Fault diagnosis method for drive motor

Abstract

The present invention discloses a fault diagnosis method for a drive motor, which relates to the technical field related to measuring electrical variables. According to the physical structure and working principle of the drive motor, a fault information library is retrieved and mined. Physical reconstruction is performed based on the physical structure and working principle to generate physical units, a derivation unit is constructed based on the fault information library, a unit mapping based on the fault mechanism is established, and a fault diagnosis module is generated. A front-end temporary data center is introduced to receive the operating data stream of the drive motor and perform preprocessing, and transmit it to the fault diagnosis module, and execute twin evolution based on the physical unit and classification decision based on the derivation unit, and integrate to determine the fault tracing list, wherein the classification decision includes linear derivation and nonlinear fractal decision under multiple decision steps. The fault tracing list is subjected to confidence verification and terminal visualization display. The technical problem that the motor fault diagnosis lacks real-time and intelligence in the prior art, resulting in low efficiency in fault identification and processing is solved.

Description

Technical Field

[0001] The present application relates to the technical field related to measuring electrical variables, and in particular to a fault diagnosis method for driving a motor. Background Art

[0002] With the widespread use of electric motors in industry and daily life, especially small drive motors, these motors are widely used in household appliances, toys and small equipment due to their small size, light weight and low cost. The importance of fault diagnosis and maintenance is becoming increasingly prominent. Due to their small size, low cost and widespread use, the failure frequency is high and the failure mode is complex. Failures of these motors will not only lead to a decline in product performance, but may also cause safety hazards. However, existing fault diagnosis methods often rely on experience and routine testing, and are unable to analyze and process the large amount of data generated by the operation of small motors in real time, resulting in insufficient timeliness and accuracy in fault identification.

[0003] Therefore, in the prior art, motor fault diagnosis lacks real-time and intelligence, resulting in technical problems of low efficiency in fault identification and processing. Summary of the invention

[0004] This application solves the technical problem that the motor fault diagnosis in the prior art lacks real-time and intelligence, resulting in low efficiency in fault identification and processing, by providing a fault diagnosis method for a drive motor. The technical effect of significantly improving the accuracy and response speed of small drive motor fault diagnosis is achieved.

[0005] The present application provides a fault diagnosis method for a drive motor, which retrieves and mines a fault information library based on the physical structure and working principle of the drive motor, wherein the fault information library contains a fault logic map and a logic rule library. Physical reconstruction is performed based on the physical structure and working principle to generate a physical unit, a derivation unit is constructed based on the fault information library, a unit mapping based on the fault mechanism is established, and a fault diagnosis module is generated. A front-end temporary data center is introduced to receive and pre-process the operation data stream of the drive motor, transmit it to the fault diagnosis module, execute the twin evolution based on the physical unit and the classification decision based on the derivation unit, and integrate to determine the fault tracing list, wherein the classification decision includes linear derivation and nonlinear fractal decision under multiple decision steps. The fault tracing list is confidence-checked and terminal visualized.

[0006] In the implementation, the retrieval and mining of the fault information library includes: retrieving motor fault records, clustering records by fault type, and determining N groups of fault records. Based on the fault scenario, multi-level fault feature and logical relationship mining is performed on the N groups of fault records to determine the fault logic map. The fault logic map is traversed to determine the basic logic rules of fault decision based on the derivation decision principle, and the logic rule library is generated.

[0007] In the implementation, the determination of the fault logic map includes: traversing the N groups of fault records, mining a one-dimensional fault map for dynamic and static dimensions. Traversing the N groups of fault records, mining a two-dimensional fault map for steady-state environments and non-steady-state environments. Traversing the N groups of fault records, mining a three-dimensional fault map for intermittent fault features, wherein a preset frequency is used as a mining standard. Integrating the one-dimensional fault map, the two-dimensional fault map and the three-dimensional fault map as the fault logic map.

[0008] In the implementation, the determination of the fault tracing list includes: identifying the operation data flow and performing pre-screening to determine the abnormal data flow, wherein the pre-screening is a fuzzy screening for abnormal data, and the screening method is a rigid standard based on standard operation conditions. The abnormal data flow is mapped twin and evolutionary decision is performed in the physical unit, and according to the unit mapping, the fault decision deduction triggered by the physical unit is performed in the derivation unit to determine the fault tracing list.

