Artificial intelligence-based asynchronous motor fault management method and system

Abstract

The present application relates to the technical field of intelligent motor state monitoring, in particular to an asynchronous motor fault management method and system based on artificial intelligence, comprising the following steps: eliminating inferior vibration signals, extracting fault features, analyzing time-frequency consistency and stability, evaluating risk levels, screening continuous abnormal equipment and writing into hidden trouble tracking blocks, comparing original records to identify evolution trend differences, marking hidden trouble sections, and outputting asynchronous motor overall health evaluation state sets; in the present application, signal integrity screening and signal-to-noise ratio judgment mechanisms are introduced to eliminate unqualified monitoring data, guarantee feature extraction based on high-quality signals, obtain stability values by combining frequency domain and time domain consistency analysis, build risk score and priority relationship, realize dynamic screening and early warning of high-risk equipment, improve the accuracy and trend judgment ability of fault identification by matching abnormal feature continuity tracking and operation records, and strengthen the real-time and response efficiency of health state evaluation.

Description

Technical Field

[0001] This invention relates to the field of intelligent motor condition monitoring technology, and in particular to an asynchronous motor fault management method and system based on artificial intelligence. Background Technology

[0002] The field of intelligent motor condition monitoring technology involves real-time monitoring and fault diagnosis of motor operating status. This ensures that abnormalities are detected promptly during motor system operation, allowing for necessary preventative or corrective measures to be taken, thereby improving motor reliability and safety. Core aspects of this field include motor fault detection, vibration analysis, current monitoring, temperature monitoring, and the assessment and prediction of motor health status using technologies such as artificial intelligence. Furthermore, it involves using sensor data acquisition, signal processing, data analysis, and machine learning technologies to achieve real-time monitoring and intelligent diagnosis of motor status. Traditional AI-based asynchronous motor fault management methods utilize artificial intelligence technology for fault detection and management. This method primarily involves collecting motor operating data, such as current, voltage, temperature, and vibration information, and using AI algorithms for analysis and processing to identify the type and location of motor faults. Traditional methods rely on rule or model establishment, identifying potential faults through feature extraction and data classification. Data processing techniques employed include data cleaning, feature selection, and fault mode recognition.

[0003] Existing technologies for monitoring and managing the condition of asynchronous motors rely on rule-based and model-based artificial intelligence algorithms for feature extraction and classification. This approach is prone to inaccurate feature extraction when the quality of the raw data is unstable or when there is interference or noise. Furthermore, the lack of preprocessing for signal quality leads to the inclusion of invalid or abnormal information in the monitoring data, interfering with fault identification results and reducing accuracy. In terms of operational status assessment, existing technologies fail to effectively evaluate the stability changes of key features in the frequency and time domains, lacking the ability to dynamically identify and respond to motor risk levels. This results in an inflexible adjustment of monitoring priorities for high-risk equipment, delaying the opportunity to address potential hazards. In addition, existing methods cannot achieve logical correlation analysis between operational data and historical health records, leading to certain biases and lags in assessing the evolution trend of motor health status. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide an asynchronous motor fault management method based on artificial intelligence, comprising the following steps:

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an asynchronous motor fault management method based on artificial intelligence, comprising the following steps:

[0006] S1: Based on the vibration sensor, collect the operation monitoring signal of the asynchronous motor, filter the original data items with abnormal signal strength fluctuation or signal-to-noise ratio below the preset range, determine whether the signal meets the integrity and quality requirements, remove the signal segments that do not meet the requirements, and obtain the effective operation signal recognition ratio.

[0007] S2: Based on the effective operating signal identification ratio, extract the key feature vectors of motor faults from the operating monitoring signals, screen the signal features with sparse feature energy distribution and low matching degree with the health standard model, analyze the consistency of key features in the frequency domain and time domain, and obtain the stability value of motor operating status.

