Motor data anomaly detection method based on Internet of Things

Abstract

The present invention relates to the technical field of motor anomaly detection, and in particular to a motor data anomaly detection method based on the Internet of Things. The method comprises the following steps: acquiring motor operation data, evaluating the motor operation demand overload and load response status, detecting the decline trend of bearing kinetic energy chain coupling and motor structure connection coordination, and then determining the dynamic magnetic field imbalance status during motor operation; based on the dynamic magnetic field imbalance, estimating the motor electromagnetic interference risk and frequency oscillation growth trend, monitoring error accumulation, and detecting motor abnormal status; combining the abnormality detection results, optimizing the configuration of motor operating parameters; the present invention achieves more stable and efficient motor operation by detecting motor data anomalies.

Description

Technical Field

[0001] The present invention relates to the technical field of motor anomaly detection, and in particular to a motor data anomaly detection method based on the Internet of Things. Background Art

[0002] As core components driving mechanical equipment, motors play a critical role in various production systems. The operating status of motors directly impacts equipment performance and production efficiency. Therefore, real-time monitoring and anomaly detection of motor operating data have become crucial for ensuring stable operation of industrial equipment. Traditional motor anomaly detection methods rely on periodic maintenance and manual inspections, resulting in long detection cycles and difficulty in achieving dynamic, real-time perception of motor operating status. These methods suffer from delayed response times and high false positive rates, making them unable to meet the demands of modern industry for efficient and intelligent equipment management. The development of the Internet of Things (IoT) technology enables real-time collection of multi-dimensional motor operating data through sensor networks, enabling online monitoring and remote management of motor status. IoT technology, integrating big data analysis, cloud computing, and edge computing, provides more comprehensive and accurate data support for motor anomaly detection, enabling rapid identification and location of motor anomalies, significantly improving detection efficiency and accuracy. However, traditional motor data anomaly detection suffers from inaccurate detection of dynamic magnetic field imbalances during motor operation. Summary of the Invention

[0003] Based on this, it is necessary to provide a motor data anomaly detection method based on the Internet of Things to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a motor data anomaly detection method based on the Internet of Things includes the following steps:

[0005] Step S1: Acquire motor operation data; perform motor operation demand overload assessment based on the motor operation data to obtain motor operation demand overload data; and detect motor operation load response status based on the motor operation data and the motor operation demand overload data;

[0006] Step S2: determining the bearing kinetic energy chain coupling decay state according to the motor operation load response state; determining the motor structure connection coordination degradation trend based on the motor operation load response state; determining the motor operation dynamic magnetic field imbalance state based on the bearing kinetic energy chain coupling decay state and the motor structure connection coordination degradation trend;

[0007] Step S3: estimating the growth trend of the electromagnetic interference risk of the motor according to the dynamic magnetic field imbalance of the motor; determining the growth status of the motor operating frequency oscillation according to the growth trend of the electromagnetic interference risk of the motor;

[0008] Step S4: detecting the accumulated motor operation error based on the motor operation frequency oscillation growth condition; performing motor abnormality detection based on the accumulated motor operation error to obtain motor operation abnormality data; performing motor configuration optimization processing on the motor operation data according to the motor operation abnormality data to obtain motor configuration optimization processing data.

[0009] The present invention systematically analyzes motor operating data to accurately assess the motor's load response, structural coordination, and kinetic energy chain status. It effectively identifies and quantifies the dynamic magnetic field imbalance and electromagnetic interference risks during motor operation, provides early warning of frequency oscillation trends, and accurately monitors error accumulation and abnormal operating conditions. This method not only improves the ability to perceive changes in the motor's internal structure and load, but also enhances the ability to control electromagnetic interference and frequency anomalies, significantly improving the stability and reliability of the motor's operation. Based on the anomaly detection results, targeted configuration optimization is implemented to optimize the motor's overall operating performance, extend the equipment's service life, reduce failure rates, and enhance system safety. The overall solution implements closed-loop management from data acquisition and status assessment to anomaly warning and optimization processing, improving the efficiency of intelligent motor diagnosis and operation and maintenance, and ensuring efficient and stable operation of the motor under complex operating conditions. Therefore, the present invention optimizes traditional motor data anomaly detection, resolving the problems of traditional motor data anomaly detection, such as inaccurate detection of dynamic magnetic field imbalance conditions during motor operation, and improving the accuracy of dynamic magnetic field imbalance detection during motor operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 The figure is a flowchart of a method for detecting abnormality in motor data based on the Internet of Things;

[0011] Figure 2 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0012] Figure 3 for Figure 1 Detailed implementation steps of step S4 in FIG.

[0013] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0014] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0015] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0016] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0017] To achieve this, please refer to Figures 1 to 3 , a motor data anomaly detection method based on the Internet of Things, comprising the following steps:

[0018] Step S1: Acquire motor operation data; perform motor operation demand overload assessment based on the motor operation data to obtain motor operation demand overload data; and detect motor operation load response status based on the motor operation data and the motor operation demand overload data;

[0019] In an embodiment of the present invention, the real-time operating data of the motor is collected by a multi-dimensional sensor network deployed at key locations of the motor. The sensors include current sensors, voltage sensors, speed sensors, temperature sensors, and vibration sensors. A high-speed data acquisition unit is used to collect the signals of each sensor at a sampling rate of not less than 1kHz to ensure that the dynamic changes of the motor are captured. The collected data is transmitted to the edge computing node through the Internet of Things communication module. The edge computing node uses a digital signal processor to pre-process the original signal, including denoising filtering (using a bandpass filter to filter out 50Hz and its harmonics of the power grid noise), signal normalization, and feature extraction. For the overload assessment of the motor's operating requirements, the actual output power of the motor is calculated based on the collected current, voltage, and speed data. , through the formula:

[0020] ;

[0021] in is the line voltage, is the line current, is the power factor, The coefficient that represents the relationship between the line voltage and phase voltage of a three-phase system, combined with the resulting power and the rated power of the motor Compare and calculate the load rate L= / . When the load rate exceeds the preset threshold (such as 1.0), it is determined to be an overload state. At the same time, the temperature and vibration data are combined to analyze whether the motor is in a continuous overload condition. The motor operation demand overload data is formed by recording the time series of the load rate, including parameters such as load rate peak, duration and overload frequency. Based on the motor operation data and demand overload data, the motor operation load response status is calculated. Specifically, by comparing the correlation between the load rate change trend and the speed fluctuation and vibration acceleration response, the cross-correlation function is used to analyze the instantaneous change of the motor load response, and the load response curve and response coefficient are obtained to reflect the dynamic adaptability and response speed of the motor to load changes, and output a set of quantitative indicators of the motor operation load response status.

[0022] Step S2: determining the bearing kinetic energy chain coupling decay state according to the motor operation load response state; determining the motor structure connection coordination degradation trend based on the motor operation load response state; determining the motor operation dynamic magnetic field imbalance state based on the bearing kinetic energy chain coupling decay state and the motor structure connection coordination degradation trend;

[0023] In an embodiment of the present invention, the coupling degradation state of the bearing kinetic energy chain is determined by analyzing the spectral characteristics of the vibration signal during the motor load response. Specifically, the method uses a high-speed Fourier transform (FFT) to process vibration acceleration data, extracting bearing characteristic frequencies, including the fundamental rotation frequency, outer race fault frequency, inner race fault frequency, and rolling element fault frequency. The energy amplitudes of these characteristic frequencies are compared with those in a historically healthy state to calculate the energy degradation ratio. A significant decrease in energy amplitude and a shift in the spectral peak, combined with a weakening load response trend, indicates coupling degradation of the bearing kinetic energy chain. The trend of declining motor structural connection coordination is determined by analyzing data from strain sensors installed at multiple points. By calculating the strain amplitude change rate at the structural connection and combining it with vibration phase difference analysis of the motor load response, it is determined whether the coordination of the connection components is abnormal. A coherence function is used to calculate the phase synchronization between different sensors. If the coherence decreases beyond a set threshold, it indicates a decline in connection coordination. The data on the decaying state of the bearing kinetic energy chain coupling and the decreasing trend of the structural connection coordination are integrated, combined with the magnetic flux density distribution collected by the magnetic field sensor inside the motor. Through the magnetic field reconstruction algorithm, the symmetry and uniformity of the magnetic field inside the motor are analyzed. The dynamic magnetic field imbalance is quantified using the magnetic field unevenness index, and the numerical parameters and trend maps of the dynamic magnetic field imbalance during motor operation are output.

[0024] Step S3: estimating the growth trend of the electromagnetic interference risk of the motor according to the dynamic magnetic field imbalance of the motor; determining the growth status of the motor operating frequency oscillation according to the growth trend of the electromagnetic interference risk of the motor;

[0025] In this embodiment of the present invention, electromagnetic compatibility analysis methods are used to assess electromagnetic interference (EMI) risk based on numerical parameters of dynamic magnetic field imbalance. The finite element method (FEM) is used to simulate the impact of changes in the motor's magnetic field on the surrounding electromagnetic environment. Combined with field-collected electromagnetic radiation intensity data, the trend of the EMI power spectral density (PSD) is calculated. The EMI risk growth trend is determined by curve fitting the rate of change of the PSD over time, quantifying the risk level of increasing EMI. This EMI risk growth trend is further mapped to determine the frequency oscillation condition of the motor. High-resolution time-frequency analysis techniques, such as wavelet transform, are used to detect frequency oscillations in the motor's current and speed signals, extracting the oscillation amplitude and frequency change rate. Based on the increasing EMI risk, the increase in oscillation amplitude and frequency is considered a direct indicator of frequency oscillation growth. A dynamic monitoring curve for frequency oscillation growth is generated, and parameters such as the oscillation growth rate and abnormal frequency range are output for subsequent error accumulation detection.

[0026] Step S4: detecting the accumulated motor operation error based on the motor operation frequency oscillation growth condition; performing motor abnormality detection based on the accumulated motor operation error to obtain motor operation abnormality data; performing motor configuration optimization processing on the motor operation data according to the motor operation abnormality data to obtain motor configuration optimization processing data.

[0027] In this embodiment of the present invention, the accumulated motor operating error is calculated based on the frequency oscillation growth. An integral calculation method is used to calculate the cumulative effect of the frequency oscillation amplitude over time as an error accumulation index. The error trend is determined by combining the slope and shape of the error integral curve. The accumulated error value is compared with the historical healthy state error threshold to determine the degree of abnormality. Motor abnormality detection, based on the error accumulation index, employs multidimensional data fusion technology to combine the accumulated frequency oscillation error data with vibration characteristics, temperature anomalies, and electromagnetic interference risk data. A decision tree classification algorithm is used to identify abnormal conditions and output a motor operation abnormality dataset, including the abnormality category, abnormality time point, and abnormality index value. Based on the motor operation abnormality data, motor configuration optimization processing is performed on the motor operation data. Specifically, this process automatically adjusts the abnormal operating conditions by adjusting the motor controller's operating parameters, such as current limit, voltage regulation strategy, and load distribution scheme. The optimization processing module feeds the adjusted parameters back to the motor control system and simultaneously generates motor configuration optimization processing data, including optimized parameter values, adjustment timestamps, and system response status, to ensure improved motor stability during subsequent operation.

