Knowledge-driven steering engine fault rapid diagnosis method and system

Abstract

The invention discloses a steering engine fault rapid diagnosis method and system based on knowledge driving, and belongs to the technical field of fault identification, and the method comprises the steps: S1, building an electric steering engine fault case library; s2, classifying and storing the fault information of the electric steering engine based on a distributed architecture; s3, carrying out discretization processing and labeling processing on continuous fault indexes in the fault information of the electric steering engine; s4, based on the discrete data obtained in the step S3, using an Apriori association rule algorithm to mine association rules among the fault information of the electric steering engine, and generating a steering engine fault knowledge rule base; and S5, according to fault information input by a user, performing fault case reuse based on the electric steering engine fault case library, and performing fault knowledge reasoning based on the steering engine fault knowledge rule library. According to the method, systematic accumulation and intelligent reuse of steering engine fault knowledge are realized, the efficiency and accuracy of steering engine fault diagnosis are remarkably improved, the dependence on artificial experience is reduced, and the method is suitable for fault prevention and rapid maintenance of various aircraft steering engine systems.

Description

Technical Field

[0001] This invention relates to the field of fault identification technology, and in particular to a knowledge-driven method and system for rapid diagnosis of servo motor faults. Background Technology

[0002] Electromechanical servo systems, as a crucial component of aircraft, are a key link in achieving high-quality aircraft design and manufacturing. Among them, electric servo motors, as the core actuators of the electromechanical servo system, directly affect the aircraft's handling performance and flight safety. Currently, electric servo motors experience frequent failures during development, testing, and use. Common faults include motor stall, abnormal feedback data, and large angle fluctuations, posing serious challenges to product quality control and maintenance. Preventing the recurrence of quality problems with electric servo motors, and quickly identifying the causes of failures when they occur, and proposing highly reliable solutions, are paramount in the design of high-quality, high-efficiency electromechanical servo systems.

[0003] Existing fault diagnosis methods for electric servos mainly rely on the experience of relevant technicians, lacking systematic scientific knowledge support and data-driven decision-making, and have the following shortcomings: 1. Long diagnostic cycle: During the diagnosis of faults, manual troubleshooting is required. It often takes several hours or even days to go from the fault symptoms to the location of the cause. This consumes a lot of manpower, material resources and time, which seriously affects the delivery and maintenance efficiency of aircraft products. 2. High reliance on experience: Fault knowledge is difficult to accumulate in a structured way, and staff turnover can easily lead to experience gaps, resulting in the recurrence of similar faults; 3. Lack of scientific basis: The diagnostic process is highly subjective, easily limited by personal experience, and it is difficult to form reusable diagnostic rules; 4. Insufficient data utilization: The multi-source data such as voltage, current, and temperature generated during the operation of the servo motor have not been systematically mined, which cannot support intelligent diagnosis; 5. Unsuitable for batch fault handling: As the number and types of electromechanical servo systems continue to increase, for example, with the mass deployment of aircraft, servo motor fault data surges, making it difficult for traditional manual methods to achieve rapid response and batch processing.

[0004] Therefore, in order to quickly locate quality problems in electromechanical servo products and take certain preventive measures, thereby improving the quality of electromechanical servo products, it is an important part of the current development of electromechanical servo products. There is an urgent need for a dedicated fault diagnosis solution for servo motors that can systematically accumulate fault knowledge, intelligently discover fault patterns, and quickly respond to fault diagnosis needs. Summary of the Invention

[0005] The purpose of this invention is to overcome the problems existing in the prior art and to provide a knowledge-driven method and system for rapid diagnosis of servo motor faults.

[0006] The objective of this invention is achieved through the following technical solution: Firstly, a knowledge-driven rapid servo motor fault diagnosis method is provided, including: S1. Establish a database of electric steering gear failure cases; S2. Classify and store electric servo motor fault information based on a distributed architecture; S3. Discretize and label the continuous fault indicators in the electric servo motor fault information. The continuous fault indicators include servo motor voltage, current and temperature. The discretization process combines the elbow method and the K-value comprehensive evaluation algorithm to determine the optimal number of clusters for each continuous fault indicator. S4. Based on the discrete data obtained in step S3, use the Apriori association rule algorithm to mine the association rules between electric servo motor fault information and generate a servo motor fault knowledge rule base. S5. Based on the fault information input by the user, reuse fault cases based on the electric servo motor fault case library, and perform fault knowledge reasoning based on the servo motor fault knowledge rule library.