[0009] In the implementation, the fault decision deduction triggered by the physical unit is performed in the deduction unit, including: identifying the abnormal data flow, dividing the linear part and the nonlinear part. Based on the unit mapping, the deduction unit performs linear analysis of the linear part and fractal principle decision of the nonlinear part to determine the fault tracing list.

[0010] In the implementation, the derivation unit performs linear analysis of the linear part and fractal principle decision-making of the nonlinear part, including: for the linear part, introducing a linear function, performing a spectrum decision and fault prediction, and determining the first fault data. For the nonlinear part, extracting fractal features, performing a spectrum decision and fault prediction, and determining the second fault data. Integrating the first fault data and the second fault data to generate the fault tracing sheet.

[0011] In an implementation, the method further includes: determining a fault decision step. Identifying a first decision step, traversing the logic rule base, performing rule matching, and determining a step rule set. Executing rule competition of the step rule set to determine a first decision rule. Combining the first decision rule with a fault map, performing decision derivation of the first decision step.

[0012] In the implementation method, an untrusted fault tracing sequence of a preset period is obtained, a relearning sample is generated and a learning rate is marked, and the fault diagnosis module is updated and learned.

[0013] The fault diagnosis method for the drive motor proposed in this application is intended to retrieve and mine the fault information library through the physical structure and working principle of the drive motor, wherein the fault information library contains a fault logic map and a logic rule library. Physical reconstruction is performed based on the physical structure and working principle to generate a physical unit, a derivation unit is constructed based on the fault information library, a unit mapping based on the fault mechanism is established, and a fault diagnosis module is generated. A front-end temporary data center is introduced to receive the operation data stream of the drive motor and pre-process it, and transmit it to the fault diagnosis module, execute the twin evolution based on the physical unit and the classification decision based on the derivation unit, and integrate and determine the fault tracing list, wherein the classification decision includes linear derivation and nonlinear fractal decision under multiple decision steps. The fault tracing list is confidence-checked and terminal visualized. The technical problem that the motor fault diagnosis lacks real-time and intelligence in the prior art, resulting in low efficiency in fault identification and processing, is solved. The technical effect of significantly improving the accuracy and response speed of small drive motor fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0015] Figure 1 A schematic diagram of a fault diagnosis method for a drive motor provided in an embodiment of the present application;

[0016] Figure 2 A schematic flow chart of generating the logic rule base for the fault diagnosis method for a drive motor provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0018] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0019] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0020] The present application provides a fault diagnosis method for a drive motor, such as Figure 1 As shown, the method includes:

[0021] According to the physical structure and working principle of the drive motor, a fault information library is retrieved and mined, wherein the fault information library includes a fault logic map and a logic rule library.

[0022] Physical reconstruction is performed based on the physical structure and working principle to generate physical units, a derivation unit is constructed based on the fault information library, a unit mapping based on the fault mechanism is established, and a fault diagnosis module is generated.

[0023] With the widespread use of small motors in home appliances, toys and portable devices, the importance of their fault diagnosis and maintenance has become increasingly prominent. In particular, small motors with an output power of less than 200 watts have a high fault frequency and complex fault modes due to their small size, low cost and widespread use. Failures in this type of motor will not only lead to a decline in product performance, but may also cause safety hazards. However, existing fault diagnosis methods often rely on experience and conventional testing, and are unable to analyze and process the large amount of data generated during the operation of small motors in real time, resulting in insufficient timeliness and accuracy in fault identification. According to the physical structure and working principle of the drive motor, the physical structure refers to the hardware components of the drive motor, such as stators, rotors, bearings, coils, etc. The working principle refers to the basic physical and electromagnetic processes of the drive motor when it is working, such as the current passing through the coil to generate a magnetic field, driving the rotor to move. Retrieve and mine the fault information library, which records the physical structure and working principle of various types of drive motors and the fault information generated during operation. The fault information in the fault information library also includes a fault logic map and a logic rule library. The fault logic map is used to describe various possible faults and the logical relationships between them. The logic rule library is a decision rule generated based on these fault relationships. The decision rule can quickly infer the corresponding fault cause based on the specific motor operation data. Based on the physical structure and working principle of the drive motor, the working process of the motor is physically reconstructed and several physical units are generated. Each physical unit represents a subsystem or component of the drive motor, such as a rotor, a stator, etc. Based on the fault logic map and the logic rule library in the fault information library, a derivation unit is constructed. For example, the rotor of the drive motor is physically reconstructed into a physical unit, and the corresponding derivation unit contains the fault logic map and logic rules corresponding to the fault types such as rotor imbalance and rotor winding short circuit, establishes a unit mapping based on the fault mechanism, and generates a fault diagnosis module.