[0028] Preferably, step S1 includes the following steps:

[0029] Step S11: obtaining motor operation data based on the Internet of Things;

[0030] In this embodiment of the present invention, a multi-dimensional data acquisition system for the motor's operating status is constructed by deploying multiple types of sensors, including current sensors, voltage sensors, speed sensors, temperature sensors, and acceleration sensors, at key locations on the motor. The analog signals collected by each sensor are converted to digital signals via an analog-to-digital converter (ADC), with a sampling frequency uniformly set to at least 1kHz to ensure that the motor's dynamic characteristics are captured. The acquisition equipment is equipped with an IoT communication module, supporting various wireless communication protocols such as Wi-Fi, NB-IoT, and LoRa, and uploading the collected data in real time to an edge computing node. The edge node uses a real-time database to store the data in a time-series manner, ensuring data integrity and continuity. During the acquisition process, CRC checksums and timing synchronization mechanisms are implemented to ensure the accuracy and real-time nature of data transmission. This generates a set of raw motor operating data, including multiple indicators such as current, voltage, speed, temperature, and vibration, providing basic data support for subsequent analysis.

[0031] Step S12: performing motor operation demand overload evaluation based on the motor operation data to obtain motor operation demand overload data;

[0032] In the embodiment of the present invention, based on the motor operation data obtained in step S11, the motor output power is calculated in real time, and the power calculation formula is:

[0033] ;

[0034] in is the line voltage, is the line current, is the power factor, The coefficient representing the relationship between the line voltage and the phase voltage of the three-phase system, all parameters are derived from sensor data. The calculated power is then compared with the rated power of the motor. Compare and calculate the load rate L= / If the load factor continuously exceeds 1, an overload state is determined. Time series analysis is performed on the load factor data to extract features such as overload duration, overload peak value, and overload frequency to generate motor operation demand overload data. This data is stored in a structured format, including the overload event timestamp, overload intensity, and duration. The entire process is completed by edge computing nodes, using a sliding window method to analyze load factor changes and identify overload events in real time, ensuring the accuracy and timeliness of overload assessment data.

[0035] Step S13: performing a motor operation intensity assessment on the motor operation data based on the motor operation demand overload data to obtain motor operation intensity data;

[0036] In an embodiment of the present invention, the motor operation data is evaluated for operation intensity based on the motor operation demand overload data obtained in step S12. The operation intensity evaluation reflects the load accumulation of the motor within a given time range by multiplying the cumulative load rate by the operation time. The specific operation is to multiply the load rate at each sampling moment by the duration of that moment and then add them up to obtain the overall load intensity. This method takes into account both the load amplitude and the load duration, thereby quantifying the load usage intensity of the motor. The operation intensity data not only includes the total cumulative load, but also includes load change trend information, which helps to identify the state of long-term high-load operation. The evaluation results are stored in the database for use in the next step of load response status detection. Time synchronization technology is used in the data processing process to ensure the consistency of the load rate and time data to avoid calculation errors.

[0037] Step S14: detecting the motor operation load response status based on the motor operation data, the motor operation intensity data, and the motor operation demand overload data.

[0038] In an embodiment of the present invention, based on the motor operation data collected in step S11, the motor operation intensity data obtained in step S13, and the motor operation demand overload data calculated in step S12, the motor operation load response status detection is performed. The detection is completed by analyzing the indicators of three dimensions: load change amplitude, change rate, and overload frequency, and the maximum and minimum values ​​of the load rate are counted to calculate the load fluctuation amplitude to reflect the range of motor load changes. Secondly, the difference and time interval of the load rates of adjacent sampling points are calculated to obtain the load change rate, which reflects the dynamic characteristics of the load response. Finally, combined with the overload frequency data, the frequency of load response and the overload risk are analyzed. All test results are presented in the form of time series data, and combined with graphical display to help identify abnormal load response status. This step uses time synchronization and multi-data fusion technology to ensure the accuracy and completeness of the load response status detection results, and provide data support for subsequent motor abnormality diagnosis and maintenance.

[0039] Preferably, step S12 includes the following steps:

[0040] Step S121: performing motor operation demand timing analysis on the motor operation data to obtain motor operation demand timing data;

[0041] In an embodiment of the present invention, a timing analysis of the motor operation requirements is performed based on the motor operation data collected in step S11. The timing analysis is completed by processing the time series data of the motor operation status parameters such as current, voltage, speed, etc. The specific operations include: sorting the motor operation data according to the timestamp to ensure the integrity and continuity of the time series; using signal processing technology to eliminate noise interference, such as smoothing the sampled data through a filter to eliminate occasional errors in the sampling process. Subsequently, the processed time series data is segmented and counted according to the preset time window to obtain the motor load condition and change trend in each time window. Through this timing analysis, the motor operation requirement timing data that accurately reflects the change of the motor operation requirement over time is obtained, providing basic input data for subsequent frequency change statistics and preprocessing steps. This step ensures the timeliness and accuracy of the timing data through data synchronization and timestamp alignment technology.

[0042] Step S122: Counting the frequency changes of the motor operation demand based on the motor operation demand time series data;

[0043] In an embodiment of the present invention, based on the motor operation demand time series data obtained in step S121, the frequency change of the motor operation demand is statistically analyzed. In specific implementation, the motor operation demand time series data is decomposed into a series of characteristic points, and the characteristic points reflect the moments of significant changes in the motor load. The statistical process includes extracting the peak value, valley value and stable interval of the load change, calculating the frequency of the characteristic points appearing in unit time, and forming a frequency distribution. Through frequency change statistics, the fluctuation frequency of the motor load demand and its changing trend are analyzed. The statistical results are presented in the form of a frequency change curve, reflecting the dynamic fluctuation of the motor load demand. This step uses frequency statistical analysis technology, combined with the segmented processing of the time series, to ensure the accuracy and continuity of the statistical data, and provide valid input data for the frequency data preprocessing step.

[0044] Step S123: performing frequency change data preprocessing on the motor operation demand time series data and the motor operation demand frequency change to obtain motor operation demand frequency preprocessing data;

[0045] In an embodiment of the present invention, the frequency change of the motor operation demand obtained in step S122 is combined with the time series data of the motor operation demand in step S121 to perform frequency change data preprocessing. The preprocessing process includes data denoising, outlier removal and normalization. The specific operations are: by setting a threshold value to filter out abnormal points that are abnormally high or abnormally low in the frequency change, so as to avoid the interference of abnormal data on subsequent analysis; using the sliding window method to smooth the frequency change data, reduce the amplitude of data fluctuations, and highlight the real trend; normalize the time series data and the frequency change data so that the two are in the same numerical range, which is convenient for fusion analysis. The data structure of the motor operation demand frequency preprocessing data formed after preprocessing is complete, the fluctuation trend is clear, and it can accurately reflect the real changes in the motor load demand. This step ensures the reliability and consistency of the preprocessed data through multi-dimensional data fusion technology, laying the foundation for the construction of the fluctuation change graph.

[0046] Step S124: constructing a motor operation demand fluctuation change graph using the motor operation demand frequency preprocessing data;

[0047] In an embodiment of the present invention, the motor operation demand frequency preprocessing data obtained in step S123 is used to construct a motor operation demand fluctuation change graph. The specific implementation is: the preprocessing data is plotted into a fluctuation curve graph in chronological order. The curve graph uses time as the horizontal axis and the frequency change value as the vertical axis to fully reflect the fluctuation of the motor operation demand over time. The fluctuation change graph uses a high-resolution time axis resolution to capture subtle changes in load demand, which facilitates the subsequent accurate extraction of the fluctuation growth interval. During the drawing process, digital drawing tools are used to smoothly connect the data points to ensure the continuity and smoothness of the fluctuation curve. The fluctuation change graph intuitively presents the fluctuation amplitude and change rate of the motor load demand, provides visual support for the fluctuation growth interval acquisition step, and ensures the accuracy of the analysis.

[0048] Step S125: collecting motor operation demand fluctuation growth interval information based on the motor operation demand fluctuation change graph;

[0049] In an embodiment of the present invention, based on the motor operation demand fluctuation change graph constructed in step S124, information on the motor operation demand fluctuation growth interval is collected. The collection process is: by scanning the fluctuation change curve, the time interval in which the load demand shows a continuous upward trend is identified, and it is determined to be a fluctuation growth interval. The specific operation is to set an incremental threshold, traverse the data points of the fluctuation change graph, and detect whether the increase of adjacent data points exceeds the threshold continuously. If the continuous time exceeds the preset minimum interval length, the interval is calibrated as a fluctuation growth interval. The collected fluctuation growth interval includes the start time, end time and fluctuation amplitude within the interval, providing basic data for subsequent slope calculation. This process is based on time series analysis technology and threshold judgment algorithm to achieve automatic interval identification and ensure the accuracy and completeness of the collected information.

[0050] Step S126: calculating the maximum slope information of the motor operation demand fluctuation according to the motor operation demand fluctuation growth interval information;

[0051] In an embodiment of the present invention, the maximum slope information of the motor operation demand fluctuation is calculated based on the motor operation demand fluctuation growth interval information collected in step S125. The calculation process is: differential analysis is performed on the time and load demand data in each fluctuation growth interval, and the maximum load change rate between any two time points is found as the maximum slope of the interval. The specific operation includes traversing all adjacent sampling points in the interval, calculating the ratio of the load change to the time difference, and recording the maximum value. The maximum slope reflects the most drastic change rate of the motor load demand fluctuation. This calculation process is implemented through time series differentiation technology to ensure that the results accurately reflect the extreme conditions of load demand changes and provide key parameters for subsequent overload assessment.

[0052] Step S127: calculating the average slope data of the demand fluctuation growth interval according to the motor operation demand fluctuation growth interval information;

[0053] In an embodiment of the present invention, based on the motor operation demand fluctuation growth interval information obtained in step S125, the average slope data of the demand fluctuation growth interval is calculated. The calculation steps are: in each fluctuation growth interval, the arithmetic mean of the load demand change rate between all sampling points is counted. The specific operation is: for each pair of consecutive sampling points in the interval, the ratio of the load change amount to the time difference is calculated, and the average value of all ratios is obtained as the average slope of the interval. The average slope reflects the overall trend and intensity of the load demand change in the fluctuation growth interval. This calculation is completed by time series differentiation and average value statistical technology. The result is used to judge the degree of fluctuation of the motor load demand, providing an important basis for overload assessment.

[0054] Step S128: When the average slope data of the demand fluctuation growth interval exceeds 1.43 times / h² and the maximum slope information of the motor operation demand fluctuation exceeds 8.4 times / h², a motor operation demand overload assessment is performed to obtain motor operation demand overload data.