[0007] In some embodiments, the method of combining the elbow method and the K-value comprehensive evaluation algorithm to determine the optimal number of clusters for each continuity failure index includes: Preliminary calculation of the number of clusters based on the elbow method; The K-value comprehensive evaluation algorithm is used to optimize the initially calculated number of clusters, specifically including: The optimal number of clusters is determined by comprehensively evaluating the sum of squared errors within clusters, silhouette coefficient, inter-cluster dispersion, and intra-cluster dispersion, and by weighted summation.

[0008] In some embodiments, the tagging process includes: Different labels are assigned to each discrete interval, and different fault phenomena within each discrete interval are labeled with different labels.

[0009] In some embodiments, the reuse of fault cases based on the electric servo motor fault case library includes: Calculate similarity using different methods for different types of data to obtain sub-similarity for each type of data. The combined similarity scores are obtained by integrating the results of each similarity calculation. Based on this comprehensive similarity, fault cases are screened, and the comprehensive similarity and similar fault cases are output.

[0010] In some embodiments, the formula for calculating the comprehensive similarity is: Among them, the and These are the weighting coefficients. This represents the sub-similarity of corresponding data. This represents the sub-similarity corresponding to text-based data.

[0011] In some embodiments, the reuse of fault cases based on the electric servo motor fault case library further includes: The weights of each sub-similarity are dynamically adjusted based on user feedback.

[0012] Secondly, a knowledge-driven rapid fault diagnosis system for servo motors is provided, including: The fault case library construction module is used to build a fault case library for electric servo motors; The data classification and storage module is used to classify and store electric steering gear fault information based on a distributed architecture. The data processing module is used to discretize and label the continuous fault indicators in the electric servo motor fault information. The continuous fault indicators include servo motor voltage, current and temperature. The discretization process combines the elbow method and the K-value comprehensive evaluation algorithm to determine the optimal number of clusters for each continuous fault indicator. The fault knowledge rule base construction module is used to mine the association rules between electric servo motor fault information based on the discrete data obtained from the data processing module using the Apriori association rule algorithm, and generate a servo motor fault knowledge rule base. The knowledge reuse and reasoning module is used to reuse fault cases based on the electric servo motor fault case library according to the fault information input by the user, and to perform fault knowledge reasoning based on the servo motor fault knowledge rule library.

[0013] It should be further noted that the technical features corresponding to the above-mentioned options and embodiments can be combined or substituted with each other to form new technical solutions without conflict.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs a fault case library for electric servo motors in electromechanical servo systems, and simultaneously uncovers strong correlations between fault information, ultimately enabling knowledge reuse, rapid fault diagnosis and precise fault location. This reduces the reliance on human personnel during the development of electromechanical servo systems, thereby achieving high-quality and high-efficiency design and promoting the development of the electromechanical servo system field. Specifically, it includes: 1. Systematically accumulate knowledge of servo motor malfunctions: By building a dedicated servo motor malfunction case library and knowledge rule library, scattered malfunction experience is transformed into structured knowledge, supporting continuous knowledge accumulation and iterative optimization.

[0015] 2. Improve the efficiency of servo motor fault diagnosis: By combining case reuse and rule reasoning, the fault can be quickly matched and located, reducing the diagnosis time from several hours to minutes, and significantly improving the maintenance response speed.

[0016] 3. Improve diagnostic accuracy and scientific rigor: Based on data-driven association rule mining, provide objective and verifiable diagnostic evidence, reduce human error, and improve the credibility of diagnostic results.

[0017] 4. Reduce reliance on expert experience: By building a knowledge base and rule base, expert experience is solidified in the system, reducing the impact of personnel turnover on fault diagnosis capabilities.

[0018] 5. Supports multi-source data fusion diagnosis: It comprehensively processes operating data such as servo motor voltage, current, and temperature with fault text descriptions to achieve intelligent diagnosis through multi-dimensional information fusion.

[0019] 6. Excellent engineering applicability: The method is specifically designed for servo systems and optimized for their common failure modes and performance characteristics. It can be directly applied to the development, testing, and maintenance of aircraft servo systems.

