Motor driver anomaly detection method, device, equipment and medium

Abstract

The application relates to the motor driver monitoring technical field and discloses a motor driver anomaly detection method, device, equipment and medium. The method comprises the following steps: acquiring power dimension parameters, pulse dimension parameters and equipment dimension parameters of a motor driver; comparing the power dimension parameters, the pulse dimension parameters and the equipment dimension parameters with corresponding self-adaptive changeable parameter thresholds respectively, so as to determine corresponding abnormal power dimension parameters, abnormal pulse parameters and / or abnormal equipment parameters based on comparison results; based on a preset parameter correlation rule, performing correlation analysis on the abnormal power dimension parameters, the abnormal pulse parameters and / or the abnormal equipment parameters to obtain corresponding correlation analysis results; and based on the correlation analysis results and the abnormal power dimension parameters, the abnormal pulse parameters and / or the abnormal equipment parameters, generating an anomaly detection result for the motor driver. The application can improve the deep processing performance of detection parameters and the correlation analysis performance of complex faults.

Description

Technical Field

[0001] This application relates to the field of motor driver monitoring technology, and in particular to a method, device, equipment and medium for detecting abnormalities in motor drivers. Background Technology

[0002] In related technologies, anomaly monitoring solutions for motor drives include external centralized monitoring systems and simple monitoring systems integrated into the motor drive itself. External centralized monitoring systems rely heavily on direct comparison between preset fixed parameter thresholds and the collected data for diagnostic logic. Simple integrated monitoring systems typically only collect a limited number of electrical parameters such as voltage and current, and trigger alarms through simple fixed thresholds, lacking the ability to perform in-depth data processing and correlate complex faults. Summary of the Invention

[0003] The purpose of this application is to provide a method, apparatus, device, and medium for detecting abnormalities in motor drives, aiming to improve the performance of in-depth processing of detection parameters and the performance of correlation analysis of complex faults.

[0004] This application provides a method for detecting abnormalities in a motor driver, including:

[0005] Obtain the power supply dimension parameters, pulse dimension parameters, and device dimension parameters of the motor driver;

[0006] The power dimension parameters, pulse dimension parameters, and device dimension parameters are compared with corresponding parameter thresholds to determine the corresponding abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on the comparison results; the parameter thresholds are adaptively changed based on the real-time status information and fault history information of the motor driver.

[0007] Based on preset parameter association rules, association analysis is performed on the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters to obtain the corresponding association analysis results.

[0008] Based on the correlation analysis results, the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters, anomaly detection results are generated for the motor driver.

[0009] In some embodiments, before comparing the power dimension parameter, the pulse dimension parameter, and the device dimension parameter with their respective parameter thresholds, the method further includes:

[0010] The power dimension parameters, pulse dimension parameters, and device dimension parameters are sequentially subjected to time alignment, wavelet denoising, and adaptive filtering to obtain preprocessed power dimension parameters, pulse dimension parameters, and device dimension parameters.

[0011] In some embodiments, the adaptive change method for the parameter threshold includes:

[0012] Based on the motor driver, determine the corresponding original parameter thresholds and original bias values;

[0013] Based on the real-time status information and the fault history information, the original bias value is adjusted to obtain the parameter bias value;

[0014] The parameter threshold is calculated based on the original parameter threshold and the parameter bias value.

[0015] In some embodiments, adjusting the original bias value based on the real-time status information and the fault history information includes:

[0016] Based on the real-time status information, the parameter range of the current linear parameter is determined;

[0017] Based on the fault history information and the parameter range, the current linear parameters are adjusted to obtain updated linear parameters;

[0018] Based on the updated linear parameters, a linear operation is performed on the original bias value to obtain the parameter bias value.

[0019] In some embodiments, adjusting the current linear parameter based on the fault history information and the parameter range includes:

[0020] When the fault history information meets the preset parameter update conditions, an objective function with the linear parameter as the variable is established. Within the parameter range, the objective function is optimized with the goal of maximizing the anomaly detection rate and minimizing the false alarm rate. After optimization, the updated linear parameter is obtained.

[0021] In some embodiments, the correlation analysis of the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters based on preset parameter correlation rules includes:

[0022] The abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters are combined to obtain a co-occurring abnormal parameter group;

[0023] Based on the preset mapping relationship between co-occurrence anomaly parameters and anomaly items, a number of anomaly items corresponding to the co-occurrence anomaly parameter group are determined.

[0024] Anomaly tracing reasoning is performed on each of the aforementioned anomalies to determine the root cause anomaly among them, thereby obtaining the correlation analysis results.

[0025] In some embodiments, the anomaly tracing and reasoning for each of the anomalous items includes:

[0026] Construct all possible combinations of anomaly root causes using each of the aforementioned anomaly items;

[0027] Calculate the joint likelihood of each of the aforementioned combinations of anomalies;

[0028] The overall confidence level of each of the anomaly root source combinations is evaluated based on the joint likelihood and the prior probability of the root source anomaly items in the anomaly root source combinations.

[0029] The combination of anomalies with the highest overall confidence is output as the result of the association analysis.

[0030] This application embodiment also provides a motor driver malfunction detection device, including:

[0031] The first module is used to obtain the power supply dimension parameters, pulse dimension parameters, and device dimension parameters of the motor driver;

[0032] The second module is used to compare the power dimension parameters, the pulse dimension parameters, and the device dimension parameters with corresponding parameter thresholds, respectively, so as to determine the corresponding abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on the comparison results; the parameter thresholds are adaptively changed based on the real-time status information and fault history information of the motor driver.

[0033] The third module is used to perform correlation analysis on the abnormal power supply dimension parameters, the abnormal pulse parameters and / or the abnormal device parameters based on preset parameter correlation rules, and to obtain the corresponding correlation analysis results.

[0034] The fourth module is used to generate anomaly detection results for the motor driver based on the correlation analysis results, the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters.

[0035] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described motor driver anomaly detection method.

[0036] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described motor driver anomaly detection method.

[0037] The beneficial effects of this application are as follows: Adaptive parameter thresholds are used to determine abnormal parameters among the multi-dimensional parameters of the motor driver. Then, correlation analysis is performed through parameter association rules to identify the inherent relationships between these abnormal parameters, thereby locating the root cause of the fault and serving as the final anomaly detection result. Therefore, by introducing adaptive parameter thresholds, threshold detection can better adapt to the complex and ever-changing operating environment of the motor driver, improving the accuracy and robustness of the detection. Simultaneously, the introduction of correlation analysis based on preset parameter association rules after threshold detection allows for the identification of the inherent relationships between multiple abnormal parameters when they are detected, accurately locating the root cause of the fault, providing users with clear decision-making basis, and improving the performance of in-depth parameter processing and correlation analysis of complex faults. Attached Figure Description

[0038] Figure 1 This diagram illustrates the application environment of the motor driver anomaly detection method provided in this embodiment.

[0039] Figure 2 This is a flowchart of the motor driver anomaly detection method provided in the embodiments of this application.

[0040] Figure 3 This is a flowchart of the adaptive change method for parameter thresholds provided in the embodiments of this application.

[0041] Figure 4 This is a flowchart of a method for performing correlation analysis on abnormal power supply dimension parameters, abnormal pulse parameters, and / or abnormal device parameters, provided in an embodiment of this application.

[0042] Figure 5 This is a schematic diagram of the structure of the motor driver abnormality detection device provided in the embodiments of this application.