[0024] like Figure 2 As shown, the method provided by the embodiment of the present application also includes: retrieving motor fault records, clustering records by fault type, and determining N groups of fault records. Based on the fault scenario, multi-level fault feature and logical relationship mining is performed on the N groups of fault records to determine the fault logic map. The fault logic map is traversed, and the basic logic rules of fault decision are determined based on the derivation decision principle to generate the logic rule library.

[0025] Retrieve the motor fault records in the history, which include various fault data and corresponding monitoring data of the motor under different environments. According to the type of fault, cluster these records, and divide the fault records into N groups of different categories according to the fault category, such as bearing fault, winding short circuit, overload fault, cooling system fault, etc., to determine N groups of fault records. Based on the fault scenario, multi-level fault feature and logical relationship mining is performed on the N groups of fault records to determine the fault logic map. Traverse the fault logic map, and determine the basic logical rules of fault decision based on the derivation decision principle. The derivation decision principle is to derive possible fault causes according to the fault logic map based on existing data and known rules, determine the basic logical rules of fault decision, and then generate the logic rule base. The logic rule base is a database composed of multiple logic rules, which contains decision rules under different fault logic maps of the system. The logic rule base can quickly match and obtain the corresponding fault cause according to specific motor operation data or data features. Exemplarily, for bearing faults, the following decision rules are generated: if the vibration frequency is greater than a certain set value, and the bearing temperature rise rate is greater than the set value, then the corresponding fault cause is bearing wear.

[0026] The method provided in the embodiment of the present application also includes: traversing the N groups of fault records, mining a one-dimensional fault map for dynamic and static dimensions. Traversing the N groups of fault records, mining a two-dimensional fault map for steady-state environments and non-steady-state environments. Traversing the N groups of fault records, mining a three-dimensional fault map for intermittent fault features, wherein a preset frequency is used as a mining standard. Integrating the one-dimensional fault map, the two-dimensional fault map and the three-dimensional fault map as the fault logic map.

[0027] Traverse N groups of fault records, analyze each group of fault records one by one, and extract fault features for dynamic and static dimensions. For example, in the static dimension, the temperature, position and other information of certain parts of the motor when they are not working. In the dynamic dimension, the state change characteristics of the motor such as speed and vibration frequency are analyzed, and a one-dimensional fault map is constructed. For example, for a winding overheating fault, the system traverses the temperature change data in the record and finds that in the dynamic dimension, the temperature rises rapidly when the motor starts, while in the static dimension, the temperature gradually drops when the motor stops working. Through the above traversal data, a curve chart of temperature changes over time is generated, that is, a one-dimensional fault map.

[0028] After completing the mining of the one-dimensional fault map, traverse the N groups of fault records, and mine the two-dimensional fault map for the steady-state environment and the non-steady-state environment. The steady-state environment refers to the performance of the current, voltage, temperature and other parameters of the motor under the stable operating state, while the non-steady-state environment refers to the performance of the current, voltage, temperature and other parameters of the motor under non-stable conditions such as starting, stopping or sudden load changes. For example, when a motor is running steadily with a constant load, its current remains in a certain stable range, but in a non-steady-state environment such as when the load suddenly increases, the current fluctuates instantly and exceeds the normal range. By analyzing the data under these different environments, a two-dimensional fault map is generated to show the characteristic curve of current fluctuations as the environment changes.

[0029] Traverse the N groups of fault records, analyze the intermittent fault features, and mine the three-dimensional fault map, wherein the preset frequency is used as the mining standard to mine the features of the motor at discontinuous time points. When the frequency of the features at discontinuous time points meets the preset frequency, it is identified as an intermittent fault feature. For example, during the continuous operation of a motor, the abnormal current fluctuation only occurs in certain time periods. By analyzing the data of these time periods, the system finds that the current will suddenly change every about 5 minutes. By setting the preset frequency such as at least 3 times, the system finally generates a three-dimensional fault map to show the fluctuation of the current in different time periods. Finally, the above-mentioned one-dimensional fault map (dynamic and static dimensions), two-dimensional fault map (steady-state and non-steady-state environment) and three-dimensional fault map (fault intermittent features) are integrated to form a complete fault logic map as the fault logic map. The fault logic map shows the fault features under all dimensions and reveals the logical connection between them, providing data support for the analysis of subsequent faults.