[0008] S3: Based on the stability value of the motor operating status, extract the risk level score and fault monitoring priority configuration rules, filter out devices with risk levels higher than the set threshold, adjust the monitoring priority mapping relationship, and generate a fault monitoring priority index table;

[0009] S4: Based on the motor ID marked in the fault monitoring priority index table, extract the original operation record data, filter motor equipment with more than a set value of consecutive abnormal feature manifestations, write the motor equipment information into the fault hazard tracking block, and obtain the motor fault management hazard tracking and handling list.

[0010] As a further aspect of the present invention, the effective operating signal identification ratio includes signal amplitude compliance rate, spectrum distribution deviation, and sampling period integrity matching degree; the motor operating status stability value includes characteristic frequency distribution uniformity, fault feature point matching degree, and operating parameter recording continuity; the fault monitoring priority index table balances equipment risk level identification, monitoring frequency update item, and key motor node mark; and the motor fault management hidden danger tracking and handling list includes hidden danger motor number, characteristic abnormality frequency statistics, and hidden danger block index.

[0011] As a further aspect of the present invention, the step of determining the effective operating signal recognition ratio specifically includes:

[0012] S101: The operation monitoring signals are collected by a sensor group deployed at a predetermined position on the stator housing of the asynchronous motor, the instantaneous amplitude and background noise intensity information of each signal are extracted, the peak factor and noise interference of the signal are checked, signal segments that exceed the preset interference range are eliminated, and the data are summarized and statistically analyzed to obtain the number of signal samples that meet the quality requirements.

[0013] The operation monitoring signals include stator current signal, rotor vibration signal, and winding temperature signal;

[0014] S102: Based on the signal-to-noise ratio and waveform integrity information of each sample in the number of signal samples that meet the quality requirements, compare it with the preset signal quality standard range, eliminate abnormal signals outside the range, count the number of remaining signal samples and calculate its proportion to the total number of collected signal samples to obtain the effective operating signal recognition ratio.

[0015] As a further aspect of the present invention, the step of determining the stability value of the motor operating state specifically includes:

[0016] S201: Based on the effective operating signal identification ratio, the key feature vectors of the asynchronous motor in the operating monitoring signal are extracted using fast Fourier transform and wavelet packet decomposition. Signal features with feature energy density lower than the preset level or with cross-correlation coefficient with the health standard model are selected. The distribution pattern and matching degree of the feature vectors within the predetermined time window are compared to obtain an abnormal feature vector record list.

[0017] S202: Based on the abnormal feature vector record list, identify the degree of dispersion of the motor operating state per unit time, and determine the degree of deviation between the motor operating state and the standard state by calculating the Euclidean distance between the real-time feature vector and the health standard model, and generate a motor operating state stability measure.

[0018] S203: Based on the motor operating state stability metric and combined with the preset motor operating stability grading standard, perform stability correlation analysis, and calculate the stability offset comparison value after comparing the real-time operating state stability metric with the preset stability threshold to obtain the motor operating state stability value.

[0019] As a further aspect of the present invention, the key feature vector includes the stator current characteristic frequency component, the rotor bar breakage characteristic frequency offset, and the bearing vibration envelope feature.

[0020] As a further aspect of the present invention, the steps for establishing the fault monitoring priority index table are as follows:

[0021] S301: Based on the stability value of the motor's operating status, extract the real-time risk level score and monitoring priority dynamic configuration rules of the asynchronous motor, perform data matching by motor equipment ID, filter motor records with risk level scores higher than the current monitoring priority, and obtain the risk motor statistics;

[0022] S302: Based on the risk motor statistics, call the preset monitoring priority mapping algorithm to identify motor records with priority lag or mismatch, extract the corresponding motor equipment number and the working condition block number, and generate a monitoring priority mapping adjustment sequence.

[0023] S303: Call the motor equipment number and operating condition block number in the monitoring priority mapping adjustment sequence, replace the corresponding monitoring priority mapping field in the database, identify the mapping relationship between the motor equipment number and the risk level score, and generate a fault monitoring priority index table.