[0043] Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0045] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and drawings are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0047] The motor driver anomaly detection method provided in this application can be executed by a computer device, which can be a terminal device or a server. The terminal device includes, but is not limited to, mobile phones, computers, smart home appliances, vehicle terminals, aircraft, and processors built into the motor driver. The server can be a standalone physical server, a server cluster consisting of multiple physical servers, a distributed system, or a cloud server. Furthermore, the information, data, and signals involved in this application's embodiments are all authorized by the relevant parties or have been fully authorized by all parties, and the collection, use, and processing of related data comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0048] In traditional motor drive anomaly monitoring technologies, diagnostic logic primarily relies on direct comparison of preset fixed parameter thresholds with collected data. These thresholds cannot be dynamically adjusted based on the motor drive's real-time operating status and historical fault data. Furthermore, anomaly detection for power supply, pulse, and equipment-level parameters is handled independently, lacking a multi-dimensional parameter collaborative analysis mechanism based on preset association rules. This limits the accuracy and comprehensiveness of anomaly detection results, thus reducing system reliability. For example, during the operation of an industrial robot joint drive system, when the power supply voltage experiences a momentary shift due to grid fluctuations, a fixed threshold triggers a false alarm, while simultaneously, abnormal pulse signal frequency and increased equipment casing temperature are not correlated or identified. Moreover, in this scenario, because the parameter thresholds are not adaptively adjusted in conjunction with real-time load conditions, the anomaly detection process cannot distinguish between normal operating condition fluctuations and actual faults, resulting in the omission of potential bearing wear faults and increasing the risk of unplanned production line downtime.

[0049] If the above problems are not solved, the abnormal detection capability of the motor drive will not be able to adapt to complex operating conditions, which may lead to the omission or misjudgment of critical faults, and thus cause a chain of equipment damage. Among them, the lack of parameter correlation analysis makes fault tracing difficult, maintenance response time is prolonged, and system availability continues to decline.

[0050] Based on this, embodiments of this application provide a method, apparatus, device, and medium for detecting anomalies in a motor driver. By using adaptive parameter thresholds to determine abnormal parameters among the multi-dimensional parameters of the motor driver, and then performing correlation analysis through parameter association rules, the final anomaly detection result is generated. This can improve the performance of in-depth processing of detection parameters and the correlation analysis performance of complex faults.

[0051] Figure 1 This diagram illustrates the application environment of the motor driver anomaly detection method provided in this embodiment. (See attached diagram.) Figure 1 This method is applied to a motor driver anomaly detection system. The system includes a terminal 110 and a server 120. The terminal 110 and server 120 are connected via a network. The terminal 110 can be a desktop terminal or a mobile terminal; the mobile terminal can be at least one of a mobile phone, tablet, or laptop. The server 120 can be a standalone server or a server cluster consisting of several servers. The terminal 110 sends power dimension parameters, pulse dimension parameters, and device dimension parameters of the motor driver to the server 120. The server 120 acquires these parameters, compares them with corresponding parameter thresholds, determines the corresponding abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on the comparison results, performs correlation analysis on the abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on preset parameter association rules, obtains corresponding correlation analysis results, and generates anomaly detection results for the motor driver based on the correlation analysis results and the abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters. The parameter thresholds adaptively change based on the real-time status information and fault history information of the motor driver.

[0052] It should be understood that Figure 1 The application scenarios shown are merely examples. In practical applications, the motor driver anomaly detection method provided in this application embodiment can also be applied to other scenarios. For example, the above-described motor driver anomaly detection method can be directly applied to terminal 110. Terminal 110 is used to obtain the power dimension parameters, pulse dimension parameters, and device dimension parameters of the motor driver, compare the power dimension parameters, pulse dimension parameters, and device dimension parameters with the corresponding parameter thresholds, and determine the corresponding abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on the comparison results. Based on preset parameter association rules, association analysis is performed on the abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters to obtain the corresponding association analysis results. Based on the association analysis results and the abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters, anomaly detection results for the motor driver are generated.

[0053] See Figure 2 In one embodiment, a method for detecting abnormalities in a motor driver is provided. The execution subject of this method may be a terminal or a server, including but not limited to steps S201 to S204.

[0054] Step S201: Obtain the power supply dimension parameters, pulse dimension parameters, and device dimension parameters of the motor driver.

[0055] Power supply parameters refer to various electrical parameters related to the power supply of the motor driver, such as input voltage, input current, power factor, and harmonic content. These parameters reflect the quality of the power supply and the load characteristics of the motor driver on the power supply.

[0056] Pulse dimension parameters refer to parameters related to the control signals output from the motor driver to the motor or the motor feedback signals, such as the duty cycle, frequency, pulse sequence integrity, and encoder feedback pulse count of the pulse width modulation (PWM) signal. These parameters directly affect the motor's operating state and control accuracy.

[0057] Device-level parameters refer to parameters related to the hardware or software operating status of the motor driver itself, such as internal temperature, cooling fan speed, control board voltage, firmware version, error codes, and operating mode. These parameters reflect the health status and operating environment of the motor driver.

[0058] Step S202: Compare the power dimension parameters, pulse dimension parameters, and device dimension parameters with the corresponding parameter thresholds, respectively, to determine the corresponding abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on the comparison results.

[0059] The parameter thresholds adaptively change based on the real-time status information and fault history information of the motor driver. In essence, a parameter threshold is a boundary value used to determine whether various parameters are within the normal range. When the actual collected parameter values ​​exceed this threshold, they are considered abnormal. This threshold is not fixed but can be dynamically adjusted according to the actual operating conditions of the motor driver.

[0060] Real-time status information refers to various operating data collected in real time by sensors or internal monitoring systems during the operation of the motor drive, such as current load, ambient temperature, running time, speed, and torque. This information reflects the current operating condition of the motor drive.

[0061] Fault history information refers to records of past motor drive failure events, including fault type, occurrence time, operating parameters at the time, environmental conditions, and maintenance records. This information is of great value for analyzing fault modes, predicting future faults, and optimizing anomaly detection strategies.

[0062] Step S203: Based on preset parameter association rules, perform association analysis on abnormal power supply dimension parameters, abnormal pulse parameters, and / or abnormal equipment parameters to obtain the corresponding association analysis results.

[0063] Parameter association rules refer to a predefined series of logical relationships or patterns used to describe the possible causal relationships or co-occurrence patterns between different abnormal parameters. These rules can be used to correlate multiple independent abnormal parameters to infer deeper causes of failures.

[0064] Correlation analysis refers to the process of comprehensively analyzing detected abnormal power supply parameters, abnormal pulse parameters, and / or abnormal equipment parameters using parameter correlation rules. Its purpose is to identify potential correlations among multiple abnormal phenomena, thereby pinpointing the root cause of the fault or predicting its development trend.

[0065] Step S204: Based on the correlation analysis results and abnormal power supply dimension parameters, abnormal pulse parameters and / or abnormal device parameters, generate abnormal detection results for the motor driver.

[0066] This embodiment provides a method for detecting abnormalities in a motor driver, the main technical features of which are described in detail below:

[0067] First, the method includes acquiring power-level parameters, pulse-level parameters, and device-level parameters of the motor driver. Acquiring these parameters is fundamental to anomaly detection. For example, power-level parameters can be acquired in real-time by installing voltage and current sensors at the motor driver input, or obtained through the driver's internal power management module. Pulse-level parameters can be obtained by monitoring the PWM signal output from the motor driver to the motor, or through feedback signals from the encoder connected to the motor. Device-level parameters can be acquired by reading temperature sensors, fan speed sensors, and internal diagnostic registers within the motor driver. In one implementation, these parameters can be manually input into the actuator by an operator, or acquired through a simple periodic sampling mechanism.