[0055] In an embodiment of the present invention, an overload assessment of the motor operation demand is performed based on the maximum slope information of the motor operation demand fluctuation calculated in step S126 and the average slope data of the demand fluctuation growth interval calculated in step S127. The specific operation is: compare the maximum slope with the threshold value of 8.4 times / hour squared, and compare the average slope with the threshold value of 1.43 times / hour squared. If both exceed the corresponding threshold value, it is determined that the motor is in an operation demand overload state. Subsequently, the judgment result and the corresponding maximum slope and average slope values ​​are integrated to form a complete motor operation demand overload data set. The data set includes detailed information such as the overload start time, end time, overload amplitude and duration, providing a data basis for subsequent operation intensity evaluation. The overload assessment is based on a strict threshold judgment method to ensure the accurate identification of the overload state and the integrity of the data record.

[0056] Preferably, step S14 includes the following steps:

[0057] Step S141: extracting the motor operation power surge state based on the motor operation demand overload data and the motor operation intensity data;

[0058] In an embodiment of the present invention, time series analysis techniques are used to detect power change trends by collecting motor operating demand overload data and motor operating intensity data. The specific operation process is as follows: Real-time motor operating demand overload data is acquired from the Internet of Things system. This data reflects the overload state of the motor when the load demand exceeds the designed load and includes timestamps and overload magnitude information. Simultaneously, motor operating intensity data is acquired, which refers to the real-time measurements of the motor load current and voltage, reflecting the current load level of the motor. The two types of data are aligned using a unified time base to construct a multidimensional time series data set. Subsequently, a sliding window technique is used to gradually calculate the incremental change in motor operating power over a set time period. The calculation method is based on the basic electrical relationship that power equals current multiplied by voltage. Threshold detection is performed on the incremental power sequence to identify power surge states. In the specific technical implementation, a power increase threshold is set. By comparing the threshold with the power surge threshold, the time period and magnitude of the power surge are marked. Finally, a motor operating power surge state data set containing the power surge time nodes and corresponding magnitudes is output.

[0059] Step S142: identifying the motor high power demand data based on the motor operating power sudden increase state exceeding 41.5% and the motor operating demand overload data;

[0060] In an embodiment of the present invention, based on the motor operating power surge state extracted in step S141 and combined with the motor operating demand overload data, a determination is made as to whether the motor is in a high power demand state. The specific operation is to perform a percentage calculation on each power surge value in the motor operating power surge state, that is, to compare the current power surge value with the reference power of the previous normal operating state, and calculate the power surge ratio. If the ratio exceeds the set threshold of 41.5%, the motor is deemed to be in an abnormally high power demand state. At the same time, the power surge time node is time-matched with the motor operating demand overload data to ensure the synchronization of the power surge and demand overload events. Based on this matching result, all time periods that meet the power surge ratio exceeding 41.5% and correspond to demand overload are filtered and marked to form a motor operating high power demand data set. This data set contains detailed information such as timestamp, high power amplitude, corresponding overload intensity, etc., which is convenient for subsequent processing.

[0061] Step S143: collecting the motor running rated power threshold value according to the motor running data to obtain the motor running rated power threshold value data;

[0062] In an embodiment of the present invention, the rated power threshold of the motor is collected from the actual operating data, and this is used as a benchmark for subsequent power over-limit judgment. In specific operations, the voltage and current data of the motor under normal operating conditions are collected, and the data at multiple time points are statistically analyzed using the power calculation formula (power = voltage × current × power factor). In order to eliminate the impact of short-term abnormal fluctuations on the threshold, a segmented statistical method is adopted, that is, the collected data is divided into several time periods (such as hours or days), and the average power value in each time period is calculated. The maximum value of the average power of all time periods is then taken as the rated power threshold of the motor. This threshold reflects the maximum operating power level of the motor under normal load conditions, ensures that the reference power is measured without overloading, and outputs the motor rated power threshold data containing the threshold data and the statistical time range for subsequent power over-limit judgment.

[0063] Step S144: estimating the motor operating power over-limit condition by using the motor operating rated power threshold data when the motor operating high power demand data exceeds 5.8 kW;

[0064] In an embodiment of the present invention, based on the motor operation high power demand data obtained in step S142, combined with the motor operation rated power threshold data collected in step S143, it is determined whether the motor has a power overlimit condition. The specific implementation logic is to filter out events in which the power value exceeds 5.8kW in the high power demand data. 5.8kW is the preset power overlimit starting point, which is a specific value determined in combination with the motor design power and operating experience. The filtered data is compared with the motor operation rated power threshold data one by one. If the high power demand value exceeds the rated power threshold, it is determined that the motor is in a power overlimit state at that time point. The power overlimit condition includes detailed information such as timestamp, overlimit amplitude (i.e., the difference between the actual power and the rated power threshold) and duration, which facilitates the assessment of the severity and persistence of the power overlimit. Finally, a data set of motor operation power overlimit condition is formed as the basic data for subsequent nonlinear load trend detection.

[0065] Step S145: detecting a nonlinear load growth trend of the motor based on the motor operating power exceeding limit condition;

[0066] In an embodiment of the present invention, the motor operating power overrun status data from step S144 is used to analyze the nonlinear growth trend of the motor load over time, arrange the power overrun status in chronological order, and extract the overrun power amplitude and corresponding time information. A numerical differentiation method is used to calculate the rate of change of the overrun power value over time, focusing on whether the overrun power amplitude growth exhibits nonlinear characteristics, such as exponential growth or polynomial growth. In conjunction with time series analysis techniques, a growth curve of the overrun power amplitude over time is plotted, and the shape of the curve is analyzed to determine whether it shows an accelerating growth trend. If the growth trend conforms to nonlinear characteristics, the trend is marked as a nonlinear growth trend of the motor load. At the same time, statistical methods (such as curve fit goodness assessment) are used to verify the stability and significance of the nonlinear growth trend. The output contains descriptive data of the nonlinear growth trend, such as the growth rate, time interval, and fitting parameters, for subsequent load response status detection.

[0067] Step S146: Detecting the motor operation load response condition based on the nonlinear load growth trend of the motor operation and the motor operation power exceeding limit condition.

[0068] In an embodiment of the present invention, the motor nonlinear load growth trend data obtained in step S145 and the power overlimit status data in step S144 are used to detect the motor load response status, align the nonlinear load growth trend and the power overlimit data time axis, and analyze the time delay and response amplitude in the load response process. Using the time series correlation analysis technology, the correlation coefficient between the load growth trend and the power overlimit event is calculated to evaluate the triggering effect of the load change on the power overlimit. By comparing and analyzing the response time interval, the dynamic characteristics of the load response are identified, including fast response, delayed response and other states. Further, combined with the motor operating parameters, the linear or nonlinear properties of the load response are determined to distinguish between normal response and abnormal response. The motor operation load response status data is output, including indicators such as response type, response intensity and response duration, providing a key basis for the detection of motor operation anomalies.

[0069] Preferably, determining the bearing kinetic energy chain coupling decay state in step S2 includes:

[0070] Detect the motor shaft over-rotation impact load according to the motor operation load response condition;

[0071] In an embodiment of the present invention, the real-time load response data of the motor shaft system is collected through the Internet of Things sensor network, including multi-dimensional data such as speed, torque, vibration frequency and axial acceleration. The load response status is specifically collected by a high-precision acceleration sensor and a torque sensor installed at the motor bearing seat, and is pre-processed in real time by the edge computing node to eliminate environmental noise and normal fluctuations in the mechanical transmission process. The over-rotation impact load of the motor shaft system is detected by a joint analysis method of time domain and frequency domain. The time domain determines the sudden load event by the load peak and duration, and the frequency domain uses the fast Fourier transform (FFT) to identify the characteristic frequency component of the impact load, with special attention to the frequency segment corresponding to the high-frequency impact peak. By defining the threshold function of the over-rotation impact load, such as:

[0072] ;

[0073] in, is the peak value of the impact load, is the load reference value. When the load exceeds a preset percentage, it is considered an over-rotation impact load. This threshold is derived from historical operating data and updated in real time to adapt to changes in motor load. The test results generate motor shaft over-rotation impact load status data, which serves as input for subsequent steps.

[0074] Measure the bearing rotation pressure growth based on the motor shaft over-rotation impact load and the motor operating load response;

[0075] In an embodiment of the present invention, a comprehensive analysis is performed based on the obtained over-rotation impact load event data in combination with the bearing rotation pressure data collected by the bearing's built-in pressure sensor. The pressure sensor is located in the lubrication cavity of the bearing and collects the lubricating oil pressure fluctuations caused by the dynamic load. Through time synchronization processing, the time point of the impact load event is matched with the pressure fluctuation curve, and the pressure change rate is calculated using the sliding window technology. By setting the pressure growth threshold, the abnormal period of pressure growth is identified. A filtering algorithm (such as Kalman filtering) is used to eliminate measurement errors to ensure accurate reflection of the pressure growth trend. The bearing rotation pressure growth curve and its statistical parameters are obtained as the basis for judging gear deformation.

[0076] Determine the motor bearing gear deformation based on the bearing rotation pressure growth;

[0077] In this embodiment of the present invention, pressure growth data is combined with strain signals collected by gear tooth surface strain sensors to analyze gear deformation. Strain signals are collected by strain gauge sensors installed at key locations on the gear tooth surfaces. The signal sampling frequency is more than 10 times the gear rotation frequency, ensuring real-time capture of deformation changes. The pressure data and strain data are input into the deformation calculation formula:

[0078]

[0079] in, is the gear deformation, is the small deformation length of the gear tooth surface, is the initial length, is the function relationship between pressure and strain, is the pressure on the gear, is a function of the state parameters. Multivariate regression analysis is used to determine the quantitative relationship between pressure growth and gear deformation. Deformation data is dynamically monitored, and deformation stages exceeding the material's elastic limit are identified as gear deformation states. This deformation data is used for subsequent gear mesh conflict detection.

[0080] Detect the motor bearing gear meshing conflict condition based on the motor bearing gear deformation condition;

[0081] In an embodiment of the present invention, based on gear deformation data, high-frequency vibration sensors and acoustic emission sensors are used to jointly detect conflicts during gear meshing. Gear meshing conflicts manifest as abnormal vibrations and impact sound signals. The sensor collects signals and extracts instantaneous impact characteristics through time-frequency analysis methods (such as wavelet transform). A conflict condition is defined as a vibration amplitude that exceeds a specific multiple (for example, 3 times) of the normal gear meshing vibration mean and the duration of the impact signal exceeds a preset time threshold. The signal processing algorithm eliminates normal mechanical operation noise and accurately identifies the conflict period. Detection accuracy is improved by cross-validating vibration and acoustic emission signals. The detection results output gear meshing conflict status data, which serves as a basis for predicting gear breakage trends.