[0030] A temporary data center is introduced to receive and pre-process the operation data stream of the drive motor, transmit it to the fault diagnosis module, execute the twin evolution based on the physical unit and the classification decision based on the derivation unit, and integrate to determine the fault tracing list, where the classification decision includes linear derivation and nonlinear fractal decision under multiple decision steps. The fault tracing list is verified with confidence and visualized on the terminal.

[0031] The front temporary data center is used to receive the operation data stream of the drive motor and perform preprocessing. The temporary data center is used to receive the real-time operation data stream of the drive motor, such as sensor data such as current, voltage, speed, temperature, etc. These data are first preprocessed to remove noise or outliers for subsequent analysis. The processed data is transmitted to the fault diagnosis module, and the twin evolution based on the physical unit and the classification decision based on the derivation unit are executed to integrate and determine the fault tracing list, wherein the classification decision includes linear derivation and nonlinear fractal decision under multiple decision steps. After the fault tracing list is generated, a confidence check is performed on it, that is, historical data related to the fault is collected, including known fault cases and corresponding fault confidence, and historical fault records and corresponding confidence are input to train the machine learning model. The confidence of the fault in the fault tracing list is obtained based on the trained machine learning model, and the accuracy of the fault tracing is verified by threshold judgment to ensure the reliability of the decision result, and the diagnosis result is displayed visually on the terminal, so that engineers can view the source of the fault and its derivation process. The technical problem of the lack of real-time and intelligence in the motor fault diagnosis in the prior art, which leads to low efficiency in fault identification and processing, is solved. The technical effect of significantly improving the accuracy and response speed of small drive motor fault diagnosis.

[0032] The method provided in the embodiment of the present application also includes: identifying the operating data flow and performing pre-screening to determine the abnormal data flow, wherein the pre-screening is a fuzzy screening for abnormal data, and the screening method is a rigid standard based on standard operation conditions. The abnormal data flow is mapped twin and evolutionary decision is performed in the physical unit, and according to the unit mapping, the fault decision deduction triggered by the physical unit is performed in the derivation unit to determine the fault tracing list.

[0033] The determination of the fault tracing list includes: identifying the operation data stream and performing pre-screening to separate normal data from abnormal data that may have faults. The pre-screening is a fuzzy screening for abnormal data. During the screening, the corresponding parameters of the acquired operation data stream are screened based on the various parameter standards of the normal operation of the motor, that is, the rigid standards of the standard operation conditions, and the data corresponding to the rigid standards that do not meet the standard operation conditions are obtained. The fuzzy screening is completed to determine the abnormal data stream. Subsequently, the abnormal data stream is mapped twin and evolutionary decision in the physical unit, that is, after the abnormal data stream is screened out, these abnormal data are mapped to the corresponding physical unit, and the operation state is simulated based on the real-time data of the physical unit through the twin technology to obtain the impact of the abnormal data on the various components of the motor, thereby obtaining the evolution process of the fault, and then obtaining whether the abnormal data will lead to further development of the fault. For example, the simulated temperature increase will cause a certain component to overheat, so based on the operation simulation, the fault that may be caused by the operation simulation is obtained, and the evolutionary decision is obtained. According to the unit mapping, the fault decision deduction triggered by the physical unit is performed in the deduction unit to determine the fault tracing list.

[0034] The method provided by the embodiment of the present application further includes: identifying the abnormal data flow, dividing the linear part and the nonlinear part. Based on the unit mapping, the derivation unit performs linear analysis of the linear part and fractal principle decision of the nonlinear part to determine the fault tracing list.

[0035] The fault decision deduction triggered by the physical unit is performed in the deduction unit, including: dividing the received abnormal data stream into a linear part and a nonlinear part, the linear part refers to the part where the change trend in the data presents a straight line relationship or a simple proportional relationship, which can usually be analyzed by a conventional linear model. For example, the current shows a linear growth as the load increases. The nonlinear part refers to the part where the change in the data cannot be described by a linear relationship, which usually manifests as complex fluctuations or irregular phenomena and may involve multiple influencing factors. For example, temperature fluctuations under different loads present complex nonlinear growth or mutations.

