Data and knowledge driven circuit breaker electronic trip multi-parameter state determination method

Abstract

The application provides a data and knowledge driven multi-parameter state determination method for a circuit breaker electronic release, and relates to the technical field of electrical equipment state evaluation, which comprises the following steps: S1, setting the state of the circuit breaker electronic release, screening key performance parameters and screening degradation characteristics of the circuit breaker electronic release; S2, constructing a hierarchical electronic release state evaluation model to determine the degradation state of the multi-element electronic release; S3, optimizing the global parameters of the hierarchical electronic release state evaluation model to determine the state division threshold; and S4, constructing a double-path electronic release state determination model to determine the state of the electronic release. The application uses a hierarchical electronic release state evaluation framework to quantitatively evaluate the state of the low-voltage circuit breaker electronic release under multi-dimensional representation parameters, optimizes the hierarchical electronic release state evaluation model by using an optimization algorithm, and constructs a double-path fusion decision model to perform adaptive evaluation and fault identification on the state of the low-voltage circuit breaker electronic release.

Description

Technical Field

[0001] This invention relates to the field of electrical equipment condition assessment technology, specifically to a data and knowledge-driven method for determining the multi-parameter condition of an electronic trip unit of a circuit breaker. Background Technology

[0002] Low-voltage circuit breakers, as core control and protection equipment in low-voltage power distribution systems, are widely used in industrial, building, and transportation applications where high power supply reliability is required. Long-term continuous service and dynamic load fluctuations can lead to degradation of internal component coupling, reducing equipment reliability and directly impacting the stable operation of the power distribution system. Electronic trip unit condition assessment is a crucial means of quantifying equipment reliability, identifying early performance degradation, and predicting fault risks. Therefore, regularly conducting electronic trip unit condition assessments on low-voltage circuit breakers can promptly identify potential problems and effectively improve equipment operational reliability.

[0003] Currently, reliability research on low-voltage circuit breakers mainly focuses on mechanical operation and contact arc extinguishing systems, while research on electronic trip units is still in its early stages. In particular, the power module, as the core of its energy supply, is prone to degradation, directly causing functional unit abnormalities and becoming a major cause of controller failure. Therefore, selecting the power module as the research object can effectively characterize the overall electronic trip unit status of low-voltage circuit breakers. Current electronic trip unit status assessment methods often employ mechanistic analysis, however, this method is difficult to accurately model. While data-driven methods do not require complex physical modeling, the actual operating data of the electronic trip unit's power module is limited and uncertain. Interpretable knowledge-driven methods suffer from strong subjectivity in parameter setting, and knowledge blind spots easily lead to errors. Hybrid-driven methods reduce the limitations of single methods, but generally suffer from poor heterogeneous information fusion performance and insufficient handling of conflicting evidence from multiple sources of degradation features, failing to effectively adapt to the complex degradation characteristics of power modules. Therefore, a high-accuracy and highly applicable method for assessing the status of low-voltage circuit breaker electronic trip units is needed. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention aims to provide a data- and knowledge-driven method for determining the multi-parameter status of electronic trip units in circuit breakers. By quantifying the random uncertainties caused by component degradation and the cognitive uncertainties arising from conflicting evidence of multi-source degradation features, a hierarchical electronic trip unit status assessment framework based on dendritic neurons and evidence theory is used to quantitatively assess the status of low-voltage circuit breaker electronic trip units under multi-dimensional representation parameters. A global adaptive optimization model based on biogeographical optimization algorithms is employed to jointly optimize key parameters and electronic trip unit status classification thresholds in the hierarchical electronic trip unit status assessment model. Furthermore, a dual-path fusion decision model is constructed based on continuous state parameter assessment paths and rapid fault determination paths for out-of-bounds faults, enabling adaptive and refined assessment of the status of low-voltage circuit breaker electronic trip units and rapid identification of extreme faults.

[0005] Specifically, the present invention provides a data and knowledge-driven method for determining the multi-parameter status of an electronic trip unit of a circuit breaker, which includes the following steps: S1: Set the circuit breaker electronic trip unit status, perform sensitivity analysis to determine the sensitivity of component parameters. The degradation characteristics of the circuit breaker's electronic trip unit are screened as follows: the rate of change of capacitance of electrolytic capacitor C5, the rate of change of capacitance of electrolytic capacitor C9, the rate of change of on-resistance of the metal-oxide-semiconductor transistor, the rate of change of power supply output voltage amplitude, and the rate of change of switching cycle. The status levels of the electronic trip unit are then classified and identified, and a complete finite set of the electronic trip unit status identification framework is established. ; S2: Construct a hierarchical electronic trip unit (EPU) state assessment model based on dendritic neurons and Dempster-Shafer evidence theory; construct the EPU state... The basic probability assignment function is obtained by refining the confidence level and obtaining the basic probability assignment function for a single feature. ; by sequentially performing Dempster combinations, the fusion basic probability assignment function is obtained. Perform state confidence probability transformation and state measurement of the electronic trip unit to generate continuous state assessment parameters for the electronic trip unit. ; S3: Assume the global parameter set to be optimized in the model. The objective function for electronic trip unit (EPU) status assessment was determined using a biogeographical optimization algorithm. The hierarchical EPU status assessment model obtained in S2 was optimized to obtain the set of continuous status assessment parameters corresponding to the EPU. Determine the electronic trip unit state classification threshold based on the principle of minimum misclassification; determine the optimal boundary threshold between three groups of adjacent electronic trip unit states. The set of thresholds for classifying the state of the electronic trip unit is obtained. ; S4: Map the continuous state evaluation parameters in S3 to the corresponding electronic trip unit state levels, and output the electronic trip unit state determination results corresponding to the training samples. Based on engineering rule constraints, a rapid fault determination path is obtained to determine if the current equipment is in a fault state; dual-path decision determines the status of the electronic trip unit.

[0006] Preferably, step S2 specifically includes: S21: Determine the Gaussian membership function corresponding to the state of the electronic trip unit, obtain the matching degree of the degradation feature state of each electronic trip unit state to the i-th degradation feature, and use evidence theory to construct the basic probability assignment function and correct the credibility. S22: Integrate the basic probability allocation function of multi-source features; determine the conflict coefficient K and clarify the state of the electronic trip unit. The corresponding basic probability allocation values; fusion yields the state-to-be-assigned evidence items. The basic probability assignment values ​​are obtained by sequentially combining the basic probability assignment functions of the individual features corresponding to the five degenerate features to obtain the fused basic probability assignment function. ; S23: Perform confidence-transformation probability transformation on the fusion basic probability allocation function. Based on the confidence-transformation probability of the electronic trip unit's state and the state scale, determine the membrane output potential M and generate continuous state evaluation parameters for the electronic trip unit. .