[0082] Predict the motor bearing gear fracture trend based on the motor bearing gear meshing conflict condition and motor bearing gear deformation condition;

[0083] In the embodiment of the present invention, the prediction of the fracture trend is based on the comprehensive analysis of the gear conflict frequency, impact strength and gear deformation accumulation. The gear strain fatigue accumulation method based on historical data statistics and real-time collection is used to calculate the critical time of gear fracture. :

[0084]

[0085] in, is the number of cycles of fatigue life of gear material, The number of collisions currently detected. Fatigue life is determined based on experimental data for the gear material. Gear deformation is factored into the calculation as a fatigue acceleration factor; greater deformation reduces fatigue life. Time series analysis of collision and deformation data identifies critical trends in fracture. This predicted data is used for subsequent backlash misalignment assessments.

[0086] Evaluate the motor bearing gear backlash imbalance based on the motor bearing gear cracking trend;

[0087] In this embodiment of the present invention, gear backlash misalignment is monitored using a gear backlash sensor that measures the actual backlash during gear meshing. The cracking trend is used as a warning parameter, combined with real-time backlash measurements, and a time series filtering algorithm is used to determine the backlash change trend. Backlash misalignment is defined as the deviation of the backlash from the standard design value exceeding a certain threshold, specifically:

[0088]

[0089] in, To measure the backlash in real time, To design the backlash, The maximum allowable deviation. Misalignment trends are confirmed through multiple consecutive measurements to prevent single-point errors from affecting judgment. Output data on gear backlash misalignment status provides a basis for detecting abnormal lateral thrust in bearings.

[0090] Detect abnormal lateral thrust growth of the bearing based on the side clearance imbalance of the motor bearing gear;

[0091] In this embodiment of the present invention, abnormal lateral thrust is detected by a force sensor installed in the axial direction of the bearing. This sensor collects real-time data on bearing force changes. Combined with the detected side clearance imbalance, the fluctuation characteristics of the lateral thrust are analyzed. Thrust growth is determined by the thrust change rate calculation formula:

[0092]

[0093] in, For the moment The lateral thrust value. If the thrust growth continues to exceed the set percentage, it is considered abnormal. The data acquisition module and real-time alarm system synchronously record abnormal events, forming an abnormal lateral thrust growth curve, which serves as an important parameter for determining the degradation state of the bearing kinetic energy chain coupling.

[0094] The bearing kinetic energy chain coupling degradation state is determined based on the abnormal lateral thrust growth of the bearing and the side clearance imbalance of the motor bearing gear.

[0095] In this embodiment of the present invention, a multi-index fusion algorithm is used to combine the lateral thrust growth data and the gear backlash misalignment data to achieve a comprehensive determination of the bearing kinetic energy chain coupling degradation state. The fusion algorithm converts the two types of state data into a unified state score through weighted averaging and normalization:

[0096]

[0097] in, Score abnormal lateral thrust, Score the gear backlash misalignment, weight coefficient represents the weight distribution of lateral thrust score in the overall coupling degradation judgment, Represents the weight distribution of the side clearance misalignment score in the overall coupling degradation judgment, which is empirically assigned based on historical experimental data and meets = 1. When the score exceeds the preset threshold, the bearing kinetic energy chain coupling is deemed to have entered a decaying state. This state data is stored in the IoT cloud platform database, providing decision support for motor anomaly detection.

[0098] Preferably, determining the decreasing trend of the motor structure connection coordination in step S2 includes:

[0099] Detecting the eccentricity of the motor shaft based on the load response of the motor;

[0100] In an embodiment of the present invention, during the operation of the motor, a periodic voltage-current-torque correlation analysis is performed based on the load response condition of the motor operation, focusing on extracting three groups of key characteristic parameters: current peak duration, torque output mean, and power fluctuation rate, and normalizing them into an average dynamic response vector within a unit time series. Subsequently, based on the rated operating conditions of the motor and the load fluctuation sensitivity threshold standard, the periodic asymmetric peaks and fluctuation deviation amplitudes appearing in the dynamic response vector are analyzed. If the power fluctuation deviates from the mean by more than 10% of the rated value, and the peak position shifts or jumps within the period, it is determined that there is an eccentric operation feature of the shaft. This conclusion is synchronously recorded in the shaft status mark table and used as an input parameter for subsequent steps.

[0101] Determine the rotational symmetry imbalance of the motor shaft according to the eccentric running condition of the motor shaft;

[0102] In an embodiment of the present invention, after obtaining the eccentric operating state mark of the rotating shaft, the rotational symmetry analysis is further performed in combination with the angular displacement data collected by the rotary encoder at the end of the rotating shaft and the instantaneous vibration vector sequence captured by the three-axis acceleration sensor. The specific operation is: within the unit rotation period, the displacement data is segmented and reconstructed at an angular resolution of 0.5°, and the difference is compared with the theoretical constant velocity circular trajectory. The area where the deviation vector is greater than 1% of the shaft radius will be marked as a local eccentric point. The three-axis vibration data is then used for time matching verification, and the rotational symmetry breakpoint confirmation is achieved by identifying the periodicity and offset of the acceleration extreme points. If there are multiple asymmetric peaks in the same period and the position is stable, it is determined that the rotational symmetry is unbalanced. This flag is written into the vibration symmetry evaluation cache queue to provide basic support data for subsequent inertia moment analysis.

[0103] Test the unevenness of the motor shaft inertia moment based on the imbalance of the motor shaft's rotational symmetry;

[0104] In an embodiment of the present invention, when a rotational symmetry imbalance of the rotating shaft is detected, the structural parameters of the rotating shaft (mass distribution, radial dimensions, etc.) and the three-dimensional vibration superposition vector data during operation are retrieved, and inertia distribution mapping is performed in the embedded structural dynamic analysis unit. The specific method is as follows: a full cycle of rotational motion is divided into 10ms frames, the equivalent force changes at each point on the rotating shaft within each frame are calculated, and the distribution trend of the moment of inertia is calculated by numerical integration accumulation. If the same cross-section shows an asymmetric distribution of inertia in different time frames (i.e., there is a difference of more than 30% in the instantaneous moment of inertia), it is recorded as an uneven moment of inertia phenomenon. At the same time, the eccentricity and asymmetric imbalance data of the previous stage are superimposed, the inertia change curve is filtered and cleaned, and the inertia offset degree vector (unit: kg·m² change rate) is output as the input data item for the periodic vibration trend assessment.

[0105] Detect the periodic vibration trend of the motor shaft based on the unevenness of the motor shaft inertia moment and the imbalance of the motor shaft rotational symmetry;

[0106] In an embodiment of the present invention, this rotational symmetry imbalance mark and the output inertia offset degree vector are used to extract the periodic vibration frequency and amplitude characteristics through a time series analysis module (such as an FFT-based spectrum density analyzer). The monitoring time window is set to 5s and the step size is 100ms. The window is slid segment by segment to extract the main frequency component and its corresponding amplitude change. If the main frequency peak shows a steady increase in amplitude and the amplitude growth rate exceeds 20% of the previous cycle, it is defined as a periodic vibration enhancement trend. The system also records parameters such as the starting frequency, peak frequency, and peak growth time point, and forms a periodic vibration trend curve, which is output to the vibration trend database.

[0107] Detect the increasing looseness of the connection parts based on the periodic vibration trend of the motor shaft;

[0108] In an embodiment of the present invention, based on periodic vibration trend data, micro-strain gauges located at the connection point between the motor end cap and the base, as well as a displacement sensing array on the motor's fixed base, are used to monitor stress release and displacement offset at the connection. The main frequency of the periodic vibration over a continuous 24-hour period is mapped to the time series of the micro-displacement of the connection point to identify the increase trend of the connection displacement extreme point. If the connection point displacement extreme value shows a continuous upward trend within each unit cycle (10 minutes) and the corresponding main frequency is synchronously offset, the connection is deemed to have a loosening trend. This is determined by setting a threshold (e.g., displacement increase > 0.05 mm / hour) and a strain release ratio greater than 20%, and outputting the loosening trend status.

[0109] According to the increasing looseness of the connection parts, the motor shaft periodic vibration trend is used to predict the growth trend of the motor structure resonance frequency;

[0110] In an embodiment of the present invention, under the premise of detecting the trend of loose connection, the high-frequency vibration frequency distribution fitting is performed on the curve in combination with the previously obtained periodic vibration curve. During the spectrum analysis process, the frequency drift trajectory in the sub-high frequency band (generally ranging from 1.5 to 2.5 times the main frequency) induced by looseness is specifically extracted, and the average growth rate of its starting value and ending value is tracked. Combined with the motor housing stiffness information and the mounting platform response frequency curve, it is evaluated whether there is a trend of the structural resonance frequency shifting to the high-frequency side. If the frequency deviation trend persists for more than three analysis cycles (20 minutes per cycle) and the increase exceeds 0.5Hz, it is determined that the structural resonance frequency has an increasing trend, and the frequency change sequence and its growth slope information are output for subsequent structural coordination analysis.

[0111] The decreasing trend of the motor structure connection coordination is determined based on the growth trend of the motor structure resonance frequency and the unevenness of the motor shaft inertia moment.

[0112] In an embodiment of the present invention, the dynamic coordination of the overall structure of the motor is judged based on the superposition analysis of the uneven degree of inertia moment and the frequency growth trend value. The analysis method is: extract the time point of the peak value of the uneven vector of the inertia moment of the rotating shaft, and match it with the peak value of the resonance frequency change on the time axis. If the two have an overlap of more than 80% on the time axis, it indicates that the inertia change is related to the structural response. Then, using the structural coupling analysis module, the synchronous response frequencies between the four structures of the housing-stator-bearing-base in the motor structure are compared. If the coupling frequency offset between the structures exceeds 10Hz, and the cross-structure delay propagation time increases significantly, it is determined that the structural coordination has decreased. Output the coordination attenuation mark, and form a structural risk trend graph, which is transmitted to the monitoring terminal for abnormal warning display.

[0113] Preferably, determining the imbalance state of the dynamic magnetic field during operation of the motor in step S2 includes:

[0114] The degree of reduction in the motor's operating force transmission capability is tested based on the bearing kinetic energy chain coupling decay state and the decreasing trend of the motor structure connection coordination;

[0115] In one embodiment of the present invention, during motor operation, a high-frequency, three-dimensional vibration acceleration sensor array, located at the connection between the motor housing and the rotating shaft, records dynamic vibration data of the bearing in multiple directions in real time. Combined with the bearing's heat accumulation as measured by the temperature rise sensor and the axial and radial output power trends monitored by the torque sensor, the kinetic energy transfer efficiency of the bearing's kinetic energy chain is accurately assessed. The degradation of the kinetic energy chain coupling performance is quantified by statistically comparing the vibration phase angle changes of the bearing's inner and outer rings, the rolling element vibration frequency stability, and the lubrication status response time under different operating conditions. Subsequently, micro-deformation data is collected from the motor's structural joints (such as the base and end cap, and the housing and bearing seat). Based on the deformation curves of these joints generated by the induction strain gauges and the fiber Bragg grating array, the coordination of the structural connections is determined to determine whether synchronous imbalance behavior has occurred. This, in turn, establishes a dynamic coordination degradation trend criterion at the structural level. Finally, the bearing kinetic energy chain coupling degradation index is coupled with the structural coordination degradation trend factor for analysis to determine the degree of reduction in the motor's force transfer capability under the current operating conditions and generate a transfer capability degradation level parameter.

