessment module 206 is used to input the complex electromechanical system data into the optimized complex electromechanical system health assessment model for assessment to obtain the health status of the complex electromechanical system.

[0130] Embodiment 1:

[0131] In this embodiment, a health assessment is performed on an aircraft engine. An aircraft engine is a sophisticated and complex dynamic system composed of electromagnetic units and mechanical units. Its health status is affected by high temperature, high pressure and strong stress changes, and is uncertain. Due to the particularity of aircraft engines, they cannot be tested frequently, so the test data samples obtained are limited. The health status of an aircraft engine is the key to aircraft safety. Using limited data samples and knowledge to make an interpretable and reliable estimate of the health status of the engine can reduce safety hazards in engineering practice. In this experiment, using the internal vibration sensor of the aircraft engine, 5,000 sets of vibration data were collected when the high-altitude valve spring pressure was reduced by 0%, 5%, 10% and 20% respectively (the corresponding health states are normal state (N), minor fault state (S), moderate fault state (M) and severe fault state (SE). The original vibration data is as follows. Figure 3 -(a) shows that the kurtosis and skewness of vibration data are used as features to describe the health status of aircraft engines:

[0132]

[0133] For every 100 data, the features are calculated using formula (15), such as Figure 3 -(b) and 3-(c). In each state, 30 data are selected as training sets and all data are used as test sets.

[0134] The specific implementation steps for aircraft engine health assessment are as follows:

[0135] Step 1: Build an initial evaluation model.

[0136] Degradation of key aircraft engine components such as valve springs can lead to slight changes in the engine vibration signal, which manifests as changes in kurtosis and skewness. In engineering practice, kurtosis and skewness are described by four semantic values: "low (L)", "medium (M)", "high (H)" and "very high (VH)". In addition, based on the knowledge accumulated from long-term aircraft engine tests, reference values ​​and initial attribute weights are determined, as shown in Table 1, which are consistent with the design principles.

[0137] Table 1 Attribute reference values ​​and weights

[0138]

[0139] In order to describe the relationship between the characteristics and the health status, it is necessary to give the initial rules through the analysis of the aircraft engine. In the normal state, the kurtosis and skewness of the aircraft engine vibration are both at the VH level. In this case, the health status becomes worse because the degree of skewness becomes "M". When the skewness level is "M" and the kurtosis level is "L" or "M", the health status is at the "SE" level. In general, as the kurtosis and skewness levels decrease, the health status becomes worse. Based on the above empirical knowledge, the initial rules are shown in Table 2.

[0140] Table 2 Initial rule base

[0141]

[0142]

[0143] Step 2: Build an optimized model considering interpretability.

[0144] Due to the limitations of expert knowledge, it is necessary to use monitoring data to optimize the initial model. First, determine the final parameter vector In this case, the rule activation factor is calculated using the training set and formulas (10)-(12), which is expressed as

[0145] ω={ω 1 ,...,ω k} = {1,1,1,0,1,1,1,1,1,1,1,0,0,1,1} (16)

[0146] This means that rules 4, 13, and 14 will not be activated by the training set, that is, the corresponding parameters do not need to be optimized. The final parameter vector is expressed as

[0147]

[0148] The objective function is determined by equation (13), and the output error is calculated as the mean square error. The parameters of P-DE-I are shown in Table 3.

[0149] Table 3 P-DE-I parameters

[0150]

[0151] Step 3: Optimized model and test results.

[0152] The optimized evaluation model is shown in Table 4. The optimized attribute weights are 1 and 0.7864 respectively.

[0153] Table 4 Optimized evaluation model

[0154]

[0155]

[0156] In the testing part, the test dataset is taken as input. The health status of the evaluation is as follows Figure 4 -(a) shows that the mean square error is 0.0054, indicating that the optimized BRB can accurately estimate the health status of the aircraft engine. The confidence level of the corresponding health status is as follows Figure 4 -(b) shows that the aircraft engine can still operate normally in normal state and minor fault state. In order to verify the robustness of P-DE-I, the experiment was conducted 25 times. The variance of MSEs (1.826E-07) is much smaller than the mean of MSEs (0.00564).

[0157] Step 4: Comparative study.

[0158] In order to verify the effectiveness of the above model, a comparative study was carried out on the initial model (denoted by BRB0), the model optimized by the P-DE-I algorithm (denoted by BRB-1), the model optimized by the traditional DE algorithm (denoted by BRB-2), the fuzzy rule base model (FRB) and the extreme learning machine (ELM).

[0159] In terms of modeling accuracy, the corresponding mean square error is shown in Table 5. It can be seen that the optimized BRB models (BRB1, BRB2) have higher modeling accuracy than other models. Compared with FRB and ELM, the modeling accuracy of BRB1 is improved by 41.30% and 15.63%. BRB2 has similar modeling accuracy to BRB1.