[0007] Preferably, the method for obtaining the degradation characteristic state matching degree of the electronic trip unit state in step S21 is as follows: ; ; in, Let be the degree of matching between the i-th degradation feature and the j-th type of electronic trip state; Here are the denominator calibration parameters; j is the electronic trip unit status number; Let be the membership degree of the i-th degradation feature to the j-th type of electronic trip state; The normalized monitoring feature value of the i-th term; The mean of the Gaussian membership function corresponding to the state of the electronic trip unit of the i-th degradation feature is denoted as ; The standard deviation of the Gaussian membership function corresponding to the state of the j-th type of electronic trip unit for the i-th degradation feature; is the temperature coefficient; e is the natural constant.

[0008] Preferably, the method for obtaining the comprehensive basic probability allocation function in step S22 is as follows: ; in, To fuse the basic probability assignment function; To integrate the basic probability assignment function for the state term The basic probability allocation value; The basic probability assignment function for each of the five degenerate features; The basic probability assignment function corresponding to the first degenerate feature is applied to the state term. The basic probability allocation value; The basic probability assignment function corresponding to the second degenerate feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the third degenerate feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the fourth degradation feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the fifth degradation feature for the state term The basic probability allocation value; This refers to the state item to be evaluated.

[0009] Preferably, the method for obtaining the continuous state evaluation parameters of the electronic trip unit in step S23 is as follows: ; , ; in, Parameters for continuous state evaluation of electronic trip units; State of the j-th type electronic trip unit The probability of confidence conversion; To integrate the basic probability assignment function for the state term The basic probability allocation value; To integrate the basic probability assignment function with the evidence items to be assigned to the state The basic probability distribution value; M is the output membrane potential of the membrane layer; This is the state scale corresponding to the state of the j-th type of electronic trip unit; The slope parameter of the activation function; This is the activation threshold parameter.

[0010] Preferably, step S3 specifically includes: S31: Assume the global parameter set to be optimized in the model. With the goal of maximizing the accuracy of electronic trip unit status assessment and minimizing the average conflict degree, an objective function for comprehensive optimization is established. S32: Initialize and set up the biogeographical optimization algorithm; update parameters for non-elite habitats based on migration rate, perform habitat migration and mutation, and redetermine the objective function for electronic trip unit state assessment; backfill the optimal parameter solution set into the hierarchical electronic trip unit state assessment model to obtain the continuous state assessment parameter set corresponding to the electronic trip unit. ; S33: Determine the electronic trip unit state division threshold based on the minimum misclassification principle; determine the optimal boundary threshold between three groups of adjacent electronic trip unit states. The set of thresholds for classifying the state of the electronic trip unit is obtained. .

[0011] Preferably, the objective function for minimizing the average conflict degree in step S31 is as follows: ; ; in, To minimize the objective function; λ is the average parameter conflict coefficient; λ is the conflict penalty weight. To improve the accuracy of electronic trip unit status assessment; Here, N is the indicator function; N is the total number of training samples. The actual electronic trip unit status label for the nth training sample; Predict the status tag for the electronic trip unit; The global parameter set to be optimized.

[0012] Preferably, the method for obtaining the optimal boundary threshold in step S33 is as follows: ; ; in, The optimal boundary threshold for the state; To minimize the number of misclassified samples; Candidate threshold selection operator that obtains the minimum value; The number of misclassified samples corresponding to the candidate boundary threshold t; For the continuous state evaluation parameters corresponding to the nth sample; Degenerate state The set of continuous state evaluation parameters; Degenerate state The set of continuous state evaluation parameters; m is the state category of the electronic trip unit.

[0013] Preferably, step S4 specifically includes: S41: Determine the electronic trip unit status assessment path based on the continuous status assessment parameters; obtain the corresponding continuous status assessment parameters using the electronic trip unit status assessment model, and map them to the corresponding electronic trip unit status levels; obtain the electronic trip unit status determination results corresponding to the training samples. ; S42: Obtain a rapid fault determination path based on engineering rule constraints; define the degradation feature vector of the sample to be evaluated. Set limit boundary thresholds for key degradation features. The device is determined to be in a faulty state. S43: Dual-path decision-making determines the final electronic trip unit status assessment result. .

[0014] Preferably, in step S41, the mapping is to the corresponding electronic trip unit status level, specifically as follows: ; in, The result of the electronic trip unit status determination; The electronic trip unit is in operation. In detection status; This is an abnormal state; The condition is faulty; This is the first dividing threshold; This is the second boundary threshold; This is the third dividing threshold.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention characterizes the state of the electronic trip unit of the circuit breaker by extracting multidimensional degradation features; through sensitivity simulation analysis and engineering reliability constraints of power module components, five key degradation features such as the change rate of electrolytic capacitor value and the change rate of conduction resistance are extracted from the power module, overcoming the problem that the traditional electronic trip unit state assessment method relies on a single monitoring parameter and has limited state characterization performance, thereby improving the effectiveness of the state assessment of the electronic trip unit of the low voltage circuit breaker.

[0016] (2) This invention uses a hierarchical electronic trip unit state assessment model to enhance uncertainty handling performance; in response to random uncertainty caused by fluctuations in monitoring data and cognitive uncertainty caused by conflicting evidence of multi-source degradation features, a hierarchical electronic trip unit state assessment model based on dendritic neuron DNM and Dempster-Shafer evidence theory is constructed. At the synaptic layer, random uncertainty is characterized by constructing a basic probability allocation function and credibility correction. At the dendritic layer, multi-source degradation features are fused through the Dempster combination rule and the Dempster combination rule, which effectively alleviates the cognitive uncertainty caused by conflicting evidence of multi-source degradation features.

[0017] (3) This invention improves the accuracy of electronic trip unit status assessment through global optimization and dual-path decision model; adopts biogeographical optimization algorithm to globally adaptively optimize the key parameters of the hierarchical electronic trip unit status assessment model and the electronic trip unit status division threshold, thereby improving the model's adaptability to actual degradation modes; and combines the dual-path decision model of continuous status parameter assessment path and out-of-bounds fault rapid judgment path to reduce the risk of misjudgment of electronic trip unit status. Attached Figure Description

[0018] Figure 1 The flowchart shows the data and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers proposed in this invention. Figure 2 This is an overall framework diagram of the electronic trip unit status assessment method of the present invention; Figure 3 This is a circuit diagram of the power supply module for the low-voltage circuit breaker electronic trip unit of the present invention; Figure 4 This is a schematic diagram of the hierarchical electronic trip unit state assessment model based on dendritic neurons and evidence theory of the present invention. Figure 5 This is a diagram showing the original data for the five features of this invention; Figure 6 This is a graph showing the accuracy of electronic trip unit status assessment during the training process of this invention. Figure 7 This is an error diagram showing the status evaluation results of the electronic trip unit of the present invention. Detailed Implementation

[0019] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0020] This invention provides a data- and knowledge-driven method for determining the multi-parameter status of an electronic circuit breaker trip unit. The overall process is as follows: Figure 1 As shown, the circuit breaker electronic trip unit status is set, key performance parameters are screened, and degradation characteristics of the circuit breaker electronic trip unit are screened; a hierarchical electronic trip unit status assessment model is constructed to determine the degradation status of multiple components of the electronic trip unit; the key parameters of the hierarchical electronic trip unit status assessment model are optimized to determine the status classification threshold; a dual-path electronic trip unit status determination model is constructed to determine the electronic trip unit status; this embodiment of the invention screens degradation characteristics through circuit simulation, classifies status levels, constructs a hierarchical electronic trip unit status assessment model based on dendritic neuron (DNM) and Dempster-Shafer evidence theory, and uses a biogeographical optimization algorithm for global optimization combined with a dual-path decision model to assess the status of low-voltage circuit breaker electronic trip units, as shown. Figure 2 As shown, the specific implementation steps are as follows: S1: Set the status of the circuit breaker's electronic trip unit, filter key performance parameters, and filter the degradation characteristics of the circuit breaker's electronic trip unit.