[0068] Secondly, the method compares the acquired power dimension parameters, pulse dimension parameters, and device dimension parameters with their corresponding parameter thresholds to determine the corresponding abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on the comparison results. These parameter thresholds adaptively change based on the real-time status information and fault history information of the motor driver. Specifically, the comparison between parameters and thresholds can be done through direct numerical comparison; for example, a parameter value exceeding a preset upper limit or falling below a preset lower limit is marked as abnormal. In one implementation, the parameter threshold can be set to a fixed value, for example, based on the rated operating range of the motor driver. However, to adapt to the operating characteristics of the motor driver under different operating conditions, the parameter threshold can be adaptively adjusted. For example, the threshold can be initially dynamically adjusted based on the current load rate, ambient temperature, and other real-time status information of the motor driver, as well as the statistical distribution of parameters in historical operating data.

[0069] Furthermore, based on pre-defined parameter association rules, this method performs correlation analysis on the abnormal power supply dimension parameter, the abnormal pulse parameter, and / or the abnormal device parameter to obtain the corresponding correlation analysis results. When multiple parameters are abnormal simultaneously, there may be an inherent connection between these anomalies. For example, an abnormal power supply voltage may cause the internal temperature of the motor driver to rise, thereby affecting the stability of the pulse output. Parameter association rules can be pre-defined as a series of logical judgment statements, such as "If the power supply voltage is abnormal and the device temperature is abnormal, then there may be a power overload problem." In one implementation, these association rules can be manually configured by domain experts based on experience, or matched using a simple lookup table.

[0070] Finally, based on the correlation analysis results, the abnormal power supply dimension parameter, the abnormal pulse parameter, and / or the abnormal device parameter, the method generates an anomaly detection result for the motor driver. This anomaly detection result is a comprehensive judgment of the current abnormal state of the motor driver. For example, if the correlation analysis results indicate a power module failure, and an abnormal power supply dimension parameter is detected simultaneously, a detection result of "power module failure" can be generated. In one implementation, the anomaly detection result can be a simple alarm signal, such as an indicator light illuminating or a buzzer sounding, or a text message containing a description of the basic anomaly type.

[0071] The following example will provide a more detailed explanation of the above technical solution:

[0072] Imagine an industrial production line where a motor driver controls a critical conveyor motor. During daily operation, this driver may encounter various anomalies, such as power fluctuations, changes in motor load, and aging internal components. These conditions can lead to decreased equipment performance or even shutdown. Traditional monitoring systems, due to their reliance on fixed thresholds and lack of correlation analysis capabilities, often struggle to accurately identify these complex or intermittent faults, potentially resulting in false alarms or missed alarms.

[0073] The anomaly detection method in this embodiment is applied to the motor driver. First, the execution entity continuously acquires the power supply dimension parameters (e.g., input voltage, current), pulse dimension parameters (e.g., PWM output duty cycle, frequency), and device dimension parameters (e.g., internal temperature, fan speed) of the motor driver. These parameters are acquired in real time through sensors and control units integrated inside the motor driver.

[0074] During the parameter comparison phase, the executing entity compares these real-time acquired parameters with the corresponding parameter thresholds. Unlike traditional fixed thresholds, the parameter thresholds in this embodiment are adaptively variable. For example, when the conveyor belt motor is running under heavy load, its normal operating current and temperature will be higher than when it is under no-load. At this time, the executing entity will dynamically adjust the current and temperature thresholds based on real-time load information (a type of real-time status information) and the parameter distribution under heavy load conditions in historical operating data (a type of fault history information). Thus, under heavy load conditions, even if the current and temperature increase, as long as they remain within the adjusted normal range, they will not be misjudged as abnormal. Conversely, if the motor driver experiences abnormally high temperatures under light load conditions, even if it does not reach the traditional fixed threshold, but exceeds the adaptive threshold under light load conditions, it will be promptly identified as abnormal.

[0075] When the actuator detects multiple parameters simultaneously exceeding their adaptive thresholds—for example, slightly high input current, fluctuating PWM duty cycle, and increased internal temperature—these anomalous parameters are flagged. The actuator then performs correlation analysis on these anomalous parameters based on pre-defined parameter correlation rules. For example, the pre-defined rules might include logic such as, "If the input current is abnormal and the PWM duty cycle fluctuates, there may be excessive motor load or a driver output stage fault." Through this correlation analysis, the actuator can link seemingly independent anomalies, thereby inferring deeper causes of faults. For instance, if the correlation analysis indicates a "driver output stage fault," and anomalies are simultaneously detected in both pulse-dimensional and device-dimensional parameters, the correlation analysis result is further confirmed.

[0076] Finally, based on the correlation analysis results and the labeled abnormal power supply dimension parameters, abnormal pulse dimension parameters, and device dimension parameters, the executing entity generates anomaly detection results for the motor driver. For example, the executing entity may output detailed diagnostic information such as "Driver output stage fault, it is recommended to check the power module," rather than a simple "alarm" signal. This detection result provides maintenance personnel with clear fault location and handling suggestions, enabling rapid response and problem resolution, and avoiding prolonged production line downtime. In this way, the method of this embodiment can effectively solve the false alarm and missed alarm problems caused by fixed thresholds in traditional solutions and improve the diagnostic capability for complex faults.

[0077] Based on the above examples, the method provided in this embodiment demonstrates significant technical contributions.

[0078] Compared to related technologies that rely on preset fixed parameter thresholds, this embodiment overcomes the limitations of traditional solutions that suffer from false alarms or missed alarms under different operating conditions by introducing parameter thresholds that adaptively change based on real-time status information and fault history information of the motor driver. For example, in the conveyor belt motor example above, when the motor load changes, this method can dynamically adjust the threshold to ensure that false alarms are not triggered under normal operating condition changes, while being able to identify actual anomalies in a timely and accurate manner. This adaptive capability allows the actuator to better adapt to the complex and ever-changing operating environment of the motor driver, improving the accuracy and robustness of detection.

[0079] Furthermore, the method in this embodiment acquires not only power supply and pulse parameters but also device parameters, enabling multi-dimensional and comprehensive status monitoring of the motor driver. This contrasts with the simple monitoring systems in the background that only collect a limited number of electrical parameters. By comprehensively analyzing parameters from the three dimensions of power supply, pulse, and device, this method obtains more comprehensive information, providing a more solid data foundation for subsequent anomaly detection.

[0080] More importantly, this embodiment introduces correlation analysis based on preset parameter association rules. When multiple abnormal parameters are detected, this method can perform in-depth analysis of these anomalies, identify the inherent relationships between them, and thus locate the root cause of the fault. This is significantly superior to solutions in the background art that lack the ability to perform in-depth data processing and complex fault correlation analysis. In the example above, through correlation analysis, the executing entity can infer the underlying cause of "driver output stage failure" from multiple anomalies, rather than simply reporting multiple independent parameter out-of-limits errors. This capability allows this method to delve into the essence from the surface, providing more instructive diagnostic results.

[0081] Finally, this embodiment generates anomaly detection results for the motor drive based on the correlation analysis results and abnormal parameters. These results not only indicate the existence of the anomaly but also provide possible fault types and causes, offering users clear decision-making support. This contrasts sharply with the prior art, which only triggers alarms using simple fixed thresholds, lacking the ability to perform correlation analysis on complex faults, resulting in ambiguous alarm information that is difficult to guide actual maintenance work. This method significantly improves the efficiency and effectiveness of motor drive fault diagnosis by providing more specific and accurate anomaly detection results.