[0116] Determine the reduction in bearing centripetal stability based on the reduction in the motor's operating force transmission capability;

[0117] In an embodiment of the present invention, the centripetal stability performance of the bearing during operation is further precisely tested based on the conduction capacity degradation level parameter obtained in the test. A high-sensitivity laser interferometer displacement measurement module is set on the inner ring of the bearing to monitor the slight changes in the radial offset of the inner ring of the bearing during the rotation cycle. At the same time, it is combined with the eddy current displacement sensor installed at the rolling element channel position to detect whether the running trajectory of the rolling element shows irregular offset or jumping behavior. By synchronously analyzing the centripetal force retention curves under the motor startup, steady-state operation and sudden stop processes, the force balancing ability of the bearing under different loads is evaluated. Combined with the force conduction degradation level parameter and the structural response lag time difference, the degree of stability reduction of the bearing in centripetal constraint is determined, and a centripetal stability level coefficient is generated for use in the next step of rotor motion trajectory deformation analysis.

[0118] Predicting rotor ellipticity based on bearing centripetal stability degradation;

[0119] In an embodiment of the present invention, based on the obtained bearing centripetal stability grade coefficient, the rotor running trajectory is further analyzed for elliptical trend. A dual-axis high-resolution magnetostrictive displacement encoder is used, with radial measurement points arranged at both ends of the shaft. By analyzing the statistical distribution of the radial displacement amplitude within each rotation cycle, the main direction displacement change amplitude and the secondary direction displacement change amplitude are extracted to form a cross-sectional diagram of the actual rotation trajectory. The data is superimposed and compared with the ideal circular trajectory, and the main eccentric path and the major-minor axis ratio of the rotating ellipse are extracted. If the major-minor axis ratio exceeds the set threshold, combined with the support imbalance phenomenon caused by the decrease in the bearing centripetal stability, it is determined that the rotor has obvious elliptical rotation behavior, and the elliptical offset parameter is generated.

[0120] Determine the motor rotor position disturbance condition based on the bearing centripetal stability degradation and rotor rotation elliptical phenomenon;

[0121] In an embodiment of the present invention, the spatial position disturbance of the rotor during a complete operating cycle is comprehensively evaluated based on the elliptical offset parameters and the centripetal stability grade coefficient. Multi-point laser interferometer locators are deployed in three-dimensional space to continuously sample the position vector of the rotor within any time period to obtain a rotor center trajectory disturbance cloud map. Combined with the micro-oscillation frequency of the bearing support point and the dynamic offset trend caused by the asymmetric axial load, the rotor position disturbance amplitude, disturbance frequency and directional stability index are extracted to form a three-dimensional vector set of rotor position disturbance characteristics and a disturbance dynamic grade label to provide input for the next step of rotor-stator gap analysis.

[0122] Monitor the rotor-stator constant distance fluctuation according to the motor rotor position disturbance;

[0123] In this embodiment of the present invention, based on the generated disturbance dynamic level labels and three-dimensional disturbance vector sets, real-time data on the actual operating gap change between the outer edge of the rotor and the inner edge of the stator is collected. Using an array of capacitive gap sensors equidistantly arranged on the inner wall of the stator, the gap change curve during rotor motion is periodically sampled to extract the gap fluctuation data at different rotor angular positions. By comparing the sampling curves with ideal equidistant spacing, gap contraction points and gap expansion points are identified, and frequency distribution statistics are performed to calculate the fluctuation frequency, amplitude, and period, and generate a gap fluctuation level index.

[0124] Estimate the air gap flux density disturbance based on the rotor-stator constant spacing fluctuation;

[0125] In an embodiment of the present invention, the rotor-stator gap fluctuation level index is used as input data, combined with the motor's current operating voltage, current, and frequency parameters, to measure the air gap magnetic flux density disturbance under the influence of gap changes. Relying on a high-precision Hall effect magnetic flux density sensor array arranged in the stator slots, the air gap magnetic flux change trajectory within one rotor cycle is densely sampled. The rate of change of magnetic flux density under equidistant and non-equidistant gaps is compared to identify abnormal mutation points and continuous disturbance trends. An air gap magnetic flux disturbance intensity index is constructed based on dimensions such as the flux disturbance density change rate, disturbance amplitude, and duration.

[0126] Calculate the degree of deviation of the motor's multi-pole electromagnetic pull according to the air gap magnetic flux density disturbance condition;

[0127] In this embodiment of the present invention, based on the magnetic flux disturbance intensity index, the variation in electromagnetic pull between each pole pair within the motor is calculated, combining the number of stator pole pairs and the distribution of rotor magnetic poles. An embedded axial magnetic sensor unit measures the pull differences between different pole pairs within the motor over the operating cycle. Combined with the aforementioned air gap flux disturbance data, this method assesses whether the electromagnetic attraction exhibits periodic fluctuations or deviations. A multi-pole electromagnetic pull deviation index is generated based on the maximum difference in electromagnetic pull between each pole pair, the inter-pole non-uniformity coefficient, and the pull variation frequency.

[0128] Predict the center deviation trend of the electromagnetic force vector based on the motor's multi-pole electromagnetic pull deviation and air gap flux density disturbance;

[0129] In this embodiment of the present invention, the electromagnetic force deviation level indicator is coupled with the air gap flux disturbance intensity indicator to generate an electromagnetic vector force field distribution map. Using vector superposition, the center of force within the motor's overall electromagnetic field is calculated over a complete operating cycle, and its dynamic trajectory in spatial coordinates is tracked as a time series. Based on the center trajectory deviation amplitude, deviation frequency, and trajectory stability, a trend map of the electromagnetic force vector center deviation is generated, providing a quantitative basis for dynamic magnetic field balance assessment.

[0130] The dynamic magnetic field imbalance condition of the motor is determined based on the center deviation trend of the electromagnetic force vector and the air gap magnetic flux density disturbance condition.

[0131] In this embodiment of the present invention, a model for evaluating the spatial stability of the magnetic field during motor operation is established based on the electromagnetic force center deviation trend map and the magnetic flux disturbance intensity index, constructing a three-dimensional dynamic magnetic field distribution map. Using magnetic field concentration, flux path deviation, and pole symmetry changes as evaluation dimensions, an indicator of the dynamic magnetic field imbalance level during motor operation is extracted by comparing it to ideal uniform magnetic field conditions. This indicator, which encompasses the electromagnetic disturbance frequency, the degree of deviation from the electromagnetic vector rotation path, and the vector stability coefficient, serves as a key input parameter for subsequent torque fluctuation and structural anomaly diagnosis.

[0132] Preferably, step S3 includes the following steps:

[0133] Step S31: estimating the growth trend of the electromagnetic interference risk of the motor according to the dynamic magnetic field imbalance condition of the motor;

[0134] In this embodiment of the present invention, during motor operating status monitoring, the motor's dynamic magnetic field imbalance has been determined in the previous step. To estimate the resulting electromagnetic interference risk growth trend, precise detection of the motor's peripheral electromagnetic radiation field within a high-frequency bandwidth is required. The specific operational process is as follows: an array of equidistant EMI detection antennas (operating in the 150kHz-30MHz frequency range) is deployed in multiple locations outside the motor housing. High-sensitivity radio frequency interference analysis equipment (such as a Tektronix RSA500 series real-time spectrum analyzer) is used to continuously sample the electromagnetic interference intensity around the magnetic field imbalance area. The time-synchronized sampling data from each antenna is transmitted to the central processing unit via the RS485 bus. Time-domain / frequency-domain coupled analysis is performed in conjunction with air gap magnetic flux density disturbance distribution data and rotor eccentricity fluctuation data. An interference risk level quantification matrix is ​​established based on the interference frequency growth rate, the amplitude of electromagnetic radiation power variation, and the degree of spatial distribution unevenness. The electromagnetic interference risk growth trend value (in dBμV / Hz / s) is output as the output data for this step.

[0135] Step S32: detecting the motor operating current waveform distortion based on the motor electromagnetic interference risk growth trend and the motor operating dynamic magnetic field imbalance;

[0136] In this embodiment of the present invention, electromagnetic interference risk growth trend parameters and magnetic field imbalance parameters are used as input factors to detect motor current waveform distortion during operation. To ensure data accuracy and synchronization, full-scale, high-sampling-rate current transformers (e.g., the Lem HAH3DR series, with a bandwidth of up to 300kHz) are deployed at the motor input and output terminals. A differential current waveform analysis module (with a 1MHz sampling frequency and 16-bit accuracy) is used to upload current waveform data in real time via the CAN-FD industrial communication bus. During waveform analysis, a fifth-order wavelet decomposition method (based on Daubechies basis functions) is used to perform segmented filtering and local feature extraction on the current signal. A baseline template of the normal current waveform is set and compared with the real-time decomposed waveform to identify distorted sections caused by magnetic field disturbances. The distortion rate is output as a percentage of the total distortion (THD%) of the current signal. The dominant frequency, amplitude, and harmonic order of the distorted signal are recorded to serve as the basis for determining amplitude and frequency offset in subsequent steps.

[0137] Step S33: determining the motor operating amplitude drift condition based on the motor operating current waveform distortion condition;

[0138] In this embodiment of the present invention, based on the acquired current waveform distortion data, the focus is on analyzing the current amplitude offset within the stable operating voltage range. This step utilizes synchronized three-phase current channels and three-axis Hall sensor data for fusion processing. High-frequency sampling (e.g., a 5MHz sampling rate) is used to obtain the waveform peak value and effective value variation range, and a time series fluctuation curve of the peak amplitude of each phase is recorded. The peak-to-peak rate of change (in V / ms) and the three-phase amplitude difference are used as evaluation indicators to quantify the amplitude change before and after waveform distortion. Furthermore, the proportion of harmonic energy in the waveform is combined to determine whether the amplitude drift exhibits nonlinear growth characteristics. If the amplitude offset exceeds the ±5% variation limit of the rated amplitude and is accompanied by frequent and dramatic amplitude fluctuations, significant operating amplitude drift is determined under these operating conditions. This drift data is output as the three-phase maximum offset amplitude, drift duration, and interphase offset difference, and is annotated as timestamp sequence data for subsequent frequency analysis steps.

[0139] Step S34: determining the oscillation growth condition of the motor operating frequency according to the motor operating amplitude drift condition and the motor operating current waveform distortion condition.