[0024] As a further aspect of the present invention, the steps for establishing the motor fault management hidden danger tracking and handling list are as follows:

[0025] S401: Call the motor ID marked in the fault monitoring priority index table, locate the original operation record data in the distributed storage, extract the feature spectrum of the motor in the target frequency band, and identify the frequency position and amplitude deviation of the abnormal features by comparing the difference between the abnormal frequency components in the feature spectrum and the health standard model to obtain the abnormal feature sequence position information.

[0026] S402: Based on the abnormal feature sequence location information, sort the feature spectrum by motor equipment ID in time series, identify the cumulative number of times abnormal features appear in the continuous monitoring period, compare it with the preset abnormal frequency threshold, filter motor equipment with the cumulative number of times exceeding the threshold, extract the associated operating feature status information, remove data that does not meet the continuous abnormal conditions, and generate a continuous abnormal motor status sequence.

[0027] S403: Call the continuous abnormal motor state sequence, match the structured information of the original signal acquisition, write all motor equipment information identified as continuously abnormal into the fault hazard tracking block, collect the motor equipment ID and the corresponding abnormal feature description, and obtain the motor fault management hazard tracking and handling list.

[0028] As a further aspect of the present invention, the method further includes step S5:

[0029] S5: Based on the operation records of the corresponding motor equipment in the motor fault management hidden danger tracking and handling list, identify the logical correlation between the current monitoring results and the original records, determine the difference in the evolution trend between the motor operation status and the original health record, mark the difference results to the hidden danger section, and output the overall health assessment status set of the asynchronous motor.

[0030] The overall health assessment status set of the asynchronous motor includes fault evolution trend difference identifiers, abnormal motor mark numbers, and monitoring result comparison index items.

[0031] As a further aspect of the present invention, the steps for the overall health assessment state set of the asynchronous motor are as follows:

[0032] S501: Call the original operation record of the corresponding motor equipment in the motor fault management hidden danger tracking and handling list, extract the motor operation feature vector, identify the numerical correlation between the current monitoring result and the original record, organize the correspondence between the operation feature, monitoring judgment result and the original record in chronological order, and generate a motor status association binding table;

[0033] S502: Based on the motor status association binding table, compare the real-time operating characteristics with the health benchmark information in the original record, check whether the feature offset between the two exceeds the preset safety evolution range, mark the abnormal records that exceed the range to the hidden danger section data field, and obtain the motor status difference marking result.

[0034] S503: Based on the motor sequence of the marked hidden danger sections in the motor status difference marking results, summarize all abnormal marking statuses by motor equipment ID, collect the associated management fields, and embed the summarized information into the structured metadata of the original operation record to obtain the overall health assessment status set of the asynchronous motor.

[0035] An AI-based asynchronous motor fault management system includes:

[0036] The signal verification module collects operation monitoring signals of the asynchronous motor from multiple sensors, filters raw data entries with abnormal signal strength fluctuations or signal-to-noise ratios below a preset range, determines whether the signal meets the integrity requirements, removes signal segments that do not meet the requirements, and obtains the effective operation signal recognition ratio.

[0037] The feature analysis module, based on the effective operating signal recognition ratio, uses time-frequency analysis to extract key feature vectors of the asynchronous motor in the operating monitoring signal, calculates the matching degree between the feature energy distribution and the health standard model, identifies sparse signal segments of feature distribution, analyzes the stability of the motor operating state, and establishes the stability value of the motor operating state.

[0038] The priority dynamic adjustment module, based on the motor operating status stability value, compares the motor risk level score with the monitoring priority configuration rules, filters motor equipment records with risk levels higher than the current priority, adjusts the monitoring mapping relationship value, writes the update result into the monitoring priority field, and constructs a fault monitoring priority index table.