[0082] In some embodiments, before comparing the power dimension parameters, pulse dimension parameters, and device dimension parameters with their respective parameter thresholds, the method further includes: performing time alignment, wavelet denoising, and adaptive filtering on the power dimension parameters, pulse dimension parameters, and device dimension parameters in sequence to obtain preprocessed power dimension parameters, pulse dimension parameters, and device dimension parameters.

[0083] Time series alignment aims to eliminate time discrepancies between different data sources or sensors caused by sampling frequency, transmission delay, or system synchronization errors, ensuring that all parameter data are consistent across the time axis and laying the foundation for subsequent accurate analysis. Time series alignment can be achieved in various ways. For example, data points can be mapped to a unified timestamp sequence using interpolation algorithms (such as linear interpolation and spline interpolation), or resampling can be performed based on a common time reference or synchronization signal. Dynamic time warping (DTW) algorithms can also be used to find the optimal matching path between different time series.

[0084] Wavelet denoising is used to effectively remove random noise from data while preserving the effective features and transient information of the signal, thereby improving the signal-to-noise ratio. This can be achieved by performing wavelet decomposition on the signal, selecting appropriate thresholds to process the wavelet coefficients (such as hard thresholding or soft thresholding), and then performing wavelet reconstruction, or by using multi-scale wavelet analysis to identify and remove noise components at different frequency scales.

[0085] Adaptive filtering can dynamically adjust filter parameters based on the statistical characteristics of the signal or noise to better adapt to signal changes, further remove residual noise and interference, and smooth the data. For example, a Kalman filter can be used to estimate the signal state and filter it in real time based on the system model and the statistical characteristics of the measurement noise; or adaptive algorithms such as the Least Mean Square (LMS) algorithm or the Recursive Least Squares (RLS) algorithm can be used to dynamically adjust the filter coefficients to minimize the error; or a moving average filter can be used to dynamically adjust the window size or weights based on the local characteristics of the data.

[0086] This application's solution first aligns the original power supply, pulse, and device parameters in a time sequence, ensuring consistency in the time dimension for data from different sources and sampling frequencies, thus avoiding misjudgments caused by time misalignment. Subsequently, the aligned data undergoes wavelet denoising, effectively filtering out random noise and high-frequency interference while preserving key characteristics and transient changes in the motor driver's operating state, crucial for accurate anomaly identification. Finally, the wavelet-denoised data undergoes adaptive filtering, further smoothing the data and dynamically adjusting filtering parameters according to changes in the motor driver's real-time operating conditions to adapt to different noise characteristics and signal patterns, thereby maximizing data purity and validity. The sequential implementation of these preprocessing steps allows subsequent parameter threshold comparisons to be based on highly accurate, reliable, and time-consistent data, significantly improving the accuracy and robustness of anomaly detection.

[0087] The following is a concrete example to illustrate this. After obtaining the power supply, pulse, and device parameters of the motor driver, these parameters are first time-aligned. For example, if the power supply parameters are sampled at 100Hz, the pulse parameters at 50Hz, and the device parameters at 10Hz, a unified sampling frequency, such as 20Hz, can be set. Then, linear interpolation is used to align all parameter data points to this unified time series. Next, wavelet denoising is performed on the aligned parameter data. Daubechies wavelets (e.g., db4 wavelets) can be used to decompose the data into three levels. Then, soft thresholding is used to process the decomposed wavelet coefficients, and finally, wavelet reconstruction is performed to obtain the denoised data. Finally, adaptive filtering, such as using a Kalman filter, is applied to the denoised data. The Kalman filter can estimate the true values ​​of the parameters in real time based on the operating model of the motor driver and the statistical characteristics of the measurement noise, and dynamically adjust the filter gain. This further removes residual noise and uncertainty while maintaining the dynamic characteristics of the signal, ultimately obtaining the preprocessed power supply, pulse, and device parameters.

[0088] By employing the aforementioned technical solutions, the original power supply, pulse, and device dimension parameters undergo timing alignment, wavelet denoising, and adaptive filtering, effectively eliminating noise, timing deviations, and unnecessary fluctuations in the data. This allows for subsequent comparisons of the preprocessed parameters with corresponding parameter thresholds based on cleaner and more accurate data, significantly improving the accuracy and reliability of anomaly detection. Consequently, it effectively reduces false alarm and false negative rates, enhancing the overall performance of motor driver anomaly detection.

[0089] See Figure 3 In one embodiment, the adaptive change method for parameter thresholds includes, but is not limited to, steps S301 to S303.

[0090] Step S301: Based on the motor driver, determine the corresponding original parameter threshold and original bias value.

[0091] Step S302: Based on real-time status information and fault history information, adjust the original bias value to obtain the parameter bias value.

[0092] Step S303: Calculate the parameter threshold based on the original parameter threshold and parameter bias value.

[0093] Determining the initial parameter thresholds and initial bias values ​​refers to setting a basic anomaly judgment limit and initial correction amount for various monitoring parameters of the motor drive. The initial parameter thresholds are usually predetermined based on the motor drive's design specifications, factory standards, or benchmark data collected under long-term stable operating conditions, using statistical analysis methods (e.g., based on the mean plus or minus the standard deviation of historical data, or setting percentiles). The initial bias value is the initial amount used to fine-tune the initial parameter thresholds; it may reflect the systematic correction required for the parameter thresholds under specific ideal operating conditions. For example, it can be set through expert experience, or calibrated under normal operating conditions during the initial installation and commissioning of the motor drive to ensure the rationality of the initial thresholds.

[0094] Adjusting the original bias value aims to dynamically correct it based on the current actual operating status of the motor drive (real-time status information) and past fault experience (fault history information), making it more suitable for anomaly detection requirements in the current environment. For example, a rule-based adjustment system can be built to increase or decrease the original bias value according to preset rules when real-time status information shows that the motor drive is under specific conditions such as heavy load or high temperature. Alternatively, a machine learning model can be used, taking real-time status information (such as load rate, ambient temperature, speed, etc.) and fault history information (such as under what conditions a specific fault type occurs and the duration of the fault) as input, training the model to output an adjustment factor. This adjustment factor is used to correct the original bias value, thereby obtaining a more adaptive parameter bias value.

[0095] Calculating the parameter threshold based on the original parameter threshold and parameter bias value involves combining the dynamically adjusted parameter bias value with the preset original parameter threshold to obtain the final adaptive parameter threshold used for anomaly detection. Typically, the calculation method can be a simple addition or subtraction operation; that is, the parameter threshold equals the original parameter threshold plus the parameter bias value. For example, if the original parameter threshold is X, and the adjusted parameter bias value is Y, then the final parameter threshold is X ± Y. Ultimately, the two parameter thresholds constitute a parameter interval [XY, X+Y]. This calculation method allows the parameter threshold to maintain a stable baseline while being flexibly adjusted according to actual operating conditions and historical experience.

[0096] This application's solution introduces original parameter thresholds and original bias values ​​as benchmarks. Based on these benchmarks, it dynamically adjusts the original bias values ​​using real-time status information and fault history information to ultimately calculate the parameter thresholds. This hierarchical adjustment mechanism ensures that the parameter thresholds do not simply fluctuate with real-time data, but rather have a stable benchmark that adapts to changes through an adjustable correction amount. Real-time status information provides immediate feedback on the current operating conditions, ensuring that the thresholds can respond to changes in the current operating environment; fault history information provides experience on long-term trends and fault modes, allowing threshold adjustments to learn from past lessons and be more sensitive to known fault modes. Therefore, the parameter thresholds can more accurately reflect the actual health status of the motor drive, maintaining detection stability while improving sensitivity to potential anomalies.