[0021] S11: Simulation Setup and Sensitivity Analysis; Using simulation software, build a simulation model of the power supply module for the electronic trip unit of a low-voltage circuit breaker, such as... Figure 3 As shown; sensitivity analysis was performed on candidate degraded components in the electronic trip unit power module to assess the impact of component parameter changes on the power module output performance; the component parameter sensitivity was determined to be: ; in, Sensitivity of component parameters; The output value after the input variable has changed; The output value before the input variable is changed; These are the initial input parameter values; represents the change value of the input parameter; i is the component number.

[0022] S12: Screening Key Degradation Features of Electronic Trip Units in Circuit Breakers. The feature selection method in this embodiment employs a multi-factor comprehensive evaluation strategy, using simulation software for circuit simulation. During the candidate feature screening process, the sensitivity analysis of degradation features to the electronic trip unit's status, the failure mechanism of power module components, and engineering reliability indicators are simultaneously considered. Five key characteristic parameters reflecting power module degradation are selected: the capacitance change rate of electrolytic capacitor C5, the capacitance change rate of electrolytic capacitor C9, the on-resistance change rate of a metal-oxide-semiconductor transistor (MOSFET), the output voltage amplitude change rate, and the switching cycle change rate. MOSFET stands for Metal Oxide Semiconductor Field Effect Transistor. Using a multi-factor evaluation strategy for feature selection, the sensitivity analysis results of component parameter changes to the power module's output performance, the failure mechanism of power module components, and engineering reliability indicators are considered. Features with limited contribution to the electronic trip unit's status determination or insufficient engineering feasibility are eliminated, forming a multi-dimensional degradation feature set for low-voltage circuit breaker electronic trip unit status assessment. The key degradation features identified include: the rate of change of capacitance value of electrolytic capacitor C5, the rate of change of capacitance value of electrolytic capacitor C9, the rate of change of MOSFET on-resistance, the rate of change of power supply output voltage amplitude, and the rate of change of switching cycle. Specifically, the rates of change of capacitance values ​​of electrolytic capacitors C5 and C9 are used to characterize the attenuation of the power module's energy storage and filtering capabilities due to aging; the rate of change of MOSFET on-resistance is used to characterize the increased conduction losses and thermal degradation of power switching devices; the rate of change of power supply output voltage amplitude is used to characterize the decreased power supply stability of the electronic trip unit; and the rate of change of switching cycle is used to characterize the timing drift of the switching control. These five physical quantities are normalized to form a five-dimensional degradation feature vector. This serves as the input for subsequent Gaussian state matching, evidence construction, and state determination.

[0023] S13: Classify and identify the status levels of the electronic trip unit; the working status of the electronic trip unit of the low-voltage circuit breaker is divided into 4 levels, and the description of each level is shown in Table 1. Level 1 corresponds to the normal operation status, Level 2 corresponds to the detection status, Level 3 corresponds to the abnormal status, and Level 4 corresponds to the fault status. The electronic trip unit of each status is described.

[0024] Table 1. Status Description of Electronic Trip Unit

[0025] The identification framework is a complete finite set containing all hypothetical state terms or decision outcomes, serving as the basis for subsequent Dempster-Shafer evidence theory modeling (DS evidence theory). The complete finite set of the electronic trip unit state identification framework is defined as follows: ; in, A complete finite set of frameworks for electronic trip unit status identification; The electronic trip unit is in normal condition. Status of the electronic trip unit; The electronic trip unit is in an abnormal state. The electronic trip unit is in a fault state.

[0026] To illustrate the effectiveness of the method of this invention, this embodiment generates 800 sets of samples with different degrees of degradation through simulation, of which 700 sets are used as training samples. Some of the original sample data are shown below. Figure 5 As shown, the parameters of the electronic trip unit status assessment model are optimized and the status division threshold of the electronic trip unit is determined; 100 groups are used as test samples for performance verification of the electronic trip unit status assessment.

[0027] The following describes an engineering application of this invention: The invention will be further described below with reference to the accompanying drawings and embodiments, but the scope of protection of this invention is not limited to the embodiments. This embodiment is built in simulation software. Figure 3 The electronic trip unit power module shown generates 800 sets of five-dimensional normalized degradation feature samples. Using random arrangement, 700 sets are selected as the training set, and the remaining 100 sets are used as the independent test set. Simulation is used to optimize the 47-dimensional parameters. Among the saved optimal parameters, the confidence coefficients of the five features are... Cell body layer parameters are , Optimal Gaussian mean Table 2 lists the average detection state values ​​for the five features; the optimal Gaussian standard deviation is also shown. Table 3 lists the detection state standard deviations for the five features; the state splitting thresholds are obtained from the training set. , , .

[0028] Table 2 Optimal Gaussian Mean Statistical table

[0029] Table 3 Optimal Gaussian Standard Deviation Statistical table

[0030] All five inputs are normalized data. Its actual label is the detection status. The values ​​represent the degradation of the capacitance values ​​of C5 and C9, the MOSFET on-resistance, the output voltage amplitude, and the switching cycle relative to their respective engineering references and normalized ranges, and are therefore dimensionless.