[0140] In this embodiment of the present invention, based on the current amplitude drift data and waveform distortion analysis results output in the previous step, this step determines the growth of the motor's operating frequency oscillation and analyzes its trend. A digital signal processing (DSP) chip performs spectral tracking of the collected current waveform within a short-time Fourier window to detect the trajectory of the main frequency changes. A sliding peak FFT method is introduced to compare the positions of the main frequency peaks before and after each 100ms interval. If the main frequency drift range continuously exceeds twice the set threshold (for example, the drift amplitude of a 50Hz system exceeds ±1.5Hz) and is accompanied by a concentrated shift in high-order harmonic energy density, the motor's operating frequency oscillation is determined to have entered a growth phase. A frequency oscillation trend curve is constructed by combining the frequency of frequency mutation points, the time series of amplitude drift extremes, and the number of phase reversals. The frequency oscillation increment ratio (in Hz / s) is output as a key parameter. This parameter is used as a dynamic factor in the evolution of system operational risks and fed back into the overall motor operating status map.

[0141] Preferably, step S31 includes the following steps:

[0142] Step S311: collecting multi-point dynamic magnetic flux density distribution data according to the dynamic magnetic field imbalance condition of the motor;

[0143] In one embodiment of the present invention, a distributed magnetic field measurement network is used to collect dynamic magnetic flux density distribution data during motor operation. Specifically, several magnetic flux density sensing nodes are configured, each integrating a Hall effect magnetic flux density detection unit. Each detection unit is deployed at key locations around the motor windings, at the stator ends, and within the air gap. The nodes are connected to an IoT data collection gateway via the RS485 industrial bus communication protocol. The sampling frequency is set to 10 kHz, and the sampling duration is controlled within a continuous 10-second cycle to capture the dynamic magnetic field distribution during the complete motor startup, steady-state, and load-variable cycles. Each node outputs real-time three-axis magnetic flux density vector data (Bx, By, Bz). This data is timestamped by the edge processing unit and uploaded to a local storage node. It is then simultaneously pushed to the IoT platform for unified archiving, resulting in a dynamic magnetic field multi-point distribution dataset structured as "node number - time - three-axis magnetic flux density value." The data is formatted in a standard CSV format and includes the physical location code for data collection and sensor status identification.

[0144] Step S312: Calculating the time domain variation gradient information of magnetic flux density based on the multi-point dynamic magnetic flux density distribution data;

[0145] In this embodiment of the present invention, the three-axis magnetic flux density time series data collected in step S311 is grouped by node number. For each node, the time-domain magnetic flux density variation curve for 10 seconds is extracted. The rate of change of magnetic flux density within each 0.1-second interval is calculated using the first-order difference method to form a magnetic flux density gradient time series. To enhance sensitivity to local sudden changes, median smoothing is introduced to remove data spikes before performing a second-order difference calculation to obtain second-order magnetic flux density gradient fluctuation data. The gradient fluctuation data corresponding to each node is aligned along the time dimension and integrated into a magnetic flux density time-domain gradient information matrix. The matrix structure dimensions are [number of nodes × time points × gradient components (Bx', By', Bz')]. This information matrix is ​​used in the subsequent time series window analysis.

[0146] Step S313: performing a multi-dimensional time series sliding window analysis on the time domain gradient information of the magnetic flux density to obtain the magnetic field fluctuation vector characteristics;

[0147] In this embodiment of the present invention, based on the magnetic flux density time-domain gradient information matrix obtained in step S312, a fixed window sliding analysis method is used to extract time-dimensional fluctuation characteristics. The sliding window width is set to 1 second, with a step size of 0.5 seconds. This means that 19 sliding window segments can be extracted from 10 seconds of data. Within each window segment, statistical indicators such as the mean, standard deviation, kurtosis, and skewness of the gradient fluctuation are calculated for each node dimension. A fast Fourier transform (FFT) is then performed within the window to analyze the main fluctuation frequency components. A window-level feature vector is then constructed for each node, containing multidimensional fluctuation parameters in the time and frequency domains based on the three-axis gradient components. The sliding window feature vectors of all nodes are concatenated according to the window sequence number to form a magnetic field fluctuation vector feature group with the dimensions [window sequence × number of nodes × number of features], which is used for further spectrum-level feature analysis and processing.

[0148] Step S314: performing magnetic field fluctuation spectrum feature analysis based on the magnetic field fluctuation vector feature to obtain magnetic field fluctuation spectrum feature data;

[0149] In this embodiment of the present invention, a spectral analysis method is used to extract the characteristic expression of the magnetic field fluctuation vector feature group obtained in step S313 in the frequency domain. The node fluctuation vector corresponding to each sliding window is weighted and synthesized using three-axis components to obtain a synthetic fluctuation amplitude sequence of the magnetic field changes at each window node. This sequence is then transformed into the frequency domain using a fast Fourier transform algorithm to extract the main frequency components and their energy distribution within the 0 to 5 kHz frequency band. To improve the accuracy of spectrum recognition, the Hilbert-Huang transform (HHT) is introduced to perform a secondary subdivision of the edge energy-intensive frequency band and decompose the local transient frequency drift characteristics. This forms a magnetic field fluctuation spectrum feature data set. The data structure includes multiple dimensional indicators such as the band center frequency, energy peak, frequency drift rate, and bandwidth. The data is archived according to the node and sliding window time dimensions to form a complete magnetic field spectrum distribution table.

[0150] Step S315: Calculating the peak value data of the magnetic field interference frequency band energy according to the magnetic field fluctuation spectrum characteristic data;

[0151] In an embodiment of the present invention, a frequency band energy peak extraction operation is performed on the magnetic field fluctuation spectrum characteristic data generated in step S314. The frequency interval between 1kHz and 3kHz is selected as the frequency band of interest, and the energy spectrum curve in this interval is used to capture the local peak point using the envelope detection method. In order to avoid misjudgment, the nodes in which the frequency band energy abnormal growth occurs more than twice in the same window are marked as "spectrum abnormality nodes". Then, for each abnormal node, the average energy value is calculated in the ±50Hz interval near its abnormal frequency, and compared with its static baseline (i.e., the spectrum average value within 3 seconds before startup) to obtain the magnetic field interference frequency band energy peak data. The data is stored in the form of a four-tuple [node number-window number-abnormal frequency-peak energy] for subsequent interference amplification analysis.

[0152] Step S316: determining the difference in abnormal magnetic field interference amplification according to the magnetic field interference frequency band energy peak data and the magnetic field fluctuation spectrum characteristic data;

[0153] In an embodiment of the present invention, based on the magnetic field interference frequency band energy peak data and the magnetic field fluctuation spectrum characteristic data obtained in step S315, the energy amplification ratio of each frequency band is calculated and the amplification difference is evaluated. The specific processing flow is to match the current peak energy at each node and each window frequency with its reference baseline value, and form an amplification sequence according to the energy ratio. Then, the maximum amplification frequency point and its corresponding frequency drift range are calculated according to the node dimension. A sliding comparison algorithm is introduced to identify continuous window segments with sudden amplification in the same frequency band, and an abnormal magnetic field interference amplification difference data set is constructed, which includes indicators such as frequency band number, energy amplification ratio, frequency drift change, and window interval length, and marks high-difference areas for subsequent weighted processing.

[0154] Step S317: performing interference growth factor weighting processing based on the difference in abnormal magnetic field interference amplification to obtain abnormal interference growth factor weighted data;

[0155] In an embodiment of the present invention, the abnormal magnetic field interference amplification difference data determined in step S316 is used to construct an interference growth factor using a weighted gain processing method. The frequency band energy amplification ratio is used as the main weight benchmark, and the superimposed frequency drift rate and the abnormal duration window length are used as auxiliary weight factors. The growth factor strength of each frequency point is calculated through a linear weighted combination. In order to reflect the importance of the frequency band, a spatial interference weight correction mechanism based on the geographical distribution of nodes is introduced to adjust the weighted results so that the growth factor proportion of sensor nodes close to the control cabinet or signal port is increased. The weighted data of the abnormal interference growth factor is output, with the node number and frequency band number as index fields. Each data includes a weighted value, weight composition details and a timestamp, providing an input basis for subsequent risk trend estimation.

[0156] Step S318: estimating the growth trend of the electromagnetic interference risk of the motor according to the weighted data of the abnormal interference growth factor and the difference in the abnormal magnetic field interference increase.

[0157] In this embodiment of the present invention, a criterion for determining the interference risk growth trend is constructed based on the weighted data of the abnormal interference growth factor generated in step S317 and the abnormal magnetic field interference increase difference data generated in step S316. The weighted data for each node is sorted by time window, and the rate of change of the weighted values ​​for consecutive growth window segments is extracted. Trend superposition calculations are performed by combining the corresponding frequency drift direction (increase or decrease) and energy gradient direction (increase or decrease). After normalization, the results of all nodes are integrated into an overall electromagnetic interference risk growth trend curve for the motor. The breakpoints on the trend curve reflect the time when the interference energy transitions from local to global. The risk growth rate curve is used to analyze the rhythm of interference intensification. Finally, an electromagnetic interference risk growth trend data packet is exported in CSV format, including fields such as node number, window number, growth amplitude, trend slope, and cumulative growth value. This data packet serves as a key input parameter for current waveform distortion detection in step S32.

[0158] It is particularly important that step S32 includes the following steps:

[0159] Step S321: determining the electrical stress growth situation based on the electromagnetic interference risk growth trend of the motor and the dynamic magnetic field imbalance situation of the motor;

[0160] In an embodiment of the present invention, a real-time sampling module based on current transformers (CTs) and voltage transformers (PTs) is deployed in an industrial-grade high-frequency induction motor operating environment. This module, in conjunction with an EMI signal analysis unit in an IoT edge node, collects high-frequency interference signals within the 150kHz to 30MHz frequency band during operation. A swept-frequency antenna coupling device is used to introduce the signals, and harmonic signals related to the main shaft rotation period are extracted through joint time-domain and frequency-domain analysis. Combined with an electromagnetic disturbance change rate trend identification unit, the module compares changes in interference frequency, amplitude, and the proportion of spurious components over the past 30 days to determine the interference risk growth trend. Simultaneously, three-dimensional vector change data of the magnetic flux density between the stator and rotor is collected, and a magnetic field vector balance criterion identification program is run in an edge processor to quantify dynamic magnetic field imbalances caused by structural eccentricity, winding asymmetry, and local core saturation, outputting an imbalance intensity index. The EMI risk trend index is then combined with the magnetic field imbalance intensity index to generate an electrical stress growth level through rule-based logical judgment. The level is quantified from 0 to 4, with 4 indicating a significant increase in stress. This level serves as an input parameter for estimating the degree of insulation material aging.

[0161] Step S322: estimating the degree of aging of the motor insulation material according to the electrical stress growth;

[0162] In an embodiment of the present invention, the electrical stress growth level obtained in step S321 is combined with the motor insulation structure type and material performance parameters to deduce the insulation aging aggravation trend. An insulation performance database has been established in the early stage of motor production, which contains the dielectric strength degradation rate, breakdown voltage variation law and temperature rise tolerance critical data of materials such as polyesterimide, polyimide, and mica tape. After the currently collected electrical stress level is matched with the database, the interpolation matching module retrieves the corresponding insulation degradation rate range. The withstand voltage failure time estimation program is run in the edge node, the time step is set, and the dielectric strength degradation process of the material under the current stress conditions is calculated step by step. The aging stage correction is performed in combination with the thermal shock and partial discharge event tags in the historical working conditions. An insulation material aging aggravation degree index is formed, which expresses the degree of insulation performance degradation as a percentage, ranging from 0% to 100%, and is output to the buffer area called in step S323.