[0020] 7. Supports adaptive optimization: The diagnostic weights are dynamically adjusted through a user feedback mechanism, enabling the system to have self-learning capabilities and continuously improve diagnostic accuracy and adaptability.

[0021] 8. High scalability: The fault knowledge obtained in this invention is applicable to fault diagnosis of electromechanical servo systems of various types of aircraft, effectively improving the design quality of electromechanical servo systems. The method framework can be extended to fault diagnosis of other electromechanical actuation systems, and has broad application prospects and promotional value. Attached Figure Description

[0022] Figure 1 This is a simplified flowchart of a knowledge-driven rapid servo motor fault diagnosis method according to the present invention. Figure 2 This is an algorithm flowchart of a knowledge-driven rapid servo motor fault diagnosis method according to the present invention; Figure 3 This is a flowchart illustrating the specific process of rule mining in this invention; Figure 4 This is a schematic diagram illustrating the specific process of reusing fault cases in this invention.

[0023] Figure 5 This is a schematic diagram illustrating the specific process of knowledge reasoning in this invention. Detailed Implementation

[0024] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] It should be noted that the defects in the solutions in the prior art are all the results of the inventors' practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the embodiments of this application in the following text should be the inventors' contributions to this application in the process of invention and creation, and should not be understood as technical content known to those skilled in the art.

[0026] In view of the technical problems pointed out in the background art, the present invention provides the following embodiments: In one exemplary embodiment, a knowledge-driven rapid servo motor fault diagnosis method is provided, referring to... Figure 1 This includes the following steps: S1. Establish a database of electric steering gear failure cases; S2. Classify and store electric servo motor fault information based on a distributed architecture; S3. Discretize and label the continuous fault indicators in the electric servo motor fault information. The continuous fault indicators include servo motor voltage, current and temperature. The discretization process combines the elbow method and the K-value comprehensive evaluation algorithm to determine the optimal number of clusters for each continuous fault indicator. S4. Based on the discrete data obtained in step S3, use the Apriori association rule algorithm to mine the association rules between electric servo motor fault information and generate a servo motor fault knowledge rule base. S5. Based on the fault information input by the user, reuse fault cases based on the electric servo motor fault case library, and perform fault knowledge reasoning based on the servo motor fault knowledge rule library.

[0027] The following combination Figure 2 Each step of this application is described.

[0028] In step S1, the database is built to store the fault cases of the electric servo motor, and the database is named the Electric Servo Motor Fault Case Library, which serves as the basis for subsequent association rule mining.

[0029] In step S2, the fault (quality) problem includes information such as the time, location, phenomenon, indicators, cause, rectification measures, and subsequent suggestions of the fault occurrence. In view of the complex and numerous fault information, this invention proposes an information processing method based on the Hadoop distributed big data framework based on the core concept of "divide and conquer". This method can quickly, accurately, and effectively analyze and process the information in the fault problem.

[0030] For example, a known fault problem contains basic information (such as time and location) and important information (such as cause and corrective measures). Secondary information can be temporarily disregarded; that is, secondary information does not participate in data mining. The fault problem information is divided into two parts: one part is used as the data foundation for association rule mining, and the other part merely provides supplementary explanations of the quality issues. Data with different roles is used in different computational contexts; therefore, the two types of data need to be stored separately in the cluster nodes of the distributed big data processing technology architecture to exclude the impact of some data on the association rule mining process. This invention ultimately chooses to store the fault problem information in the file management system (HDFS) under the Hadoop distributed big data processing technology architecture deployed above, specifically in three file types: apriori.txt, RedundantData.txt, and data.txt. The apriori.txt file stores information used for association rule mining; RedundantData.txt stores a small amount of information that does not participate in data mining; and data.txt stores all extracted quality problem information. The purpose of this file is to back up all quality problem information and improve system reliability.

[0031] In step S3, the fault information includes diverse representation methods, including textual and data-based information. First, the measurable indicators of the electromechanical servo system, i.e., the data-based information, are discretized. Through the analysis of the function and principle of the electromechanical servo system, it is known that voltage U, current I, and temperature T are indicators that can be directly measured and used for fault diagnosis of electric servo motors. Now, these three indicators are discretized and labeled for fault diagnosis.