[0031] S2: Construct a hierarchical electronic trip unit (EPU) state assessment model based on dendritic neuron (DNM) and Dempster-Shafer evidence theory to determine the degradation state of multiple components in the EPU. The hierarchical EPU state assessment model, abbreviated as DSDNM (Dempster-Shafer Deep Neuron Model), specifically includes four layers: the synaptic layer corresponding to the construction of EPU degradation feature evidence, the dendritic layer corresponding to the fusion of multi-source degradation features, the membrane layer corresponding to the EPU state probability transformation and state measurement calculation, and the cell body layer corresponding to the continuous state assessment numerical output. Figure 4 As shown, specifically, the synaptic layer uses the rate of change of capacitance of electrolytic capacitor C5, the rate of change of capacitance of electrolytic capacitor C9, the rate of change of resistance of MOSFET on-resistance, the rate of change of power supply output voltage amplitude, and the rate of change of switching cycle as input degradation features. Based on Dempster-Shafer evidence theory, a basic probability assignment function is constructed for each degradation feature corresponding to different electronic trip unit states. The dendritic layer fuses the single-feature evidence formed by each degradation feature using Dempster combination rules to obtain comprehensive evidence of the degradation state of multiple electronic trip unit components. The membrane layer performs confidence-based pigfinistic probability transformation on the fused comprehensive evidence and calculates the membrane potential in conjunction with the electronic trip unit state scale. The cell body layer generates continuous state assessment values ​​based on the membrane potential and uses them for subsequent electronic trip unit state level determination. Thus, while preserving the dendritic neuron (DNM) hierarchical structure, each layer corresponds to the electronic trip unit degradation feature input, evidence fusion, state measurement calculation, and state assessment output, respectively, solving the uncertainty problem caused by the degradation of multiple components in low-voltage circuit breaker electronic trip units.

[0032] S21: The synaptic layer uses Dempster-Shafer evidence theory to construct the basic probability assignment function and correct its credibility; it also uses Dempster-Shafer evidence theory to represent uncertain information. For the five input degradation features, an improved Gaussian membership function is used to construct the basic probability assignment function corresponding to each electronic trip unit degradation feature. Furthermore, an electronic trip unit state assignment evidence term U is introduced. This term is determined by the closeness of the matching degree distribution of the same degradation feature with respect to the degradation feature states of each electronic trip unit, and is expressed through the corresponding basic probability assignment function value. This characterizes the degree of overlap of the state boundaries of the degradation feature under the current monitoring sample; simultaneously, a confidence correction coefficient is introduced to modify the basic probability allocation function to characterize the random uncertainty caused by fluctuations in monitoring data, forming a single-feature basic probability allocation function corresponding to each degradation feature. The formula for determining the Gaussian membership function corresponding to the electronic trip unit state is as follows: ; in, Let be the membership degree of the i-th degradation feature to the j-th type of electronic trip state; The i-th normalized degradation characteristic value corresponds to the rate of change of capacitance value of electrolytic capacitor C5, the rate of change of capacitance value of electrolytic capacitor C9, the rate of change of resistance value of MOSFET on-resistance, the rate of change of power supply output voltage amplitude and the rate of change of switching cycle, respectively. The mean of the Gaussian membership function corresponding to the j-th type of electronic trip state for the i-th degradation feature is the typical degradation center. Let be the standard deviation of the Gaussian membership function corresponding to the j-th type of electronic trip state for the i-th degradation feature, that is, the allowable degradation fluctuation range under this state; This is a temperature coefficient used to adjust the sensitivity at state boundaries; in this embodiment, it is taken as... =2; i∈{1,2,3,4,5} corresponds to 5 features, j∈{1,2,3,4} corresponds to 4 electronic trip states; e is a natural constant.

[0033] The matching degree of the degradation feature of the i-th degradation feature to the degradation feature state of each electronic trip unit is calculated based on the membership degree corresponding to the state of each electronic trip unit. Specifically: ; in, Let be the degree of matching between the i-th degradation feature and the j-th type of electronic trip state; denominator calibration parameter; j is the electronic trip unit status number.

[0034] Substituting the normalized data X and the parameters from Tables 2 and 3 into the calculation, the Gaussian state matching degree matrix is ​​obtained as follows: ; Each row corresponds to one of the five physical degradation characteristics, and each column corresponds to one of the following characteristics. to The matching degree of the degenerate feature state is calculated row by row to obtain: ; Membership of electronic trip unit status Matching degree with degenerate feature state The input is limited to five degradation features of the electronic trip unit, and the output is not directly used as the classification result, but is used to construct a basic probability allocation function containing the evidence to be assigned. To characterize the random uncertainty caused by fluctuations in monitoring data, a state-assigned evidence term U is introduced. When the matching degree distribution of a certain degradation feature to different electronic trip unit states is relatively close, it indicates that the discrimination uncertainty of this feature is high, and it is difficult to stably distinguish between normal, detection, abnormal, or fault states. In this case, this part of the information is not forcibly assigned to a specific state, but is assigned to the basic probability allocation function value corresponding to the state-assigned evidence term U. .therefore, This is used to handle uncertain evidence arising from ambiguous states, monitoring noise, and ambiguity of degraded features. It is based on the state matching degree of the degraded features and the basic probability assignment function value corresponding to the state-to-be-assigned evidence item U. Calculate the degradation characteristic of the i-th term to clarify the state of the electronic trip unit. Basic probability assignment function value Specifically: ; in, Electronic trip unit status The value of the basic probability assignment function; The basic probability assignment function value is assigned to the i-th degenerate feature as the state-to-be-assigned evidence item U; U is the state-to-be-assigned evidence item. The electronic trip unit is in operation.

[0035] Therefore: ; Furthermore, different degradation characteristics are affected by noise, operating condition disturbances, and have varying degrees of vulnerability; therefore, the first... Item feature parameter confidence correction coefficient Its value range is: 0 < <1. Used to characterize the credibility of the i-th degenerate feature as state evidence; parts with insufficient credibility are transferred to the state to be assigned evidence item U.

[0036] The confidence level of the basic probability assignment function is adjusted, and the adjusted basic probability assignment function for definite states is as follows: ; in, The basic probability assignment function after credibility correction; Electronic trip unit status The credibility of the basic probability assignment function value is corrected. For the first The confidence correction coefficient for the characteristic parameter.

[0037] The corrected basic probability assignment function value for the state-to-be-assigned evidence item U is: ; in, The basic probability assignment function value for the evidence item U to be assigned to the state; Because of the first The basic probability assignment function parameters of the feature parameters are transferred to the state to be assigned evidence item U due to insufficient confidence of the feature parameters.

[0038] To ensure that the modified basic probability allocation function satisfies the total mass conservation constraint, it is normalized as follows: ; in, Let be the basic probability assignment function for the single feature corresponding to the i-th degenerate feature; For the i-th degradation feature, the state term The basic probability assignment value of a single feature; State item to be evaluated ; For electronic trip unit status items The credibility is corrected for the value of the basic probability assignment function.

[0039] Obtain the basic probability assignment function for each of the five degenerate features. That is, the characteristic evidence corresponding to the degradation feature.