[0163] Step S323: estimating the growth trend of the motor chip failure rate based on the degree of increasing aging of the motor insulation material and the growth of electrical stress;

[0164] In an embodiment of the present invention, the percentage of insulation aging aggravation formed in step S322 and the electrical stress level in step S321 are used to jointly construct a dual-parameter failure rate assessment link for chip fault warning. For the IGBT driver module, power modulation chip and control unit signal acquisition chip, the static voltage offset of the chip pin, the operating temperature change rate and the shell-to-ground leakage voltage and other physical data are read in real time through the multi-channel chip probe board. The fault trend tracking module compares the parameter changes over multiple days. When key parameters such as the temperature rise rate exceed 0.8°C / hour for three consecutive days, it is marked as an accelerated degradation state. This state is combined with the insulation aging percentage and the electrical stress level to form an index, and a rule base containing a three-dimensional threshold matrix and the corresponding failure rate growth rate is called to generate a chip failure rate growth trend. The result is output as a percentage of failure rate change per hour.

[0165] Step S324: determining an abnormal short circuit condition of the motor circuit board based on the motor chip failure rate growth trend and the motor chip failure rate growth trend;

[0166] In this embodiment of the present invention, the chip failure rate growth rate obtained in step S323 is combined with data from a circuit board surface temperature acquisition array and micro-arc detection array deployed on-site to comprehensively assess potential short circuit trends within the circuit board. Using an infrared temperature array with an accuracy of at least ±0.2°C, the unit area of ​​the circuit board is scanned at 1-second intervals to generate a hotspot migration map. Simultaneously, an ultraviolet micro-arc detector is activated to monitor for arcing events. When the abnormal temperature migration rate on the same path exceeds 5°C / minute and two or more micro-arc events occur consecutively, an abnormal short circuit warning is triggered. The short circuit warning signal is cross-compared with the chip failure rate growth trend. If the conditions of "medium-high failure rate + high hotspot migration + arcing events" are met, the circuit board abnormal short circuit determination module outputs a three-level short circuit risk level: "0" for normal, "1" for potential short circuit, and "2" for a clear short circuit trend. The determination result is written to the data bus for subsequent use.

[0167] Step S325: Detect the motor operating current waveform distortion according to the abnormal short circuit condition of the motor circuit board.

[0168] In this embodiment of the present invention, after the short-circuit status is determined to be "1" or "2" in step S324, current waveform distortion detection is automatically initiated. Using a current sampling module with 24-bit accuracy and a 100kHz sampling frequency, high-frequency sampling of the three-phase current signal is performed and transmitted in real time to an edge processing node via a high-speed bus. A Fourier order analysis and odd harmonic ratio calculation module is configured within the node to decompose and analyze the fundamental wave of the current waveform and the proportion of each harmonic. When the amplitude of a specific harmonic level (such as the 5th, 7th, and 11th) exceeds 5% of the rated current and the harmonic corresponds to a short-circuit path, the current distortion is determined to be caused by an abnormal circuit board short circuit. Combined with motor speed and load change data, the current distortion is verified to be synchronized with the load disturbance and to eliminate external interference. Three-phase current distortion indicators, including harmonic distortion percentage, interphase imbalance, and zero-sequence current amplitude, are output and input into the diagnostic module as abnormality judgment criteria, forming a complete abnormality traceability chain.

[0169] It is particularly important that step S33 includes the following steps:

[0170] Step S331: detecting abnormal heat growth of the motor winding based on the motor running current waveform distortion;

[0171] In this embodiment of the present invention, in an IoT environment, a thermal infrared array sensor deployed on the exterior of the motor winding collects surface temperatures across various areas of the motor winding. This is combined with a thermocouple array embedded within the motor to collect temperature rise curves per unit time at key winding locations (such as the notch, mid-section, and end-of-winding nodes). Simultaneously, using the motor current waveform distortion parameters extracted in the previous step, the distortion frequency distribution, harmonic amplitude, and crest factor within the current operating cycle are selected as comparison benchmarks. The current distortion data is then compared with the temperature growth rate at each measuring point. Using a method for analyzing the synchronization of thermal changes, abnormal spikes in the temperature curve at each measuring point are extracted to identify any periods of abnormal heat growth during periods of increased current distortion. Specifically, the thermocouple data acquisition frequency is set to 10 times per second, and the heat growth rate of each area is synchronously collected at eight measuring points located in different winding sections. Based on the comparison of thermistor array and thermocouple data, a three-dimensional heat conduction map is further constructed through the heat flow tracking algorithm, the energy density change value of the heat concentration area inside the winding is extracted, and the thermal growth slope index is calculated, finally obtaining the abnormal heat growth status of the motor winding under the current operating state.

[0172] Step S332: determining a local overheating condition of the motor based on the abnormal heat growth condition of the motor winding;

[0173] In an embodiment of the present invention, based on the heat growth slope extracted in step S331 and the heat growth data of each collection point, a dynamic area determination method is adopted to set the historical temperature rise threshold of the winding area, and the real-time heat data is compared with the average value of the historical normal heat growth rate. The area where the temperature rise rate exceeds 120% is defined as a thermal anomaly area. The area, temperature peak and duration of each abnormal area are integrated to form a local overheating index group, and the heat accumulation phenomenon is evaluated by the ratio of the thermal anomaly duration to the temperature peak. The distributed thermal field monitoring system is used to track high-temperature islands using multi-frame thermal images, and the effective local overheating point is determined by combining the trend of heat flow direction changes. According to the temperature growth amplitude, spatial scale and duration period, a complete local overheating situation data set is formed as the basis for thermal deformation determination.

[0174] Step S333: Identifying the degree of uneven thermal expansion of the motor based on the local overheating of the motor and the abnormal heat growth of the motor windings;

[0175] In an embodiment of the present invention, after obtaining the local overheating area in step S332, the thermal anomaly area is used as the initial area for thermal expansion deformation analysis. The multi-channel motor housing strain sensor is used to collect the micro-deformation data of the corresponding position of the winding every 30ms. The local thermal expansion offset model is established in combination with the thermal diffusion radius. The strain difference is standardized to generate a distribution diagram of the winding thermal response amplitude. The stator slot laser displacement sensor array is synchronously connected to measure the inner diameter micro-change in real time. The thermal expansion non-uniformity rate is extracted through multi-point linear fitting. The thermal center offset trajectory and strain field non-uniformity are combined to extract the direction of the thermal deformation vector and evaluate the directional consistency. If the thermal expansion direction of the hot spot area deviates from the axis by more than 5° per unit time and the strain amplitude difference exceeds 10 microstrains, it is determined that there is non-uniform thermal expansion. Finally, the thermal response difference and direction offset are summarized to output the motor thermal expansion non-uniformity degree parameter.

[0176] Step S334: determining the motor operation amplitude drift condition based on the degree of uneven thermal expansion of the motor.

[0177] In an embodiment of the present invention, based on the determined degree of thermal expansion non-uniformity, a high-precision rotation angle decoder and inertial measurement unit (IMU) mounted on the motor shaft end are used to monitor mechanical axis offset in real time. The change in the shaft rotation radius is measured when the motor is operating at full load, and the axial offset distribution function caused by thermal expansion is calculated based on the strain peak in the thermal expansion region. Using the difference between the end displacement and the rated operating radius as the drift reference, an axis deviation model is established through three-dimensional mechanical deflection analysis. The difference between the maximum and minimum rotation radii within one spindle cycle is simultaneously monitored. If the tolerance exceeds ±0.1mm, it is determined to be "motor operating amplitude drift." The directionality of thermal expansion, the non-uniform strain difference, and the degree of axis deviation are combined to generate an amplitude drift index for use in vibration diagnosis and fault prediction.

[0178] Preferably, step S4 includes the following steps:

[0179] Step S41: predicting damage to the motor's electronic components based on the motor's operating frequency oscillation growth;

[0180] In this embodiment of the present invention, motor operating frequency oscillation data is collected in an IoT environment. This data is monitored in real time by high-precision frequency sensors deployed at key locations on the motor. These sensors accurately record minute fluctuations and oscillation characteristics of the motor's operating frequency. The collected frequency oscillation data undergoes time-domain preprocessing, including noise removal filtering and baseline drift correction, to ensure data validity and accuracy. Subsequently, a frequency oscillation growth trend analysis module segments the time series of the frequency oscillation data and calculates the oscillation amplitude and frequency change rate within each time segment. A sliding window technique is used to gradually shift the time window, updating the oscillation growth trend curve in real time. The oscillation growth trend is compared with historical normal operation data using a threshold determination method. Multiple critical thresholds for electronic component damage are preset, and the current degree of damage to the motor's electronic components is determined based on the frequency oscillation amplitude and growth rate. Damage status parameters for the motor's electronic components are output, including a damage level score, an estimated damaged location, and a damage trend time series, for use in subsequent steps. This data collection and processing ensures that damage status predictions are based on precise physical quantities of frequency oscillation, eliminating subjective judgment based on human experience.

[0181] Step S42: detecting the accumulated error of the motor operation according to the damage status of the motor electronic components;

[0182] In an embodiment of the present invention, based on the motor electronic component damage status parameters obtained in step S41, the focus is on detecting the accumulation of motor operating errors caused by damaged electronic components. A motor operating error detection module is used, which combines the motor's actual operating speed, torque, and position feedback data to calculate the error between the actual operating state and the set target value. Error data is collected in real time by high-sampling rate sensors and stored in edge computing nodes to ensure that the error data is complete and without delay. Error accumulation is achieved using a numerical integration method, combined with a time-weighted strategy to calculate the cumulative sum of errors within a certain time range, highlighting long-term error trends. The error accumulation data is also combined with auxiliary parameters such as the motor operating environment temperature and voltage fluctuations to perform multi-dimensional correlation analysis to eliminate errors caused by external environmental interference. During the detection process, an error accumulation alarm threshold is set and correlated with the damage level of the electronic components to confirm the inherent connection between error growth and component damage. The motor operation error accumulation parameters are output, including the cumulative error value, error growth rate, and error fluctuation range, providing a quantitative basis for the abnormality detection step.