[0032] For example, the three indicators of voltage U, current I and temperature T in the fault state described in each case in the electric servo motor fault case library built in step S1 are listed separately to determine the number of clusters K, that is, the three indicators of voltage U, current I and temperature T are divided into several categories respectively.

[0033] The elbow method has limitations such as the lack of a clear elbow point or the existence of multiple candidate elbow points. Therefore, this invention proposes a K-value comprehensive evaluation algorithm to select the optimal number of clusters. This algorithm is based on the initial calculation of the number of clusters using the elbow method. Then, the K-value comprehensive evaluation algorithm is used to evaluate the number of clusters. After optimization, the final optimized number of clusters is: .

[0034] The core idea of ​​the K-value comprehensive evaluation algorithm is as follows: in: , , Indicates weight, ; This function represents the normalization process after inverting the sum of squared errors within a cluster (WCSS); This represents the function after the silhouette coefficients have been normalized. The normalized function representing the ratio of inter-cluster dispersion to intra-cluster dispersion (Calinski_Harabasz); The specific calculation formula for the above function is as follows: The meanings of the specific functions in the above formulas are as follows: The sum of squares within a cluster, which measures the cluster density, is calculated using the following formula: in: C represents the number of clusters initially calculated using the elbow method; i Let X represent the i-th cluster; X represents the sample within the i-th cluster. Let represent the centroid of the i-th cluster.

[0035] It is used to measure the similarity of a sample to its own cluster and the dissociation from other clusters. The specific calculation formula is as follows: in, This represents the average distance from a sample point to other points within the same cluster. This represents the minimum average distance from a sample point to other clusters.

[0036] in, This represents the initial cluster number calculated using the elbow method; N is the total number of samples. The trace of the between-class scatter matrix is ​​used to represent the between-class scatter, and the between-class scatter matrix is... The calculation formula is: in, It is the number of samples in cluster i. is the centroid of cluster i, and m is the global centroid.

[0037] The trace of the within-class scatter matrix is ​​used to represent the within-class scatter, and the within-class scatter matrix is... The calculation formula is: in, It is the sample set of cluster i.

[0038] The cluster numbers of the three indicators—voltage U, current I, and temperature T—were finally determined based on the elbow method and a comprehensive evaluation algorithm. The voltage U of the electromechanical servo system is divided into three categories: U1, U2, and U3; the current I is divided into four categories: I1, I2, and I3; and the temperature T is divided into four categories: T1, T2, and T3.

[0039] Furthermore, all information in the fault cases, including textual and data types, is labeled. The processed results are used for the Apriori association mining algorithm. Different labels are set for each discrete interval. Taking the fault phenomenon as an example, it is set as a dataset labeled "A". Different phenomena are labeled with different labels, namely A1, A2, A3, and A4, respectively. Finally, the fault problem information labels are shown in Table 1 below.

[0040] Table 1 Fault Information Labels In step S4, association rule analysis is performed based on the aforementioned discrete data to acquire knowledge about quality issues. The essence of association rule analysis is to find strong relationships between data. Based on the established fault case database, each given fault issue is considered a transaction; that is, the database is composed of many transactions. When some elements in a transaction exist in the form of a set, this set is called an itemset. This invention selects the Apriori association rule algorithm as the mining algorithm for acquiring knowledge about quality issues. Using itemsets... This paper takes an example to explain in detail the concepts and generalized calculation methods of support and confidence involved in the Apriori algorithm.

[0041] Support: Confidence level: Based on two important theorems in the implementation of the Apriori algorithm: (1) If an itemset is frequent, then all non-empty subsets of that itemset are frequent. (2) If an itemset is infrequent, then its supersets must be infrequent. A scan of the case library is completed. The specific implementation process of the Apriori algorithm is as follows: Figure 3 As shown.

[0042] To obtain high-quality fault knowledge, this invention uses the Apriori algorithm to obtain frequent 1-itemsets, frequent 2-itemsets, frequent 3-itemsets, and frequent 4-itemsets with support of 0.1, 0.15, and 0.2, respectively. Then, a minimum confidence level of 0.6 is set for association rule mining to obtain strong associations between quality problem information. Table 2 below lists the association rules at a minimum support of 0.15.