[0040] S22: The dendritic layer fuses multi-source feature evidence according to the Dempster combination rule. The dendritic layer uses the Dempster combination rule to fuse evidence from the basic probability assignment functions of each degenerate feature output from the synaptic layer in S21, using the conflict coefficient K to quantify the degree of inconsistency between different feature sources, thus mitigating feature evidence conflicts and the resulting cognitive uncertainty. The Dempster combination rule combines features that jointly support the same state. The evidence is strengthened, while the uncertain parts that cannot be clearly classified are retained through the state-to-be-assigned evidence item U. After fusing the five degenerate physical quantities in sequence, the fused basic probability assignment function is obtained. , used to represent the comprehensive evidence support of the electronic trip unit in each state under the combined effect of multiple physical degradation features. Let two functions be selected from the five single-feature basic probability allocation functions obtained from S21, denoted as . and ,in, Since the state-to-be-assigned evidence item U is introduced in S21, the influence of the electronic trip unit state and the basic probability assignment function corresponding to the state-to-be-assigned evidence item U must be clearly defined during feature fusion; the conflict coefficient K formula is: ; Where K is the conflict coefficient; This is the p-th state in the electronic trip unit's state; This is the qth state in the electronic trip unit's state; For single-feature basic probability assignment function Assigned to state item The value of the basic probability assignment function; For single-feature basic probability assignment function Assigned to state item The value of the basic probability assignment function; It is the first degenerate feature of the single-feature basic probability assignment function; is the second degenerate feature of the single-feature basic probability assignment function; p is the first index of the electronic trip unit state. ; q is the second index of the electronic trip unit status. .

[0041] After fusing the two sets of characteristic parameters, the status of the electronic trip unit is determined. The corresponding basic probability assignment function value is: ; in, For single-feature basic probability assignment function and Assigned to state items after being combined according to Dempster's combination rules The value of the basic probability assignment function; For single-feature basic probability assignment function Assigned to state item The value of the basic probability assignment function; For single-feature basic probability assignment function Assigned to state item The value of the basic probability assignment function; For single-feature basic probability assignment function The basic probability assignment function value assigned to the state-to-be-assigned evidence item U; For single-feature basic probability assignment function The basic probability assignment function value assigned to the state-to-be-assigned evidence item U; This is the Dempster combinatorial operator.

[0042] Evidence items to be assigned in the merged state The basic probability assignment function value is: ; in, For single-feature basic probability assignment function and Evidence items to be assigned to the state after being combined according to Dempster's combination rules The value of the basic probability assignment function.

[0043] By sequentially applying Dempster's combination to the individual feature basic probability assignment functions corresponding to the five degenerate features, a comprehensive basic probability assignment function is obtained, as follows: ; To fuse the basic probability assignment function; To integrate the basic probability assignment function for the state term The basic probability allocation value; Let i be the basic probability assignment function for the single feature corresponding to the i-th degenerate feature. ; The basic probability assignment function corresponding to the first degenerate feature is applied to the state term. The basic probability allocation value; The basic probability assignment function corresponding to the second degenerate feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the third degenerate feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the fourth degradation feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the fifth degradation feature for the state term The basic probability allocation value; This refers to the state item to be evaluated.

[0044] The state of the evidence items to be assigned is calculated according to the formula. The corresponding basic probability assignment function value get: ; Evidence items to be assigned Substituting the basic probability allocation function values ​​and confidence coefficients, we obtained the basic probability allocation function values ​​of the five features corresponding to the five degradation features, as shown in Table 4. We also statistically analyzed the basic probability allocation function values ​​of the five states.

[0045] Table 4. Statistical Table of Basic Probability Assignment Function Values ​​for Each State

[0046] The sum of each row is 1, indicating that rows with insufficient confidence have been transferred to U without violating mass conservation. Four Dempster combinations are performed in the order of features 1 to 5. The conflict coefficients for the four combinations are... The average conflict coefficient is The Dempster combination of the basic probability assignment function values ​​for each of the five degenerate features is performed sequentially to obtain the comprehensive basic probability assignment function value, as follows: ; The output of this step is the feature fusion basic probability assignment function. It integrates multi-source degradation features and alleviates cognitive uncertainty caused by conflicting evidence of multi-source degradation features.

[0047] S23: Determine the confidence level Pignistic probability transformation and membrane potential of the membrane layer. The comprehensive basic probability assignment function of the membrane layer to the dendritic layer output is subjected to a confidence level Pignistic probability transformation. This transformation is used to assign the state to the evidence items to be assigned. The basic probability allocation function values ​​are redistributed to each specific electronic trip unit (EPU) state, yielding the confidence probability value corresponding to each EPU state. The membrane potential (M) is then calculated by weighted summation using a preset EPU state scale. The membrane potential M is not a single-state probability, but a continuous state quantity comprehensively reflecting the current operating state of the EPU. The confidence probability of each EPU state is given by the formula: ; in, State of the j-th type electronic trip unit The reliability of the Pignistic probability; To integrate the basic probability assignment function for the state term The basic probability assignment values, j∈{1,2,3,4}; To integrate the basic probability assignment function with the evidence items to be assigned to the state The basic probability allocation value.

[0048] To continuously and quantitatively characterize the state of the electronic trip unit, four state scales corresponding to the four electronic trip unit states are preset. Based on the confidence probability of each electronic trip unit's state and the corresponding state scale, the membrane output potential is determined as follows: ; Where M is the output membrane potential; This is the state scale corresponding to the state of the j-th type of electronic trip unit.

[0049] In the example, the confidence level Pignistic probability was calculated to obtain: ; The output membrane potential is: .

[0050] S24: Generate continuous state evaluation parameters for the electronic trip unit; the cell somatic layer uses the sigmoid activation function to evaluate the membrane potential. Perform a nonlinear mapping to generate continuous state evaluation parameters with values ​​ranging from [0,1]. Characterizes the state of the equipment. Continuous state assessment parameters correspond to the electronic trip unit state classification thresholds determined by subsequent optimization, which are then mapped to discrete electronic trip unit state levels. The activation function Sigmoid is as follows: ; in, Parameters for continuous state evaluation of electronic trip units; The slope parameter of the activation function Sigmoid; The activation threshold parameter has a value range of 1 ≤ ≤30, 0≤ ≤1; and The result is determined by global optimization in subsequent step S3.

[0051] The continuous state assessment parameters are characterization quantities obtained by performing state matching, evidence allocation, confidence correction, multi-source fusion, and state measurement on five degraded physical quantities. The smaller the value, the higher the degree of degradation of the electronic trip unit's power module; the larger the value, the closer the electronic trip unit is to normal operation. The continuous state assessment parameters will be used in step 3, in conjunction with the optimized electronic trip unit state classification threshold, for discrete electronic trip unit state assessment.

[0052] In this embodiment, the membrane potential is substituted into the Sigmoid activation function formula to obtain the continuous state evaluation parameters of the electronic trip unit: .

[0053] S3: Optimize the global parameters of the hierarchical electronic trip unit state assessment model and determine the state division threshold. To globally optimize the key parameters of the hierarchical electronic trip unit state assessment model, this embodiment employs the Biogeography-Based Optimization (BBO) algorithm for adaptive optimization.