[0183] Step S43: performing motor abnormality detection based on the accumulated motor operation errors and the damage status of the motor electronic components to obtain motor operation abnormality data;

[0184] In an embodiment of the present invention, motor anomaly detection is performed by comprehensively utilizing the motor operation error accumulation parameters in step S42 and the electronic device damage status parameters in step S41. The two types of parameters are multi-dimensionally fused to construct an abnormal feature vector containing the error accumulation value, error growth rate, damage level, and damage trend. This feature vector is subjected to an abnormality threshold judgment. The abnormality threshold is derived based on historical normal operation status statistics and includes multi-level differentiation criteria covering minor anomalies, moderate anomalies, and severe anomalies. The anomaly detection process adopts a dynamic threshold update mechanism to adjust the threshold range according to real-time data changes to ensure the real-time and sensitivity of the detection. The detection results include an abnormal status label, anomaly severity level, and an abnormal timestamp, which fully describe the current abnormal operating status of the motor. The motor operation abnormality data is stored in a structured data format and uploaded to the central monitoring platform via the Internet of Things communication module to achieve real-time sharing of abnormal information. This step achieves accurate determination of the motor abnormal state through rigorous multi-parameter fusion and dynamic threshold technology, ensuring that the abnormal data reflects the actual operating risks.

[0185] Step S44: performing motor configuration optimization processing on the motor operation data according to the motor operation abnormality data to obtain motor configuration optimization processing data.

[0186] In this embodiment of the present invention, the motor operation anomaly data generated in step S43 is used as input to perform motor configuration optimization, aiming to adjust motor operating parameters and mitigate or correct abnormal conditions. This optimization process is based on a rules engine that includes multiple optimization strategies tailored to different anomaly levels and types. Specific strategies include adjusting the supply voltage, drive frequency, load distribution, and dynamically adjusting cooling system operating parameters. The anomaly data is classified by anomaly type to determine the set of configuration parameters requiring adjustment. Subsequently, the system automatically calculates optimized configuration values ​​based on pre-set configuration optimization rules and current environmental parameters (such as temperature and load). The optimization process is implemented through a feedback loop, which collects results in real time based on optimized motor operation data, determines the optimization effect, and then makes further adjustments, forming a closed-loop control system. The optimized configuration data includes the specific parameter adjustment values, adjustment time points, and adjustment ranges. The optimized configuration data is transmitted to the motor control unit via the Internet of Things communication module, enabling remote, real-time adjustment. This step enables dynamic adjustment of the motor configuration, improving operational stability and extending the life of electronic components, ensuring continuous and stable motor operation.

[0187] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A motor data anomaly detection method based on the Internet of Things, characterized in that: The following steps are involved: Step S1: Acquire motor operation data; Perform motor operation demand overload evaluation based on motor operation data to obtain motor operation demand overload data; Detecting the motor operation load response status based on the motor operation data and the motor operation demand overload data; Step S2: determining the bearing kinetic energy chain coupling decay state according to the motor operation load response condition; Determine the motor structure connection coordination decline trend based on the motor operation load response status; determine the motor operation dynamic magnetic field imbalance status based on the bearing kinetic energy chain coupling decay state and the motor structure connection coordination decline trend; Step S3: estimating the growth trend of the electromagnetic interference risk of the motor according to the dynamic magnetic field imbalance condition of the motor; Determine the oscillation growth of the motor operating frequency based on the growth trend of the motor's electromagnetic interference risk; Step S4: detecting the accumulated motor operation error based on the motor operation frequency oscillation growth condition; Perform motor abnormality detection based on the accumulated motor operation errors to obtain motor operation abnormality data; Motor configuration optimization processing is performed on the motor operation data according to the motor operation abnormality data to obtain motor configuration optimization processing data.

2. The method for detecting motor data anomaly based on the Internet of Things according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: obtaining motor operation data based on the Internet of Things; Step S12: performing motor operation demand overload evaluation based on the motor operation data to obtain motor operation demand overload data; Step S13: performing a motor operation intensity assessment on the motor operation data based on the motor operation demand overload data to obtain motor operation intensity data; Step S14: detecting the motor operation load response status based on the motor operation data, the motor operation intensity data, and the motor operation demand overload data.

3. The method for detecting abnormal motor data based on the Internet of Things according to claim 2, characterized in that: Step S12 includes the following steps: Step S121: performing motor operation demand timing analysis on the motor operation data to obtain motor operation demand timing data; Step S122: Counting the frequency changes of the motor operation demand based on the motor operation demand time series data; Step S123: performing frequency change data preprocessing on the motor operation demand time series data and the motor operation demand frequency change to obtain motor operation demand frequency preprocessing data; Step S124: constructing a motor operation demand fluctuation change graph using the motor operation demand frequency preprocessing data; Step S125: collecting motor operation demand fluctuation growth interval information based on the motor operation demand fluctuation change graph; Step S126: calculating the maximum slope information of the motor operation demand fluctuation according to the motor operation demand fluctuation growth interval information; Step S127: calculating the average slope data of the demand fluctuation growth interval according to the motor operation demand fluctuation growth interval information; Step S128: When the average slope data of the demand fluctuation growth interval exceeds 1.43 times / h² and the maximum slope information of the motor operation demand fluctuation exceeds 8.4 times / h², a motor operation demand overload assessment is performed to obtain motor operation demand overload data.

4. The method for detecting motor data anomaly based on the Internet of Things according to claim 2, wherein: Step S14 includes the following steps: Step S141: extracting the motor operation power surge state based on the motor operation demand overload data and the motor operation intensity data; Step S142: identifying the motor high power demand data based on the motor operating power sudden increase state exceeding 41.5% and the motor operating demand overload data; Step S143: collecting the motor running rated power threshold value according to the motor running data to obtain the motor running rated power threshold value data; Step S144: estimating the motor operating power over-limit condition by using the motor operating rated power threshold data when the motor operating high power demand data exceeds 5.8 kW; Step S145: detecting a nonlinear load growth trend of the motor based on the motor operating power exceeding limit condition; Step S146: Detecting the motor operation load response condition based on the nonlinear load growth trend of the motor operation and the motor operation power exceeding limit condition.

5. The method for detecting abnormal motor data based on the Internet of Things according to claim 1, characterized in that: The determination of the bearing kinetic energy chain coupling decay state in step S2 includes: Detect the motor shaft over-rotation impact load according to the motor operation load response condition; Measure the bearing rotation pressure growth based on the motor shaft over-rotation impact load and the motor operating load response; Determine the motor bearing gear deformation based on the bearing rotation pressure growth; Detect the motor bearing gear meshing conflict condition based on the motor bearing gear deformation condition; Predict the motor bearing gear fracture trend based on the motor bearing gear meshing conflict condition and motor bearing gear deformation condition; Evaluate the motor bearing gear backlash imbalance based on the motor bearing gear cracking trend; Detect abnormal lateral thrust growth of the bearing based on the side clearance imbalance of the motor bearing gear; The bearing kinetic energy chain coupling degradation state is determined based on the abnormal lateral thrust growth of the bearing and the side clearance imbalance of the motor bearing gear.

6. The method for detecting motor data anomaly based on the Internet of Things according to claim 1, characterized in that: Determining the decreasing trend of the motor structure connection coordination in step S2 includes: Detecting the eccentricity of the motor shaft based on the load response of the motor; Determine the rotational symmetry imbalance of the motor shaft according to the eccentric running condition of the motor shaft; Test the unevenness of the motor shaft inertia moment based on the imbalance of the motor shaft's rotational symmetry; Detect the periodic vibration trend of the motor shaft based on the unevenness of the motor shaft inertia moment and the imbalance of the motor shaft rotational symmetry; Detect the increasing looseness of the connection parts based on the periodic vibration trend of the motor shaft; According to the increasing looseness of the connection parts, the motor shaft periodic vibration trend is used to predict the growth trend of the motor structure resonance frequency; The decreasing trend of the motor structure connection coordination is determined based on the growth trend of the motor structure resonance frequency and the unevenness of the motor shaft inertia moment.

7. The method for detecting motor data anomaly based on the Internet of Things according to claim 1, characterized in that: Determining the dynamic magnetic field imbalance condition of the motor in step S2 includes: The degree of reduction in the motor's operating force transmission capability is tested based on the bearing kinetic energy chain coupling decay state and the decreasing trend of the motor structure connection coordination; Determine the reduction in bearing centripetal stability based on the reduction in the motor's operating force transmission capability; Predicting rotor ellipticity based on bearing centripetal stability degradation; Determine the motor rotor position disturbance condition based on the bearing centripetal stability degradation and rotor rotation elliptical phenomenon; Monitor the rotor-stator constant distance fluctuation according to the motor rotor position disturbance; Estimate the air gap flux density disturbance based on the rotor-stator constant spacing fluctuation; Calculate the degree of deviation of the motor's multi-pole electromagnetic pull according to the air gap magnetic flux density disturbance condition; Predict the center deviation trend of the electromagnetic force vector based on the motor's multi-pole electromagnetic pull deviation and air gap flux density disturbance; The dynamic magnetic field imbalance condition of the motor is determined based on the center deviation trend of the electromagnetic force vector and the air gap magnetic flux density disturbance condition.

8. The method for detecting motor data anomaly based on the Internet of Things according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: estimating the growth trend of the electromagnetic interference risk of the motor according to the dynamic magnetic field imbalance condition of the motor; Step S32: detecting the motor operating current waveform distortion based on the motor electromagnetic interference risk growth trend and the motor operating dynamic magnetic field imbalance; Step S33: determining the motor operating amplitude drift condition based on the motor operating current waveform distortion condition; Step S34: determining the oscillation growth condition of the motor operating frequency according to the motor operating amplitude drift condition and the motor operating current waveform distortion condition.

9. The method for detecting motor data anomaly based on the Internet of Things according to claim 8, characterized in that: Step S31 The following steps are involved: Step S311: collecting multi-point dynamic magnetic flux density distribution data according to the dynamic magnetic field imbalance condition of the motor; Step S312: Calculating the time domain variation gradient information of magnetic flux density based on the multi-point dynamic magnetic flux density distribution data; Step S313: performing a multi-dimensional time series sliding window analysis on the time domain gradient information of the magnetic flux density to obtain the magnetic field fluctuation vector characteristics; Step S314: performing magnetic field fluctuation spectrum feature analysis based on the magnetic field fluctuation vector feature to obtain magnetic field fluctuation spectrum feature data; Step S315: Calculating the peak value data of the magnetic field interference frequency band energy according to the magnetic field fluctuation spectrum characteristic data; Step S316: determining the difference in abnormal magnetic field interference amplification according to the magnetic field interference frequency band energy peak data and the magnetic field fluctuation spectrum characteristic data; Step S317: performing interference growth factor weighting processing based on the difference in abnormal magnetic field interference amplification to obtain abnormal interference growth factor weighted data; Step S318: estimating the growth trend of the electromagnetic interference risk of the motor according to the weighted data of the abnormal interference growth factor and the difference in the abnormal magnetic field interference increase.

10. The method for detecting motor data anomaly based on the Internet of Things according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: predicting damage to the motor's electronic components based on the motor's operating frequency oscillation growth; Step S42: detecting the accumulated error of the motor operation according to the damage status of the motor electronic components; Step S43: performing motor abnormality detection based on the accumulated motor operation errors and the damage status of the motor electronic components to obtain motor operation abnormality data; Step S44: performing motor configuration optimization processing on the motor operation data according to the motor operation abnormality data to obtain motor configuration optimization processing data.

Images