[0043] Table 2 Association rules with a minimum support of 0.15 The results of the Apriori algorithm are analyzed below. Taking the first association rule as an example, the physical meaning of the rule is briefly explained: A2 indicates that the motor in the electromechanical servo system is stalled; B1 indicates that the AD chip feedback data is abnormal. The first result shows that when the motor in the electromechanical servo system is stalled, the AD chip feedback data is abnormal, and the confidence level of this conclusion is approximately 0.65185.

[0044] Furthermore, the electromechanical servo system knowledge mined in the previous step is used to establish a servo motor fault knowledge rule base, namely, an association rule base, which facilitates the rapid location of fault points when faults occur in the future, and provides a scientific basis for the located fault points and related information.

[0045] In step S5, as Figure 2 As shown, case-based reasoning (CBR) can first be performed based on an electric servo motor fault case library. The core idea of ​​CBR is that similar fault problems have similar solutions. Based on CBR case reasoning, similar cases that have already occurred can be directly used for fault diagnosis. Since the fault cases contain both data types (data and text), this invention creates a comprehensive similarity calculation method. Different methods are used to calculate similarity for different types of data, namely, sub-similarity calculation and total similarity calculation. The specific implementation process is as follows... Figure 4 As shown.

[0046] Sub-similarity calculation is divided into data similarity calculation and text similarity calculation. The formula for data similarity calculation is as follows: The formula for calculating text similarity is as follows: Where A and B are vectors.

[0047] The overall similarity calculation is the comprehensive similarity calculation of the subset similarity calculation. The specific calculation method for the comprehensive similarity calculation is as follows: in, and It belongs to the weighting coefficient, and . This represents the sub-similarity (data similarity) between data types. This represents the sub-similarity (text similarity) corresponding to text-type data. A similarity threshold x is set to be used for cases with a comprehensive similarity ≥ x; otherwise, it is not used.

[0048] Furthermore, the weights are dynamically adjusted using multi-round user feedback data. An example of adjusting a certain weight is as follows: in: and These are the learning rates for data-type data and text-type data, respectively. The learning rate refers to the frequency with which a certain data appears in the historical K rounds of validation, that is, the frequency with which it is learned by the application. Utility is the performance application score of the two types of data retrieval cases. The performance application score refers to the accuracy in the latest validation application. This represents the contribution of the a-th data type in the case, where n represents the total number of data types. is the contribution of the b-th textual data in the case, where m represents the total number of m textual data. The contribution represents the average accuracy over the past K rounds of validation.

[0049] The weighted similarity weights of the weighted data type data adjusted by the above formula are: The similarity weight for text-based data is... The details are as follows: Furthermore, if no cases with high similarity meeting the threshold are found based on CBR case reasoning, knowledge reasoning is then performed based on the rule base. The specific reasoning process is detailed in [link to relevant documentation]. Figure 5 For example: A certain electromechanical servo system product is experiencing a fault phenomenon of large fluctuations in feedback angle data. Through information processing, it is known that the label for large fluctuations in feedback angle data is A1. Now, taking A1 as a known condition, each rule in the association rule base is scanned one by one to obtain other unknown fault knowledge.

[0050] The association rule set mined using the Apriori association rule algorithm is applied to real-world production and analysis for verification. For example, a certain electromechanical servo system exhibits a fault phenomenon of large fluctuations in feedback angle data. This phenomenon is used as known information to quickly locate the fault point and its cause. Information processing results show that the label for large fluctuations in feedback angle data is A1. Using A1 as a known condition, the association rule set is scanned to find fuzzy association rules related to the known condition. The specific query results are shown in Table 3 below.

[0051] Table 3 Association rules obtained based on known conditions Based on the association rules in the table above, fault analysis shows that the fault phenomenon A1 (large fluctuations in feedback angle data) is most likely caused by B1 (abnormal feedback data from the AD chip). The confidence level of this inference is 0.91262. After testing, it was confirmed that the large fluctuations in feedback data were caused by poor soldering of the AD chip pins. The reuse of fault problem knowledge can effectively shorten the design cycle of electromechanical servo systems and improve their R&D efficiency.