[0054] S31: Construct the objective function of the biogeographical optimization algorithm (BBO); simulate the migration, mutation, and evolution of biological species between habitats, mapping the model's parameters to be optimized to habitat suitability index variables (SIV), and mapping the electronic trip unit status assessment effect to the habitat suitability index (HSI). Let the global parameter set to be optimized in the model be: ; in, The global parameter set to be optimized; Let be the mean of the Gaussian membership function corresponding to the j-th type of electronic trip state for the i-th feature; Let be the standard deviation of the Gaussian membership function corresponding to the state of the j-th type of electronic trip unit for the i-th feature; For the first The confidence correction coefficient for the characteristic parameter is used to adjust the reliability of different physical quantities as evidence of state. The slope parameter of the activation function Sigmoid; The activation threshold parameter controls the mapping relationship between membrane potential and continuous state evaluation parameters. By globally optimizing these parameters, the subjectivity introduced by manually setting state boundaries, confidence weights, and output mapping parameters can be reduced.

[0055] To balance the accuracy of electronic trip unit (EPU) status determination with the stability of multi-source parameter fusion, the key parameters of the global adaptive optimization model are maximizing the accuracy of EPU status assessment and minimizing the average conflict coefficient as the objective function. Accuracy is used to constrain whether the final status label is correctly determined, and the average conflict coefficient is used to constrain whether the evidence formed by the five degenerate physical quantities contradicts each other. The optimization process selects a parameter set that combines both determination accuracy and evidence consistency. The objective function is established as follows: ; ; in, To minimize the objective function; λ is the average parameter conflict coefficient; λ is the conflict penalty weight, which is adjusted according to the actual dataset, and is taken as 0.01 in this embodiment of the invention. To improve the accuracy of electronic trip unit status assessment; This is an indicator function that takes the value 1 when the condition in parentheses is true, and 0 otherwise; N is the total number of training samples. The actual electronic trip unit status label for the nth training sample; The predicted state label for the electronic trip unit is obtained by the nth training sample according to the state mapping relationship of the hierarchical electronic trip unit state assessment model DSDNM in S4.

[0056] S32: Initialize and set up the biogeographical optimization algorithm; update parameters for non-elite habitats based on migration rate, perform habitat migration and mutation, and redetermine the objective function for electronic trip unit state assessment; backfill the optimal parameter solution set into the hierarchical electronic trip unit state assessment model to obtain the continuous state assessment parameter set corresponding to the electronic trip unit. .

[0057] S321: Initialize the biogeographical optimization algorithm (BBO algorithm); initialize the habitat population, randomly generate Q habitat individuals, where Q is the habitat population size, and each habitat corresponds to a set of parameters to be optimized for the electronic trip unit status assessment model. In this embodiment, to balance global search performance and computational complexity, the population size is set to 50 to 200. Migration parameters for the Biogeographical Optimization (BBO) algorithm are set, with both the maximum immigration rate (I) and maximum emigration rate (E) set to 1. The immigration rate monotonically decreases with increasing Habitat Suitability Index (HSI), while the emigration rate monotonically increases with increasing HSI. Habitat mutation parameters are also set, with the maximum mutation rate in this embodiment being... Set the value to 0.005. To preserve the current optimal solution, an elite retention strategy is adopted, retaining two individuals with the best habitat in each generation to directly enter the next generation. The algorithm termination condition is set to a maximum of at least 100 iterations.

[0058] S322: Habitat Migration and Abrupt Changes; Parameter updates for non-elite habitats are performed based on migration rates. Habitats with a high habitat suitability index (HSI) migrate to habitats with a low habitat suitability index (HSI) to obtain the superior suitability index variable (SIV). This is achieved by replacing the corresponding parameter components of the target habitat with the parameter components to be optimized from the source habitat, thus updating the habitat parameters. To enhance the search performance of the parameter search space, the habitat parameters are subjected to random abrupt abrupt changes; the abrupt change rate formula for the s-th habitat is: ; in, Let be the mutation rate of the s-th habitat; The maximum mutation rate; Let be the selection probability of the s-th habitat based on suitability; This represents the maximum selection probability in the population.

[0059] After randomly perturbing the corresponding habitat parameters based on the mutation rate, the objective function for assessing the state of the electronic trip unit is redefined, and the combination of habitat parameters is updated.

[0060] S323: Outputs the globally optimal parameters; after the Biogeographical Optimization (BBO) algorithm meets the termination condition, it outputs the optimal solution set of the parameter set to be optimized in the electronic trip unit state assessment model. The optimal parameter solution set is then backfilled into the hierarchical electronic trip unit state evaluation model. Calculations are performed on the training samples to obtain the continuous state evaluation parameter set corresponding to the electronic trip unit: ; in, This is a set of continuous state evaluation parameters for the electronic trip unit, used to determine the electronic trip unit state division threshold in S33; represents the continuous state evaluation parameter corresponding to the nth training sample; N is the total number of training samples.

[0061] S33: Determine the electronic trip unit state classification threshold based on the minimum misclassification principle; to determine the electronic trip unit state classification threshold corresponding to the continuous state evaluation parameters, the electronic trip unit state classification threshold is determined based on the continuous state evaluation parameters obtained from the training samples. For any two adjacent state classes and... ,in The continuous state evaluation parameter sample sets corresponding to the electronic trip unit states in the training set are extracted as follows: ; ; in, For the continuous state evaluation parameters corresponding to the nth sample; This is the status tag of the nth sample's actual electronic trip unit; The state is that of the m-th type electronic trip unit; The state is that of the (m+1)th type of electronic trip unit; Labels for real state The set of continuous state evaluation parameters corresponding to the training samples; Labels for real state The set of continuous state evaluation parameters corresponding to the training samples.

[0062] The two types of state evaluation parameter samples are merged and sorted according to their numerical values ​​from smallest to largest, resulting in an ordered set of evaluation parameters: ; in, To be and The ordered sequence of evaluation parameters obtained after merging and sorting; Let L be a sorting operator that arranges the evaluation parameters of the continuous states in ascending order; L is an ordered set. The total number of samples included; For ordered sets The first in Each sorting state evaluation parameter value, .

[0063] Based on the midpoints between adjacent evaluation parameter samples after sorting, a candidate boundary threshold set is constructed as follows: ; in, For state and The set of candidate boundary thresholds between; For ordered sets The value of the l-th sorting state evaluation parameter; For ordered sets The Middle +1 sorting status evaluation parameter value.

[0064] Since a smaller continuous state evaluation parameter indicates a higher degree of degradation of the electronic trip unit, the number of misclassified samples for a candidate threshold t is: ; in, The number of misclassified samples corresponding to the candidate boundary threshold t; For the continuous state evaluation parameters corresponding to the nth sample; This is an indicator function that takes the value 1 if the condition within the parentheses is true, and 0 otherwise. The set of continuous state evaluation parameters; Degenerate state The set of continuous state evaluation parameters.

[0065] The candidate boundary threshold that minimizes the number of misclassified samples is selected as the optimal boundary threshold between adjacent electronic trip states: ; in, For state With state The optimal boundary threshold between them; To minimize the number of misclassified samples; Candidate threshold selection operator that obtains the minimum value.