[0036] Based on the unit mapping, the abnormal data is mapped to the physical unit (such as the rotor, stator, winding, etc. of the motor) to ensure that the object of data analysis is clear. For the linear part that has been divided, the data is linearly analyzed. Through linear analysis, the linear change trend under normal and abnormal conditions is obtained, and a change spectrum is generated. The spectrum decision and fault prediction are performed. According to the change spectrum, it is detected whether a physical unit has an abnormal situation that deviates from the normal working state, and the fault prediction is completed accordingly. For example, when the load of the motor winding increases, the current rises linearly. If the system detects that the change of the winding current deviates from this linear relationship in the abnormal data stream (for example, the growth rate is too fast or the change is too slow), it can be determined through linear analysis that the current anomaly comes from a winding fault, such as winding heating or short circuit. Based on the unit mapping, the derivation unit performs linear analysis of the linear part and fractal principle decision of the nonlinear part to determine the fault tracing unit.

[0037] The method provided in the embodiment of the present application further includes: for the linear part, introducing a linear function, performing a graph decision and fault prediction, and determining the first fault data. For the nonlinear part, extracting fractal features, performing a graph decision and fault prediction, and determining the second fault data. Integrating the first fault data and the second fault data, and generating the fault tracing sheet.

[0038] For the linear part, a linear function is introduced, and the linear change trend under normal and abnormal conditions is obtained through linear analysis, and a change spectrum is generated to perform spectrum decision and fault prediction, that is, a spectrum with high similarity is obtained in the fault diagnosis module according to the change spectrum, and the spectrum decision is completed, and the fault information corresponding to the spectrum is obtained, and the fault information includes a specific fault category and possible fault causes, so as to determine the first fault data. For the nonlinear part, a fractal analysis method is used to extract complex features in the data to obtain fractal features. According to the fractal features, the fractal features include multiple fractal features such as fractal dimensions. According to the fractal features, the spectrum corresponding to the fractal features with high similarity is obtained in the fault diagnosis module, and the spectrum decision is completed. According to the spectrum, the corresponding fault information is obtained, and the fault information includes a specific fault category and possible fault causes, and the second fault data is determined, such as the fractal features in the vibration signal, such as fractal dimension, multi-fractal spectrum, etc. When a device fails, the fractal features of the vibration signal will change, such as for the current signal, the fractal dimension of the current waveform is extracted. Finally, the first fault data and the second fault data are integrated to generate the fault tracing sheet, which includes fault information and corresponding data features.

[0039] The method provided in the embodiment of the present application also includes: determining a fault decision step. Identifying a first decision step, traversing the logic rule base, performing rule matching, and determining a step rule set. Executing rule competition of the step rule set to determine a first decision rule. Combining the first decision rule with the fault map, performing decision derivation of the first decision step.

[0040] After determining the fault tracing sheet, the fault decision step is determined according to the fault category in the fault tracing sheet. Each fault category has multiple preset decision steps, and the decision step is used to make the final diagnosis and judgment of the fault cause. Subsequently, the first decision step is identified, which is the first step in the fault decision step. The logic rule base is traversed to find the most relevant rule by analyzing the current operation data and historical fault records. For example, if the current current of the motor is overloaded, all rules related to "current overload" in the rule base are searched. Rules are matched to determine the step rule set. The step rule set contains all relevant fault rules for subsequent decision-making. For example, if multiple fault causes related to "current overload" such as overheating and mechanical resistance are matched, these causes will form the step rule set. The rule competition of the step rule set is executed, that is, the step rule with the highest correlation with the fault in the step rule set is obtained. The specific correlation is obtained based on the frequent items of the fault characteristics, and the first decision rule is determined. The first decision rule is the step rule with the highest correlation with the fault. Finally, the decision derivation of the first decision step is performed by combining the first decision rule with the fault spectrum. That is, the possible causes of the fault are analyzed by combining the first decision rule with the fault map. For example, if the first decision rule is mechanical resistance, its possible causes such as poor lubrication, component wear, etc. can be analyzed through the fault map to further deduce the specific cause of the fault. Obtain an untrusted fault tracing sequence of a preset period, that is, obtain uncertain or incompletely reliable fault data collected within a specific period (such as one day, two days, etc.). Generate relearning samples and identify the learning rate. Create a data set for model retraining based on the collected untrusted fault data. These samples will contain information labeled with the fault type and related features. By inputting new relearning samples into the diagnostic module to establish a unit mapping based on the fault mechanism, the diagnostic module can better adapt to new data and improve the diagnostic accuracy. Update the fault diagnosis module for learning.