[0039] The hidden danger identification and tracking module, based on the fault monitoring priority index table, locates the risk motor ID, extracts the associated original operation records and feature spectrum, compares the real-time features with the health standard model, identifies the abnormal state within the continuous monitoring period, filters motor equipment with an abnormal cumulative number of times greater than a preset threshold, writes them into the fault hidden danger tracking block, and generates a motor fault management hidden danger tracking and handling list.

[0040] The trend assessment and labeling module, based on the motor fault management hidden danger tracking and handling list, extracts the associated monitoring results and original records, compares the trend difference between the motor operating status and the original baseline, labels the hidden danger status position in the data structure, and obtains the overall health assessment status set of the asynchronous motor.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0042] This invention effectively eliminates monitoring data that does not meet quality requirements by introducing a signal integrity screening and signal-to-noise ratio judgment mechanism, ensuring that feature extraction is based on high-quality signals for analysis. By identifying the consistency of key features in the frequency and time domains, stability values ​​are obtained to reflect the changing trend of operating status. By constructing a risk level scoring and priority mapping relationship, dynamic screening and priority processing of high-risk equipment are achieved. By tracking the continuity of abnormal features, the ability to identify and handle potential faults is strengthened. By using the logical matching of operating records and historical health data, the ability to accurately judge the evolution trend of operating status and locate hidden dangers is enhanced. Dynamic assessment and graded output of motor health status are achieved, effectively improving the accuracy, real-time performance, and response capability of fault monitoring. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a schematic diagram of the steps of the present invention;

[0045] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0046] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0047] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0048] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0049] Figure 6 This is a detailed schematic diagram of S5 of the present invention;

[0050] Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0051] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0052] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0053] Please see Figure 1 This invention provides an artificial intelligence-based method for managing asynchronous motor faults, comprising the following steps:

[0054] S1: Based on the vibration sensor, collect the operation monitoring signal of the asynchronous motor, filter the original data items with abnormal signal strength fluctuation or signal-to-noise ratio below the preset range, determine whether the signal meets the integrity and quality requirements, remove the signal segments that do not meet the requirements, and obtain the effective operation signal recognition ratio.

[0055] S2: Based on the effective operation signal identification ratio, extract the key feature vectors of motor faults from the operation monitoring signals, screen the signal features with sparse feature energy distribution and low matching degree with the health standard model, analyze the consistency of key features in the frequency domain and time domain, and obtain the stability value of motor operation status.

[0056] S3: Based on the stability value of motor operation status, extract the risk level score and fault monitoring priority configuration rules of asynchronous motor, compare with the risk level mapping relationship, filter motor equipment with risk level higher than the set threshold, adjust the monitoring priority mapping relationship, and generate a fault monitoring priority index table.

[0057] S4: Based on the motor ID marked in the fault monitoring priority index table, extract the original operation record data, filter motor equipment with a number of consecutive abnormal features exceeding the set value, write the motor equipment information into the fault hazard tracking block, and obtain the motor fault management hazard tracking and handling list.

[0058] S5: Based on the operation records of the corresponding motor equipment in the motor fault management hidden danger tracking and handling list, identify the logical correlation between the current monitoring results and the original records, determine the difference between the evolution trend of the motor operation status and the original health record, mark the difference results to the hidden danger section, and output the overall health assessment status set of the asynchronous motor.

[0059] The effective operating signal recognition ratio includes signal amplitude compliance rate, spectrum distribution deviation, and sampling period integrity matching degree. The motor operating status stability value includes characteristic frequency distribution uniformity, fault characteristic point matching degree, and operating parameter recording continuity. The fault monitoring priority index table includes equipment risk level identification, monitoring frequency update item, and key motor node mark. The motor fault management hidden danger tracking and handling list includes hidden danger motor number, characteristic abnormality frequency statistics, and hidden danger block index. The asynchronous motor overall health assessment status set includes fault evolution trend difference identification, abnormal motor mark number, and monitoring result comparison index item.