[0066] By using the above method, the optimal boundary threshold between three groups of adjacent electronic trip unit states is determined, resulting in a set of electronic trip unit state division thresholds. ;in, This is the threshold value that separates the normal state from the detection state of the electronic trip unit. To define the boundary threshold between the detected state and the abnormal state, This is the boundary threshold between abnormal and fault states, and it satisfies: .

[0067] The final output is the globally optimized parameter set. and the set of thresholds for classifying the state of the electronic trip unit. Based on a defined set of thresholds, continuous state evaluation parameters can be mapped to corresponding discrete electronic trip unit states, enabling adaptive hierarchical discrimination of electronic trip unit states.

[0068] S4: Construct a dual-path electronic trip unit status determination model based on a continuous state parameter evaluation path and a rapid fault determination path to determine the electronic trip unit status.

[0069] S41: Determine the electronic trip unit state path based on continuous state parameter evaluation; input the sample to be evaluated into the optimized hierarchical electronic trip unit state evaluation model DSDNM to obtain the corresponding continuous state evaluation parameter Score; based on the electronic trip unit state division threshold set determined in step S33... The continuous state assessment parameters are mapped to the corresponding electronic trip unit state levels, specifically as follows: ; in, The result of the electronic trip unit status determination; The electronic trip unit is in operation. In detection status; This is an abnormal state; The condition is faulty; This is the first dividing threshold; This is the second boundary threshold; This is the third dividing threshold.

[0070] For the nth training sample, its continuous state evaluation parameters are... Substituting the above state mapping relationship, we obtain the electronic trip state determination result of the hierarchical electronic trip state evaluation model DSDNM corresponding to this training sample. To maintain consistency with the predicted state label symbol in the objective function of step S31, the predicted electronic trip unit state label is set as follows: ; in, The predicted electronic trip unit status label for the nth training sample; The state level is determined by the hierarchical electronic trip unit state assessment model DSDNM continuous state parameter evaluation path for the nth training sample, and .

[0071] S42: A rapid fault determination path is obtained based on engineering rule constraints; this path is used for engineering interception of extreme degradation or obvious failure states. Let the degradation feature vector of the sample to be evaluated be: ,in Corresponding to Capacity change rate Rates of change in capacitance, MOSFET on-resistance, power supply output voltage amplitude, and switching cycle are considered. Limit thresholds are set for these key physical quantities. When any Exceed When this occurs, it indicates that the corresponding component has degraded or the power supply output performance has reached the engineering failure boundary, i.e., it satisfies: The electronic trip unit is directly determined to be in a faulty state. The final dual-path decision rule is as follows: if an out-of-bounds fault is quickly determined to trigger the path, the fault status is output; if not triggered, the status level of the path output is evaluated using continuous state assessment. The extreme boundary thresholds are determined based on the component's set operating boundary, the power module's functional failure boundary, and the equipment's engineering operation specifications. In this embodiment, the normalized extreme boundary thresholds are set as follows: ; S43: The dual-path decision determines the final electronic trip unit status as follows: ; in, This is the final electronic trip unit status determination result after dual-path fusion; This is the normalized limit boundary threshold.

[0072] The boundary threshold between the detected state and the abnormal state in the embodiment for: ; Continuous state parameter evaluation path output detection state Meanwhile, the maximum value of the five features is 0.6600, none of which exceeds the limit boundary of 1; therefore, paths exceeding the boundary do not output fault status. The final result after dual-path fusion is... .

[0073] For any sample with a normalized degradation trait greater than 1, the program directly overrides the continuous state parameter evaluation path and outputs the fault state. For example, when the normalized value of the C9 capacitance change rate is 1.5679, even if the continuous state parameter evaluation path does not output a fault state, the final result is still [fault state]. .

[0074] The change in the accuracy of electronic trip unit status assessment with the number of iterations during model training is as follows: Figure 6 As shown, with the increase of the number of iterations of the BBO algorithm, the accuracy of the electronic trip unit status assessment gradually improves and tends to stabilize. After reaching the preset termination condition, the optimal parameter combination is obtained, and the corresponding electronic trip unit status assessment accuracy reaches 97.5%, indicating that the method of the present invention can effectively assess the status of the electronic trip unit of low-voltage circuit breakers.

[0075] To verify the contribution of each technical module of this invention to the performance of electronic trip unit status assessment, a comparative experiment with different model structures was further conducted. The results are shown in Table 5. The accuracy of electronic trip unit status assessment was analyzed using four methods. After introducing the DS evidence theory parameter fusion model on the basis of the traditional dendritic neuron DNM model, the accuracy of electronic trip unit status assessment was improved. After further combining BBO global parameter optimization and dual-path electronic trip unit status determination model, the model accuracy was improved again, indicating that the technical modules proposed in this invention can synergistically improve the performance of electronic trip unit status assessment.

[0076] Table 5. Statistical table of accuracy of electronic trip unit status assessment using different methods

[0077] Error distribution of electronic trip unit condition assessment results as follows Figure 7 As shown, the evaluation error of most test samples is concentrated in the lower range, indicating that the method of the present invention can accurately distinguish the electronic trip condition corresponding to different degrees of degradation.

[0078] The beneficial effects of the embodiments of this invention are as follows: This invention proposes a data- and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers. By extracting key degradation features through simulation analysis, it overcomes the problems of traditional electronic trip unit state assessment methods relying on a single monitoring parameter and having limited state representation, thus improving the effectiveness of low-voltage circuit breaker electronic trip unit state assessment. Addressing the random uncertainty caused by fluctuations in monitoring data and the cognitive uncertainty caused by conflicting evidence of multi-source degradation features, a hierarchical electronic trip unit state assessment model is constructed, effectively mitigating the interference of uncertainty issues. The accuracy of electronic trip unit state assessment is improved through global optimization and a dual-path decision model, enhancing the model's adaptability to actual degradation modes. The use of a dual-path decision model reduces the risk of misjudging the electronic trip unit's state.

[0079] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A data- and knowledge-driven method for determining the multi-parameter status of an electronic trip unit in a circuit breaker, characterized in that: It includes: S1: Set the circuit breaker electronic trip unit status, perform sensitivity analysis to determine the sensitivity of component parameters. The degradation characteristics of the circuit breaker's electronic trip unit are screened as follows: the rate of change of capacitance of electrolytic capacitor C5, the rate of change of capacitance of electrolytic capacitor C9, the rate of change of on-resistance of the metal-oxide-semiconductor transistor, the rate of change of power supply output voltage amplitude, and the rate of change of switching cycle. The status levels of the electronic trip unit are then classified and identified, and a complete finite set of the electronic trip unit status identification framework is established. ; S2: Construct a hierarchical electronic trip unit (EPU) state assessment model based on dendritic neurons and Dempster-Shafer evidence theory; construct the EPU state... The basic probability assignment function is obtained by refining the confidence level and obtaining the basic probability assignment function for a single feature. ; by performing Dempster combinations sequentially, the fusion basic probability assignment function is obtained. Perform state confidence probability transformation and state measurement of the electronic trip unit to generate continuous state assessment parameters for the electronic trip unit. ; S3: Assume the global parameter set to be optimized in the model. The objective function for electronic trip unit (EPU) status assessment was determined using a biogeographical optimization algorithm. The hierarchical EPU status assessment model obtained in S2 was optimized to obtain the set of continuous status assessment parameters corresponding to the EPU. Determine the electronic trip unit state classification threshold based on the principle of minimum misclassification; determine the optimal boundary threshold between three groups of adjacent electronic trip unit states. The set of thresholds for classifying the state of the electronic trip unit is obtained. ; S4: Map the continuous state evaluation parameters in S3 to the corresponding electronic trip unit state levels, and output the electronic trip unit state determination results corresponding to the training samples. Based on engineering rule constraints, a rapid fault determination path is obtained to determine if the current equipment is in a fault state; dual-path decision determines the status of the electronic trip unit.