[0097] The following is a concrete example to illustrate this. Suppose we need to detect anomalies in the output current of a motor driver. First, based on the motor driver's design specifications and long-term stable operating data, the initial parameter threshold for the output current is determined to be 10A, and the initial bias value is 0A. During the operation of the motor driver, the actuator continuously acquires real-time status information. For example, it detects that the motor driver is currently under heavy load, with a load rate reaching 90%. Simultaneously, the actuator queries historical fault information and finds that in the past, under heavy load conditions, if the current threshold was set too low, false alarms were likely to occur. Based on this real-time status information and historical fault information, the actuator adjusts the initial bias value. For example, according to a preset rule, when the load rate exceeds 85%, the initial bias value is increased by 0.5A. Therefore, the parameter bias value is adjusted to +0.5A. Finally, based on the initial parameter threshold of 10A and the adjusted parameter bias value ±0.5A, the final parameter thresholds are calculated to be 10A + 0.5A = 10.5A and 10A - 0.5A = 9.5A. At this point, the output current of the motor driver will be compared with 10.5A and 9.5A to determine if there is any abnormality.

[0098] Through the above technical solution, this application can achieve fine-grained adaptive adjustment of parameter thresholds. This adjustment mechanism allows the parameter thresholds to not only respond to the current operating conditions of the motor drive but also to be optimized based on historical fault experience, thereby improving the accuracy and robustness of anomaly detection. Compared with simply adjusting the threshold based on real-time data, this solution effectively avoids false alarms or missed alarms caused by instantaneous fluctuations in the threshold, ensuring the stability and reliability of anomaly detection, and thus improving the efficiency and accuracy of motor drive fault diagnosis.

[0099] In some embodiments, adjusting the original bias value based on real-time status information and fault history information includes: determining the parameter range of the current linear parameter based on real-time status information; adjusting the current linear parameter based on fault history information and parameter range to obtain the updated linear parameter; and performing a linear operation on the original bias value based on the updated linear parameter to obtain the parameter bias value.

[0100] Determining the current linear parameter range aims to provide a reasonable search range or constraint boundary for subsequent linear parameter adjustments. Real-time status information reflects the current operating conditions of the motor drive, such as load, speed, temperature, and voltage. By analyzing this real-time data, a reasonable range of linear parameters under the current operating conditions can be inferred, thus avoiding adjustments within unreasonable ranges and improving the efficiency and accuracy of adjustments. For example, this can be achieved through a pre-established operating condition-parameter range mapping table. When the motor drive is under light load and low speed, the linear parameter range may be narrower and biased towards a specific range; while under heavy load and high speed, the parameter range may be wider or biased towards another range. The executing entity queries this mapping table based on the real-time status information to obtain the corresponding parameter range. Alternatively, it can be dynamically determined through statistical analysis of real-time status information. For example, cluster analysis or regression analysis can be performed on real-time status information over a period of time, and the upper and lower limits of the linear parameters can be dynamically calculated based on the analysis results to form the parameter range.

[0101] Adjusting the current linear parameters aims to optimize them using historical fault data, making them better adaptable to abnormal conditions in actual operation. The parameter range limits the adjustment scope, ensuring the rationality of the adjustment results. By adjusting the linear parameters, the parameter bias values ​​can more accurately reflect the actual operating conditions and potential fault trends of the motor drive. For example, optimization algorithms can be used for adjustment. Within a given parameter range, using historical fault data as a training set, the linear parameters are iteratively adjusted using optimization methods such as genetic algorithms, particle swarm optimization, or gradient descent, by minimizing objective functions such as false alarm rate and false negative rate, until the optimal or updated linear parameters that meet preset conditions are reached. Alternatively, machine learning models can also be used for adjustment. For example, a predictive model based on fault history information can be constructed. This model takes historical fault characteristics and corresponding optimal linear parameters as input and output. When new fault history information appears, the model predicts the updated linear parameters within the parameter range.

[0102] Linear operations are performed on the original bias values ​​to apply optimized linear parameters to them, resulting in the final parameter bias values ​​used for threshold calculation. Linear operations are a simple and effective adjustment method that directly reflects the effect of linear parameter adjustment on the bias values, thus influencing the adaptability of the parameter threshold. For example, a linear operation can be represented as a multiplication operation, where the parameter bias value equals the original bias value multiplied by the updated linear parameter. This method allows for proportional scaling of the original bias value. Alternatively, a linear operation can be represented as a weighted sum, where the parameter bias value equals the original bias value plus the updated linear parameter multiplied by a weighting factor. This method allows for incremental adjustment based on the original bias value.

[0103] This application's solution achieves more refined and intelligent adjustment of the original bias value by dynamically adjusting the linear parameter. First, the executing entity dynamically determines a reasonable parameter range based on the current real-time status information of the motor driver. This range limits the possible values ​​of the linear parameter, ensuring the effectiveness and rationality of subsequent adjustments. Next, combining historical fault information and the determined parameter range, the current linear parameter is finely adjusted. This adjustment process utilizes the patterns and trends inherent in historical fault data, enabling the linear parameter to better reflect the characteristics of the motor driver under different fault modes, thus obtaining an optimized and updated linear parameter. Finally, this updated linear parameter is applied to the original bias value, and a final parameter bias value is generated through linear operations. This parameter bias value is then combined with the original parameter threshold to jointly calculate an adaptive parameter threshold for anomaly detection. In this way, the adaptive change of the parameter threshold is no longer a simple coarse adjustment based on real-time status and fault history, but rather, by introducing and optimizing the linear parameter, a more refined and intelligent adjustment of the original bias value is achieved, thereby improving the accuracy and robustness of anomaly detection.

[0104] The following is a concrete example to illustrate this. When adjusting the initial bias value, we can first analyze the real-time status information of the motor driver, such as the current operating current, voltage, and temperature, and combine this with a preset operating condition model to determine a parameter range for the linear parameter. For example, when the motor is running under rated load, the value range of the linear parameter may be limited to 0.8 to 1.2. Subsequently, the execution unit will search the historical fault database to obtain historical fault data similar to the current operating condition. This data may include abnormal events such as overcurrent, overvoltage, and overheating, and their corresponding changes in motor parameters. Using this historical fault data, combined with the previously determined parameter range, an optimization model can be constructed. For example, with the objective function of minimizing the false alarm rate and false alarm rate of historical faults, a genetic algorithm can be used to search for the optimal linear parameter within the parameter range of 0.8 to 1.2. Suppose that after optimization, the updated linear parameter is 1.05. Finally, this updated linear parameter 1.05 is multiplied by the initial bias value. For example, if the initial bias value is 5, the parameter bias value is calculated as 5 × 1.05 = 5.25. This parameter bias value of 5.25 will be used for subsequent parameter threshold calculations, so that the threshold can more accurately reflect the actual operating status and potential failure risks of the current motor drive.

[0105] Through the above technical solution, when adjusting the original bias value, real-time status information and fault history information can be fully utilized. By introducing linear parameters and dynamically adjusting them, the determination of parameter bias values ​​becomes more accurate and intelligent. This refined adjustment mechanism avoids the inaccuracies that may result from simple adjustments, thereby improving the accuracy of adaptive changes in parameter thresholds. This, in turn, enhances the sensitivity and reliability of abnormal detection in the motor driver, effectively reducing the risk of false alarms and missed alarms.