[0041] The technical solution provided by the embodiment of the present invention retrieves and mines the fault information library according to the physical structure and working principle of the drive motor, wherein the fault information library includes a fault logic map and a logic rule library. Physical reconstruction is performed based on the physical structure and working principle to generate a physical unit, a derivation unit is constructed based on the fault information library, a unit mapping based on the fault mechanism is established, and a fault diagnosis module is generated. A front-end temporary data center is introduced to receive the operation data stream of the drive motor and perform preprocessing, transmit it to the fault diagnosis module, execute twin evolution based on the physical unit and classification decision based on the derivation unit, and integrate to determine the fault tracing list, wherein the classification decision includes linear derivation and nonlinear fractal decision under multiple decision steps. The fault tracing list is subjected to confidence verification and terminal visualization display. The technical problem that the motor fault diagnosis lacks real-time and intelligence in the prior art, resulting in low efficiency in fault identification and processing, is solved. The technical effect of significantly improving the accuracy and response speed of small drive motor fault diagnosis is achieved.

[0042] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A fault diagnosis method for a drive motor, characterized in that: The method comprises: According to the physical structure and working principle of the drive motor, a fault information library is retrieved and mined, wherein the fault information library includes a fault logic map and a logic rule library; Perform physical reconstruction based on the physical structure and working principle to generate physical units, build derivation units based on the fault information library, establish unit mapping based on fault mechanism, and generate fault diagnosis modules; Introducing a front-end temporary data center, receiving and preprocessing the operation data stream of the drive motor, transmitting it to the fault diagnosis module, executing the twin evolution based on the physical unit and the classification decision based on the derivation unit, integrating and determining the fault tracing list, wherein the classification decision includes linear derivation and nonlinear fractal decision under multiple decision steps; Performing confidence verification and terminal visual display on the fault tracing sheet; The determining of the fault tracing sheet includes: Identify the operation data flow and perform pre-screening to determine abnormal data flow, wherein the pre-screening is a fuzzy screening for abnormal data, and the screening method is a rigid standard based on standard operation conditions; Mapping twins and evolutionary decisions are performed on the abnormal data flow in the physical unit, and according to the unit mapping, fault decision deduction triggered by the physical unit is performed in the derivation unit to determine the fault tracing list; Performing fault decision derivation triggered by a physical unit in the derivation unit includes: Identify the abnormal data stream and divide it into a linear part and a nonlinear part; Based on the unit mapping, the derivation unit performs linear analysis of the linear part and fractal principle decision making of the nonlinear part to determine the fault tracing list.

2. The fault diagnosis method for a drive motor according to claim 1, characterized in that: The retrieving and mining the fault information database includes: Retrieve motor fault records, cluster the records by fault type, and determine N groups of fault records; Based on the fault scenario, multi-level fault feature and logical relationship mining is performed on the N groups of fault records to determine the fault logic map; The fault logic map is traversed, basic logic rules for fault decision are determined based on the derivation decision principle, and the logic rule base is generated.

3. The fault diagnosis method for a drive motor according to claim 2, characterized in that: The determining of the fault logic map comprises: Traversing the N groups of fault records, mining a one-dimensional fault map for dynamic and static dimensions; Traversing the N groups of fault records, mining two-dimensional fault maps for steady-state environments and non-steady-state environments; Traversing the N groups of fault records, mining a three-dimensional fault map based on the intermittent fault characteristics, wherein a preset frequency is used as a mining standard; The one-dimensional fault map, the two-dimensional fault map and the three-dimensional fault map are integrated as the fault logic map.

4. The fault diagnosis method for a drive motor according to claim 1, characterized in that: The derivation unit performs linear analysis of the linear part and fractal principle decision making of the nonlinear part, including: For the linear part, a linear function is introduced to perform graph decision making and fault prediction to determine the first fault data; For the nonlinear part, fractal features are extracted, graph decision making and fault prediction are performed, and the second fault data is determined; The first fault data and the second fault data are integrated to generate the fault tracing sheet.

5. The fault diagnosis method for a drive motor according to claim 4, characterized in that: The method further comprises: Identify the failure decision steps; Identify the first decision step, traverse the logic rule base, perform rule matching, and determine a step rule set; Execute the rule competition of the rule set in the step to determine the first decision rule; The decision derivation of the first decision step is performed in combination with the first decision rule and the fault spectrum.

6. The fault diagnosis method for a drive motor according to claim 1, characterized in that: An untrusted fault tracing sequence of a preset period is obtained, a relearning sample is generated and a learning rate mark is performed, and the fault diagnosis module is updated and learned.

Images