[0052] In another exemplary embodiment, based on the same inventive concept as the method, a knowledge-driven rapid servo motor fault diagnosis system is provided, comprising: The fault case library construction module is used to build a fault case library for electric servo motors; The data classification and storage module is used to classify and store electric steering gear fault information based on a distributed architecture. The data processing module is used to discretize and label the continuous fault indicators in the electric servo motor fault information. The continuous fault indicators include servo motor voltage, current and temperature. The discretization process combines the elbow method and the K-value comprehensive evaluation algorithm to determine the optimal number of clusters for each continuous fault indicator. The fault knowledge rule base construction module is used to mine the association rules between electric servo motor fault information based on the discrete data obtained from the data processing module using the Apriori association rule algorithm, and generate a servo motor fault knowledge rule base. The knowledge reuse and reasoning module is used to reuse fault cases based on the electric servo motor fault case library according to the fault information input by the user, and to perform fault knowledge reasoning based on the servo motor fault knowledge rule library.

[0053] The above detailed embodiments are a description of the present invention. It should not be considered that the specific embodiments of the present invention are limited to these descriptions. For those skilled in the art, several simple deductions and substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the protection scope of the present invention.

Claims

1. A knowledge-driven rapid fault diagnosis method for servo motors, characterized in that, include: S1. Establish a database of electric steering gear failure cases; S2. Classify and store electric servo motor fault information based on a distributed architecture; S3. Discretize and label the continuous fault indicators in the electric servo motor fault information. The continuous fault indicators include servo motor voltage, current and temperature. The discretization process combines the elbow method and the K-value comprehensive evaluation algorithm to determine the optimal number of clusters for each continuous fault indicator. S4. Based on the discrete data obtained in step S3, use the Apriori association rule algorithm to mine the association rules between electric servo motor fault information and generate a servo motor fault knowledge rule base. S5. Based on the fault information input by the user, reuse fault cases based on the electric servo motor fault case library, and perform fault knowledge reasoning based on the servo motor fault knowledge rule library.

2. The knowledge-driven rapid servo motor fault diagnosis method according to claim 1, characterized in that, The algorithm combining the elbow method and the K-value comprehensive evaluation method determines the optimal number of clusters for each continuity fault index, including: Preliminary calculation of the number of clusters based on the elbow method; The K-value comprehensive evaluation algorithm is used to optimize the initially calculated number of clusters, specifically including: The optimal number of clusters is determined by comprehensively evaluating the sum of squared errors within clusters, silhouette coefficient, inter-cluster dispersion, and intra-cluster dispersion, and by weighted summation.

3. The knowledge-driven rapid servo motor fault diagnosis method according to claim 1, characterized in that, The tagging process includes: Different labels are assigned to each discrete interval, and different fault phenomena within each discrete interval are labeled with different labels.

4. The knowledge-driven rapid fault diagnosis method for servo motors according to claim 1, characterized in that, The reuse of fault cases based on the electric servo motor fault case library includes: Calculate similarity using different methods for different types of data to obtain sub-similarity for each type of data. The combined similarity scores are obtained by integrating the results of each similarity calculation. Based on this comprehensive similarity, fault cases are screened, and the comprehensive similarity and similar fault cases are output.

5. The knowledge-driven rapid servo motor fault diagnosis method according to claim 4, characterized in that, The formula for calculating the overall similarity is: Among them, the and These are the weighting coefficients. This represents the sub-similarity of corresponding data. This represents the sub-similarity corresponding to text-based data.

6. The knowledge-driven rapid servo motor fault diagnosis method according to claim 4, characterized in that, The reuse of fault cases based on the electric servo motor fault case library also includes: The weights of each sub-similarity are dynamically adjusted based on user feedback.

7. A knowledge-driven rapid fault diagnosis system for servo motors, characterized in that, include: The fault case library construction module is used to build a fault case library for electric servo motors; The data classification and storage module is used to classify and store electric steering gear fault information based on a distributed architecture. The data processing module is used to discretize and label the continuous fault indicators in the electric servo motor fault information. The continuous fault indicators include servo motor voltage, current and temperature. The discretization process combines the elbow method and the K-value comprehensive evaluation algorithm to determine the optimal number of clusters for each continuous fault indicator. The fault knowledge rule base construction module is used to mine the association rules between electric servo motor fault information based on the discrete data obtained from the data processing module using the Apriori association rule algorithm, and generate a servo motor fault knowledge rule base. The knowledge reuse and reasoning module is used to reuse fault cases based on the electric servo motor fault case library according to the fault information input by the user, and to perform fault knowledge reasoning based on the servo motor fault knowledge rule library.

Images