[0106] In some embodiments, the current linear parameters are adjusted based on fault history information and parameter range, including: when the fault history information meets the preset parameter update conditions, establishing an objective function with linear parameters as variables, optimizing the objective function within the parameter range with the goal of maximizing the anomaly detection rate and minimizing the false alarm rate, and obtaining the updated linear parameters after optimization.

[0107] The preset parameter update conditions may include, but are not limited to: detecting a certain number of abnormal events, the system runtime reaching a preset threshold, model performance indicators (such as accuracy and recall) dropping to an alert level, or manual triggering of an update command. These conditions ensure the timeliness and necessity of parameter updates, avoid unnecessary computational overhead, and guarantee the effectiveness of the updates.

[0108] The objective function is a mathematical expression used to quantify the relationship between the optimization objective and the linear parameters. It can be a weighted sum or ratio that comprehensively considers the anomaly detection rate and the false alarm rate, and its form can be designed according to specific optimization requirements and model characteristics. This function takes the linear parameters as input and outputs an evaluation value that reflects the anomaly detection performance under the current linear parameters.

[0109] The proposed solution introduces an optimization mechanism for linear parameters when adjusting the original bias value. Specifically, when historical fault information meets preset parameter update conditions, the execution entity does not simply adjust the linear parameters based on real-time status and historical fault information. Instead, it constructs an objective function with the linear parameters as independent variables. This objective function aims to quantify the performance of anomaly detection under the current linear parameter configuration, with a core focus on comprehensively considering the anomaly detection rate and false alarm rate. By iteratively solving the objective function within a predefined parameter range using an optimization algorithm, the execution entity can intelligently explore and identify the optimal combination of linear parameters. This optimization process ensures that the updated linear parameters not only reflect the current system state and historical fault modes, but more importantly, they enable the execution entity to maximize the detection of true anomalies while effectively suppressing false alarms. Ultimately, these optimized linear parameters are used to accurately calculate parameter bias values, thereby making the adaptive adjustment of parameter thresholds more precise and efficient, significantly improving the performance and reliability of the entire anomaly detection method.

[0110] The following is a concrete example to illustrate this. When the executing entity detects that the false alarm rate exceeds 5% or the number of undetected anomalies exceeds 3 within the past 24 hours, a parameter update condition can be triggered. At this point, an objective function F(L) = w1 * DetectionRate - w2 * FalsePositiveRate can be established, where L represents a linear parameter, and w1 and w2 are preset weight coefficients used to balance the importance of anomaly detection rate and false alarm rate. For example, w1 can be set to 0.7, and w2 can be set to 0.3, indicating a greater emphasis on detection rate. The parameter range of this linear parameter L can be determined by real-time status information (such as motor load, operating mode) and historical experience data, for example, [0.5, 1.5]. The executing entity can use a genetic algorithm for optimization. First, an initial population of linear parameters L is randomly generated. Then, new populations are generated iteratively through crossover, mutation, and other operations, and selection is made based on the fitness value of the objective function F(L) until the preset number of iterations is reached or the objective function converges. After optimization, the linear parameter L with the highest fitness is used as the updated linear parameter. For example, after optimization, the linear parameter L may be updated from the initial value of 1.0 to 1.2, which means that under the current operating conditions, slightly increasing the linear parameter can better balance the detection rate and the false alarm rate.

[0111] Through the above technical solution, this application, when adjusting the original bias value, can establish and optimize an objective function with linear parameters as variables based on fault history information and parameter range, thereby obtaining a better updated linear parameter. This optimization process makes the adaptive adjustment of parameter thresholds no longer a simple linear mapping or empirical adjustment, but based on a quantitative evaluation and optimization of anomaly detection performance. This significantly improves the accuracy and robustness of anomaly detection, effectively solving the problem of low anomaly detection rate or excessive false alarm rate that may occur with traditional methods in complex and variable motor operating environments. By maximizing the anomaly detection rate and minimizing the false alarm rate, this application can ensure that anomalies of the motor drive can be detected in a timely and accurate manner, while avoiding resource waste and unnecessary downtime caused by false alarms, thereby improving the stability and reliability of the motor drive operation.

[0112] See Figure 4 In one embodiment, the method for performing correlation analysis on abnormal power dimension parameters, abnormal pulse parameters and / or abnormal device parameters includes, but is not limited to, steps S401 to S403.

[0113] Step S401: Combine the abnormal power supply dimension parameters, abnormal pulse parameters, and / or abnormal device parameters to obtain a co-occurring abnormal parameter group.

[0114] Step S402: Based on the preset mapping relationship between co-occurring anomaly parameters and anomaly items, determine several anomaly items corresponding to the co-occurring anomaly parameter group.

[0115] Step S403: Perform anomaly tracing and reasoning on each abnormal item to determine the root cause abnormal item among each abnormal item and obtain the correlation analysis results.

[0116] Combining abnormal power supply parameters, abnormal pulse parameters, and / or abnormal device parameters refers to the logical merging or aggregating of various parameters that occur simultaneously and are determined to be abnormal. This combination can be a simple logical AND operation, where multiple specific parameters exceeding a threshold are considered a co-occurring abnormal parameter group; or it can be based on statistical methods, such as cluster analysis, to automatically categorize parameters that frequently co-occur. In this way, scattered abnormal signals can be integrated into abnormal patterns with higher semantic meaning, providing more focused information for subsequent diagnosis.

[0117] The mapping relationship between co-occurring anomaly parameters and anomaly items refers to a pre-established knowledge base or rule set, which can be manually defined based on the experience and knowledge of domain experts. For example, when "power supply voltage too low" and "output current too high" occur simultaneously, it may be mapped to anomalies such as "power module failure" or "motor overload". Alternatively, this mapping relationship can be automatically constructed by mining and learning from historical fault data, such as using machine learning methods like decision trees and association rule learning.

[0118] Abnormal items are fault descriptions that are more diagnostically significant than abnormal raw parameters, such as "overvoltage", "undercurrent", "sensor failure", etc.

[0119] Various techniques can be employed to perform anomaly tracing and reasoning on various anomalies. These include causal graph models (such as Bayesian networks), fault tree analysis, expert system rule chain reasoning, or deep learning models based on historical failure patterns. The core principle is to identify, through logical reasoning or probabilistic inference, the "root cause" that is at the very top of the causal chain and directly leads to other anomalies from multiple related anomalies. For example, if both "motor overload" and "driver overheating" are identified as anomalies, and the reasoning model indicates that "motor overload" is a common cause of "driver overheating," then "motor overload" will be identified as the root cause anomaly.

[0120] This application's solution combines independent anomaly parameters to form a more diagnostically meaningful set of co-occurring anomaly parameters, and further maps these parameters to specific anomaly items. Based on this, through anomaly tracing and reasoning, the true root cause anomaly item is identified from multiple possible anomaly items. This progressive analysis method allows anomaly detection to go beyond surface phenomena and delve into the essential causes of faults. This not only improves the accuracy and efficiency of fault diagnosis but also provides clear guidance for subsequent troubleshooting and prevention, significantly enhancing the intelligence level of motor drive anomaly detection.