[0160] Table 5 Evaluation errors of five models

[0161]

[0162] In terms of model interpretability, the rule weights and confidence distributions of BRB0, BRB1, and BRB2 are compared, as shown in Figure 5 and 6 As shown. Figure 4It can be seen that the rule weights in BRB1 are similar to those in BRB0, while most of the rule weights in BRB2 are far from those in BRB0. Taking the third rule as an example, the rule weights of BRB0 and BRB1 are 0.7 and 0.62 respectively, indicating that the third rule is very important. However, the rule weight of the third rule in BRB2 is 0.0278, indicating that the rule is not important. This is completely inconsistent with the expert's initial judgment. Figure 6 As shown, the confidence distributions in BRB0 and BRB1 are close. Most of the confidence distributions in BRB2 are different from those in BRB0. Some distributions, such as the confidence distribution in the eighth rule, do not meet the interpretability requirements. The above results show that BRB1 improves modeling accuracy while fine-tuning the parameters determined by expert knowledge, that is, the optimization process of BRB1 is based on the initial judgment of experts, which effectively ensures the interpretability of the evaluation process.

[0163] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0164] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A complex electromechanical system health assessment method considering interpretability, characterized in that: include: Constructing a health assessment model for complex electromechanical systems based on a trust rule base; Based on expert knowledge, the search intensity, rule activation factor and interpretable distribution constraints are defined respectively; The search intensity is defined as: Among them, represents the search intensity, which is used to describe the sampling probability in the optimization process; p(Ω) satisfies the two properties of regularity and unimodal p(Ω) distribution. represents the credibility of expert knowledge, σ is the covariance matrix, which is used to reflect and adjust the credibility of expert knowledge; The rule activation factor is defined as: Among them, W k Represents the vector constructed by the activation weight of the kth rule, assuming that the size of the dataset is P, W k Represented as W k =(w1,...,w p ,...,w P ),k=1,...,L; p=1,...,P; The interpretable distribution constraint is defined as: Among them, C k represents the explainability distribution, β k represents the confidence of the kth rule relative to a certain health level, and L represents the total number of rules; According to the rule activation factor, selecting the activated parameters in the complex electromechanical system health assessment model; Establishing an objective function according to the interpretable distribution constraints and the activated parameters; Optimizing the activated parameters according to the search intensity, the interpretable distribution constraints and the objective function to obtain an optimized complex electromechanical system health assessment model; The complex electromechanical system data is input into the optimized complex electromechanical system health assessment model for assessment to obtain the complex electromechanical system health assessment status.

2. The complex electromechanical system health assessment method considering interpretability according to claim 1 is characterized in that: The complex electromechanical system health assessment model is constructed based on the confidence rule base, specifically including: Building a health assessment model for complex electromechanical systems based on a trust rule base: with a rule weightθ k and attribute weightsδ i in, Represents the monitoring index of complex electromechanical systems, Indicates the reference value of the i-th indicator, and the value range of i is 1 to T k , the value of j ranges from 1 to J i θ k is the weight of the kth rule, which is used to reflect the importance of this rule relative to other rules. The value range of k is 1 to L; δ i represents the weight of the i-th indicator, which is used to characterize its importance relative to other indicators; L represents the number of rules, β nk Indicates reference level D n The confidence level of n ranges from 1 to N, where N represents the number of health levels.

3. The complex electromechanical system health assessment method considering interpretability according to claim 1 is characterized in that: The step of selecting the activated parameters in the complex electromechanical system health assessment model according to the rule activation factor specifically includes: activating the rule according to the rule activating factor; When the activation rules are determined, the corresponding parameters are selected as optimization vectors; The optimization vector is expressed as in, and Indicates the activated parameter.

4. The complex electromechanical system health assessment method considering interpretability according to claim 3 is characterized in that: The objective function is: in, represents the output error of the model, It is Interpretability constraints on the confidence distribution in the rules.

5. The complex electromechanical system health assessment method considering interpretability according to claim 1 is characterized in that: The step of optimizing the activated parameters according to the search intensity, the interpretable distribution constraint and the objective function to obtain an optimized complex electromechanical system health assessment model specifically includes: The activated parameters are optimized using a differential evolution algorithm according to the search intensity, the explainable distribution constraints and the objective function to obtain an optimized complex electromechanical system health assessment model.

6. A complex electromechanical system health assessment system considering interpretability, characterized in that: The complex electromechanical system health assessment system considering interpretability is applied according to the complex electromechanical system health assessment method considering interpretability as described in any one of claims 1 to 5, comprising: Complex electromechanical system health assessment model building module, used to build a complex electromechanical system health assessment model based on a confidence rule base; A definition module is used to define the search intensity, rule activation factor and interpretable distribution constraints respectively based on expert knowledge; An activated parameter determination module, used to select activated parameters in the complex electromechanical system health assessment model according to the rule activation factor; An objective function establishment module, used for establishing an objective function according to the interpretable distribution constraint and the activated parameters; An optimization module, used for optimizing the activated parameters according to the search intensity, the interpretable distribution constraint and the objective function to obtain an optimized complex electromechanical system health assessment model; The health status assessment module is used to input the complex electromechanical system data into the optimized complex electromechanical system health assessment model for assessment to obtain the health status of the complex electromechanical system.

Images