2. The data- and knowledge-driven multi-parameter state determination method for circuit breaker electronic trip units according to claim 1, characterized in that: Step S2 is as follows: S21: Determine the Gaussian membership function corresponding to the state of the electronic trip unit, obtain the matching degree of the degradation feature state of each electronic trip unit state to the i-th degradation feature, and use evidence theory to construct the basic probability assignment function and correct the credibility. S22: Integrate the basic probability allocation function of multi-source features; determine the conflict coefficient K and clarify the state of the electronic trip unit. The corresponding basic probability allocation value; The fusion yields the state-to-be-assigned evidence items. The basic probability assignment values ​​are obtained by sequentially combining the basic probability assignment functions of the individual features corresponding to the five degenerate features to obtain the fused basic probability assignment function. ; S23: Perform confidence-transformation probability transformation on the fusion basic probability allocation function. Based on the confidence-transformation probability of the electronic trip unit's state and the state scale, determine the membrane output potential M and generate continuous state evaluation parameters for the electronic trip unit. .

3. The data- and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers according to claim 2, characterized in that: The method for obtaining the degradation characteristic state matching degree of the electronic trip unit state in step S21 is as follows: ； ； in, Let be the degree of matching between the i-th degradation feature and the j-th type of electronic trip state; Here are the denominator calibration parameters; j is the electronic trip unit status number; Let be the membership degree of the i-th degradation feature to the j-th type of electronic trip state; The normalized monitoring feature value of the i-th term; The mean of the Gaussian membership function corresponding to the state of the electronic trip unit of the i-th degradation feature is denoted as ; The standard deviation of the Gaussian membership function corresponding to the state of the j-th type of electronic trip unit for the i-th degradation feature; is the temperature coefficient; e is the natural constant.

4. The data and knowledge-driven multi-parameter state determination method for circuit breaker electronic trip units according to claim 2, characterized in that: The method for obtaining the comprehensive basic probability allocation function in step S22 is as follows: ； in, To fuse the basic probability assignment function; To integrate the basic probability assignment function for the state term The basic probability allocation value; The basic probability assignment function for each of the five degenerate features; The basic probability assignment function corresponding to the first degenerate feature is applied to the state term. The basic probability allocation value; The basic probability assignment function corresponding to the second degenerate feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the third degenerate feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the fourth degradation feature for the state term The basic probability allocation value; The basic probability assignment function corresponding to the fifth degradation feature for the state term The basic probability allocation value; This refers to the state item to be evaluated.

5. The data and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers according to claim 2, characterized in that: The method for obtaining the continuous state evaluation parameters of the electronic trip unit in step S23 is as follows: ； ， ； in, Parameters for continuous state evaluation of electronic trip units; State of the j-th type electronic trip unit The probability of confidence conversion; To integrate the basic probability assignment function for the state term The basic probability allocation value; To integrate the basic probability assignment function with the evidence items to be assigned to the state The basic probability distribution value; M is the output membrane potential of the membrane layer; This is the state scale corresponding to the state of the j-th type of electronic trip unit; The slope parameter of the activation function; This is the activation threshold parameter.

6. The data- and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers according to claim 1, characterized in that: Step S3 is as follows: S31: Assume the global parameter set to be optimized in the model. With the goal of maximizing the accuracy of electronic trip unit status assessment and minimizing the average conflict degree, an objective function for comprehensive optimization is established. S32: Initialize and set up the biogeographical optimization algorithm; update parameters for non-elite habitats based on migration rate, perform habitat migration and mutation, and redetermine the objective function for electronic trip unit state assessment; backfill the optimal parameter solution set into the hierarchical electronic trip unit state assessment model to obtain the continuous state assessment parameter set corresponding to the electronic trip unit. ; S33: Determine the electronic trip unit state division threshold based on the minimum misclassification principle; determine the optimal boundary threshold between three groups of adjacent electronic trip unit states. The set of thresholds for classifying the state of the electronic trip unit is obtained. .

7. The data and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers according to claim 6, characterized in that: The objective function for minimizing the average conflict degree in step S31 is as follows: ； ； in, To minimize the objective function; λ is the average parameter conflict coefficient; λ is the conflict penalty weight. To improve the accuracy of electronic trip unit status assessment; Here, N is the indicator function; N is the total number of training samples. The actual electronic trip unit status label for the nth training sample; Predict the status tag for the electronic trip unit; The global parameter set to be optimized.

8. The data and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers according to claim 6, characterized in that: The method for obtaining the optimal boundary threshold in step S33 is as follows: ； ； in, The optimal boundary threshold for the state; To minimize the number of misclassified samples; Candidate threshold selection operator that obtains the minimum value; The number of misclassified samples corresponding to the candidate boundary threshold t; For the continuous state evaluation parameters corresponding to the nth sample; Degenerate state The set of continuous state evaluation parameters; Degenerate state The set of continuous state evaluation parameters; m is the state category of the electronic trip unit.

9. The data- and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers according to claim 1, characterized in that: Step S4 is as follows: S41: Determine the electronic trip unit status assessment path based on the continuous status assessment parameters; use the electronic trip unit status assessment model to obtain the corresponding continuous status assessment parameters and map them to the corresponding electronic trip unit status level. Obtain the electronic trip unit status determination result corresponding to the training sample. ; S42: Obtain a rapid fault determination path based on engineering rule constraints; define the degradation feature vector of the sample to be evaluated. Set limit boundary thresholds for key degradation features. The device is determined to be in a faulty state. S43: Dual-path decision-making determines the final electronic trip unit status assessment result. .

10. The data- and knowledge-driven multi-parameter state determination method for electronic trip units of circuit breakers according to claim 9, characterized in that: In step S41, the mapping is to the corresponding electronic trip unit status level, specifically as follows: ； in, The result of the electronic trip unit status determination; The electronic trip unit is in operation. In detection status; This is an abnormal state; This is a fault condition; This is the first dividing threshold; This is the second boundary threshold; This is the third dividing threshold.

Images