[0121] In one specific implementation, during the operation of the motor driver, the executing entity first acquires its power supply dimension parameters, pulse dimension parameters, and device dimension parameters. Assuming that after comparing with parameter thresholds, it is detected that three parameters—"power supply voltage too low," "output current too high," and "abnormal motor speed"—simultaneously exceed the normal range, the executing entity will combine these three abnormal parameters into a co-occurring abnormal parameter group, for example, denoted as `{power supply voltage too low, output current too high, motor speed abnormal}`. Next, the executing entity will query a preset mapping relationship between co-occurring abnormal parameters and abnormal items. This mapping relationship may contain the rule: "If 'power supply voltage too low' and 'output current too high' and 'motor speed abnormal', then there may be abnormal items such as 'power module failure', 'motor overload', or 'driver internal short circuit'." Based on this, the executing entity determines 'power module failure', 'motor overload', and 'driver internal short circuit' as several current abnormal items. Subsequently, the executing entity performs anomaly tracing and reasoning on these abnormal items. For example, using a Bayesian network-based inference model, the model might contain the knowledge that 'motor overload' is a common cause of 'excessive output current' and 'low power supply voltage', while 'internal short circuit in the driver' can also cause similar phenomena, but its probability distribution or association pattern with other parameters may differ. By calculating the posterior probability of each anomalous item as a root cause, the executing agent might ultimately infer that 'motor overload' is the most likely root cause anomalous item and output it as the result of the association analysis.

[0122] Through the above technical solution, this application can integrate scattered abnormal signals into meaningful fault modes and further analyze the causal relationships between these fault modes, thereby accurately locating the root cause of motor drive abnormalities. This allows maintenance personnel to directly address the root cause, avoiding repeated troubleshooting of superficial problems, significantly shortening fault diagnosis and repair time, reducing maintenance costs, and effectively preventing more serious equipment damage or production interruptions caused by failure to promptly identify the root cause.

[0123] In some embodiments, anomaly tracing reasoning is performed on each anomalous item, including: constructing all possible combinations of anomalous root causes using each anomalous item; calculating the joint likelihood of each combination of anomalous root causes; evaluating the comprehensive confidence of each combination of anomalous root causes based on the joint likelihood and the prior probability of the root anomalous item in the combination of anomalous root causes; and outputting the combination of anomalous root causes with the highest comprehensive confidence as the result of the association analysis.

[0124] This approach utilizes each anomaly item to construct all possible combinations of root causes, aiming to systematically generate all possible combinations of root causes that could lead to these anomalies from the identified anomalies. This can be achieved through predefined fault tree models, cause-effect graphs, or expert knowledge bases. For example, if two anomalies, "overcurrent" and "overvoltage," are detected, possible root cause combinations could include single root causes such as "power module failure," "load short circuit," or "control board malfunction," or combinations of multiple root causes such as "power module failure and load short circuit."

[0125] Calculating the joint likelihood of various combinations of anomalous root causes aims to quantify the probability of a specific combination of anomalous root causes occurring given observed anomalous items. Joint likelihood can be based on statistical analysis of historical failure data, such as calculating the conditional probability of observing the current set of anomalous items when a specific combination of root causes occurs. Alternatively, Bayesian networks and other probabilistic graphical models can be used to derive the joint likelihood through the conditional probability distribution among nodes.

[0126] Based on joint likelihood and prior probabilities of root cause anomalies in combinations of anomalies, this study evaluates the overall confidence of each combination of root causes. The aim is to obtain a more comprehensive confidence assessment by combining likelihood information from current observation data with prior knowledge about the combinations of root causes. Prior probabilities can be derived from historical failure frequencies, expert experience, or equipment design characteristics. By fusing joint likelihood with prior probabilities, for example using Bayes' theorem, the posterior probability of each combination of root causes can be calculated, thus obtaining its overall confidence.

[0127] This application's solution achieves accurate identification of motor driver anomalies by systematically constructing all possible combinations of anomalies and incorporating data-driven likelihood calculations and prior knowledge. Specifically, after performing correlation analysis on abnormal power supply parameters, abnormal pulse parameters, and / or abnormal equipment parameters, and identifying several anomalies, this solution first uses these anomalies to exhaustively or intelligently generate all possible combinations of root causes that could lead to these anomalies. Subsequently, for each generated root cause combination, its joint likelihood is calculated under the condition that the current anomaly occurs, reflecting the consistency between the combination and the observed data. To avoid biases that may result from relying solely on current data, this solution further introduces the prior probability of each anomaly root cause combination, which reflects the frequency of occurrence of different root cause combinations in history or experience. By fusing the joint likelihood with the prior probability, for example using Bayesian inference, the comprehensive confidence of each anomaly root cause combination can be evaluated, which more comprehensively reflects the probability that the combination is the true root cause. Ultimately, the implementing entity selected the combination of anomaly root causes with the highest overall confidence level as the correlation analysis result, thereby effectively solving the problem of difficulty in accurately identifying the true root cause when there are multiple potential anomaly root causes, and significantly improving the accuracy and reliability of anomaly tracing.

[0128] As a specific implementation, when the motor driver detects two anomalies—"abnormal power supply voltage" and "fluctuation in output current"—the executing entity can first construct all possible combinations of anomaly root causes based on a pre-set fault knowledge base or cause-effect graph. For example, these combinations might include: {power module failure}, {load characteristic change}, {control algorithm parameter drift}, {sensor failure}, and {power module failure, load characteristic change}, etc. Next, the executing entity can use historical fault data to calculate the joint likelihood of each anomaly root cause combination through statistical analysis or a trained probabilistic model (such as a Naive Bayes classifier). For example, if historical data shows that when "power module failure" occurs, the probability of both "abnormal power supply voltage" and "fluctuation in output current" occurring simultaneously is high, then the joint likelihood of that combination will be correspondingly high. Simultaneously, the executing entity will obtain the prior probability of each anomaly root cause combination; for example, based on equipment operating experience, the prior probability of "power module failure" might be higher than that of "sensor failure". Then, the executing entity combines the joint likelihood with the prior probability, for example, by calculating the posterior probability P(root cause combination | anomalous item), to evaluate the overall confidence of each combination. Finally, the executing entity outputs the anomaly root cause combination with the highest overall confidence. For example, if the calculation results show that {power module failure} has the highest overall confidence, it is used as the correlation analysis result for this anomaly, guiding subsequent fault diagnosis and repair.

[0129] Through the above technical solution, this application can systematically trace the causes of abnormal items in motor drives. By constructing all possible combinations of root causes and combining joint likelihood with the prior probability of the root cause abnormal item to evaluate the overall confidence level, this solution can effectively distinguish the probability of different combinations of root causes, thereby accurately identifying the most likely root cause abnormal item when multiple potential fault causes exist. This significantly improves the accuracy and reliability of abnormal source tracing, avoids misjudgment and omission, provides more precise guidance for fault diagnosis and maintenance of motor drives, and thus improves the operational stability and maintenance efficiency of the equipment.

[0130] See Figure 5 This application also provides a motor driver anomaly detection device, which can implement the above-described motor driver anomaly detection method. The device includes:

[0131] The first module 501 is used to obtain the power supply dimension parameters, pulse dimension parameters and device dimension parameters of the motor driver;

[0132] The second module 502 is used to compare the power supply dimension parameters, pulse dimension parameters and device dimension parameters with the corresponding parameter thresholds, so as to determine the corresponding abnormal power supply dimension parameters, abnormal pulse parameters and / or abnormal device parameters based on the comparison results; the parameter thresholds are adaptively changed based on the real-time status information and fault history information of the motor driver.

[0133] The third module 503 is used to perform correlation analysis on abnormal power supply dimension parameters, abnormal pulse parameters and / or abnormal equipment parameters based on preset parameter correlation rules, and obtain the corresponding correlation analysis results;

[0134] The fourth module 504 is used to generate anomaly detection results for the motor driver based on the correlation analysis results and the abnormal power supply dimension parameters, abnormal pulse parameters and / or abnormal device parameters.

[0135] The specific implementation of the motor driver anomaly detection device is basically the same as the specific implementation of the motor driver anomaly detection method described above, and will not be repeated here.

[0136] Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application.

[0137] The following reference Figure 6 To describe an electronic device 600 according to such an embodiment of the present disclosure. Figure 6 The electronic device 600 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0138] like Figure 6As shown, the electronic device 600 is presented in the form of a general-purpose computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different system components (including storage unit 620 and processing unit 610), a display unit 640, etc.

[0139] The storage unit stores program code, which can be executed by the processing unit 610, causing the processing unit 610 to perform the steps described in the above-described motor driver anomaly detection method section of this specification according to various exemplary embodiments of this disclosure.

[0140] Storage unit 620 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 6201 and / or cache memory 6202, and may further include a read-only memory (ROM) 6203.

[0141] Storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0142] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0143] Electronic device 600 can also communicate with one or more external devices 600' (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 600, and / or with any device that enables electronic device 600 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 650. Furthermore, electronic device 600 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of electronic device 600 via bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0144] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0145] The motor driver anomaly detection method, apparatus, device, and medium provided in this application utilize adaptive parameter thresholds to determine abnormal parameters among the multi-dimensional parameters of the motor driver. Then, correlation analysis is performed through parameter association rules to identify the inherent relationships between these abnormal parameters, thereby locating the root cause of the fault and serving as the final anomaly detection result. Therefore, by introducing adaptive parameter thresholds, threshold detection can better adapt to the complex and ever-changing operating environment of the motor driver, improving the accuracy and robustness of the detection. Simultaneously, the introduction of correlation analysis based on preset parameter association rules after threshold detection allows for the identification of the inherent relationships between multiple abnormal parameters when they are detected, accurately locating the root cause of the fault and providing users with clear decision-making basis. This improves the performance of in-depth parameter processing and correlation analysis of complex faults.

[0146] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, or network device, etc.) to execute the methods described above according to the embodiments of this disclosure.

[0147] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0148] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0149] Those skilled in the art will understand that the above modules can be distributed in the device as described in the embodiments, or they can be modified accordingly and placed in one or more devices that are unique to this embodiment. The modules in the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0150] Exemplary embodiments of this disclosure have been specifically shown and described above. It should be understood that this disclosure is not limited to the detailed structures, arrangements, or implementations described herein; rather, this disclosure is intended to cover various modifications and equivalent arrangements contained within the spirit and scope of the appended claims.

Claims

1. A method for detecting abnormalities in a motor driver, characterized in that, include: Obtain the power supply dimension parameters, pulse dimension parameters, and device dimension parameters of the motor driver; The power dimension parameters, pulse dimension parameters, and device dimension parameters are compared with corresponding parameter thresholds to determine the corresponding abnormal power dimension parameters, abnormal pulse parameters, and / or abnormal device parameters based on the comparison results; the parameter thresholds are adaptively changed based on the real-time status information and fault history information of the motor driver. Based on preset parameter association rules, association analysis is performed on the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters to obtain the corresponding association analysis results. Based on the correlation analysis results, the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters, anomaly detection results for the motor driver are generated. The correlation analysis based on preset parameter association rules for the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters includes: The abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters are combined to obtain a co-occurring abnormal parameter group; Based on the preset mapping relationship between co-occurrence anomaly parameters and anomaly items, a number of anomaly items corresponding to the co-occurrence anomaly parameter group are determined. Anomaly tracing reasoning is performed on each of the aforementioned anomalies to determine the root cause anomaly among them, thereby obtaining the correlation analysis results.

2. The motor driver anomaly detection method according to claim 1, characterized in that, Before comparing the power dimension parameter, the pulse dimension parameter, and the device dimension parameter with their respective parameter thresholds, the method further includes: The power dimension parameters, pulse dimension parameters, and device dimension parameters are sequentially subjected to time alignment, wavelet denoising, and adaptive filtering to obtain preprocessed power dimension parameters, pulse dimension parameters, and device dimension parameters.

3. The method for detecting abnormalities in a motor driver according to claim 1, characterized in that, The adaptive change method for the parameter threshold includes: Based on the motor driver, determine the corresponding original parameter thresholds and original bias values; Based on the real-time status information and the fault history information, the original bias value is adjusted to obtain the parameter bias value; The parameter threshold is calculated based on the original parameter threshold and the parameter bias value.

4. The motor driver anomaly detection method according to claim 3, characterized in that, The adjustment of the original bias value based on the real-time status information and the fault history information includes: Based on the real-time status information, the parameter range of the current linear parameter is determined; Based on the fault history information and the parameter range, the current linear parameters are adjusted to obtain updated linear parameters; Based on the updated linear parameters, a linear operation is performed on the original bias value to obtain the parameter bias value.

5. The motor driver anomaly detection method according to claim 4, characterized in that, The adjustment of the current linear parameter based on the fault history information and the parameter range includes: When the fault history information meets the preset parameter update conditions, an objective function with the linear parameter as the variable is established. Within the parameter range, the objective function is optimized with the goal of maximizing the anomaly detection rate and minimizing the false alarm rate. After optimization, the updated linear parameter is obtained.

6. The method for detecting abnormalities in a motor driver according to claim 1, characterized in that, The anomaly tracing and reasoning for each of the aforementioned abnormal items includes: Construct all possible combinations of anomaly root causes using each of the aforementioned anomaly items; Calculate the joint likelihood of each of the aforementioned combinations of anomalies; The overall confidence level of each of the anomaly root source combinations is evaluated based on the joint likelihood and the prior probability of the root source anomaly items in the anomaly root source combinations. The combination of anomalies with the highest overall confidence is output as the result of the association analysis.

7. A motor driver malfunction detection device, characterized in that, include: The first module is used to obtain the power supply dimension parameters, pulse dimension parameters, and device dimension parameters of the motor driver; The second module is used to compare the power dimension parameter, the pulse dimension parameter and the device dimension parameter with the corresponding parameter thresholds, so as to determine the corresponding abnormal power dimension parameter, abnormal pulse parameter and / or abnormal device parameter based on the comparison results. The parameter thresholds are adaptively changed based on the real-time status information and fault history information of the motor driver; The third module is used to perform correlation analysis on the abnormal power supply dimension parameters, the abnormal pulse parameters and / or the abnormal device parameters based on preset parameter correlation rules, and to obtain the corresponding correlation analysis results. The fourth module is used to generate anomaly detection results for the motor driver based on the correlation analysis results, the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters. The correlation analysis based on preset parameter association rules for the abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters includes: The abnormal power supply dimension parameters, the abnormal pulse parameters, and / or the abnormal device parameters are combined to obtain a co-occurring abnormal parameter group; Based on the preset mapping relationship between co-occurrence anomaly parameters and anomaly items, a number of anomaly items corresponding to the co-occurrence anomaly parameter group are determined. Anomaly tracing reasoning is performed on each of the aforementioned anomalies to determine the root cause anomaly among them, thereby obtaining the correlation analysis results.

8. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the motor driver abnormality detection method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the motor driver anomaly detection method according to any one of claims 1 to 6.

Images