[0060] Please see Figure 2 The specific steps for effectively operating the signal recognition ratio are as follows:

[0061] S101: The operation monitoring signals are collected by a sensor group deployed at a predetermined position on the stator housing of the asynchronous motor, the instantaneous amplitude and background noise intensity information of each signal are extracted, the peak factor and noise interference of the signal are checked, signal segments that exceed the preset interference range are eliminated, and the data are summarized and statistically analyzed to obtain the number of signal samples that meet the quality requirements.

[0062] Operation monitoring signals include stator current signal, rotor vibration signal, and winding temperature signal;

[0063] A distributed sensor array deployed at predetermined physical nodes on the stator housing of the asynchronous motor performs high-frequency data acquisition. Hall effect current sensors acquire analog stator three-phase current, piezoelectric accelerometers pick up rotor axial and radial mechanical vibration waveforms, and embedded PT100 platinum resistance thermometers read real-time temperature data of windings and bearing housings. The acquisition card converts analog signals into digital sequences at a 20,000 Hz sampling rate and transmits them to the edge computing unit. An adaptive filtering algorithm is called to perform denoising processing on the original signals. Kalman filtering is used to eliminate Gaussian white noise for the stator current signal, and wavelet threshold denoising technology is used to remove environmental background noise for the vibration signal. The instantaneous amplitude of each signal is extracted and the background noise intensity is measured. The background noise intensity is taken as the root mean square value of the signal amplitude obtained by long-term statistical monitoring under motor shutdown and power-on or no-load steady state as the reference noise floor. The signal crest factor is calculated as maximum instantaneous amplitude / effective signal value, and the noise interference level is calculated as high-frequency random component energy / total energy. A preset interference range is set, based on the statistical distribution of signal fluctuations under normal operating conditions in the historical database, and the upper and lower limits are determined using the principle of three times the standard deviation. Data segments that exceed the dynamic range or have abnormally high crest factors without corresponding load change records are marked as interfered segments and removed. The number of remaining valid signal samples is then summarized. For example, in the sampling segment, the maximum instantaneous amplitude of the stator current signal is 15 amperes, the effective signal value is 10 amperes, the average background noise energy is 5 joules, and the current total signal energy is 100 joules; the crest factor is 15 / 10 = 1.5, and the noise interference level is 5 / 100 = 0.05. The results show that 1.5 < 1.8 and 0.05 < 0.1, which is considered to meet the quality requirements. Based on the statistical data of 50 asynchronous motors of the same model undergoing 500 hours of full-condition cyclic testing, the upper limit of the preset interference range for the crest factor is set to 1.8 to ensure that only high-quality signal samples are retained for subsequent processes.

[0064] S102: Based on the signal-to-noise ratio and waveform integrity information of each sample in the number of signal samples that meet the quality requirements, compare it with the preset signal quality standard range, eliminate abnormal signals outside the range, count the number of remaining signal samples and calculate its proportion to the total number of acquired signal samples to obtain the effective operating signal recognition ratio.

[0065] A refined quality assessment is performed on signal samples that meet the quality requirements. The effective power and noise power of each retained sample are obtained, and the signal-to-noise ratio (SNR) in dB is calculated as 10 × log10 (effective power / noise power). A waveform integrity evaluation mechanism is introduced, and the cross-correlation coefficient between a complete cycle waveform of the real-time acquired signal and a standard sine wave or a reference healthy waveform is calculated to quantify the degree of waveform distortion. A cross-correlation coefficient close to 1 indicates high waveform integrity. The calculated SNR and waveform integrity values ​​are compared one by one with the preset signal quality standard range. The standard range requires that the SNR > a specific dB value and the waveform integrity coefficient > a specific limit value. Signals outside the range are considered abnormal or low-quality signals and are removed. The number of remaining high-quality signal samples is counted, and the effective operating signal recognition ratio is calculated as (number of remaining samples / total number of acquired samples) × 100%. For example, the effective signal power of the vibration signal sample is 500 mW, and the noise power is 5 mW. After screening, the number of samples that meet the standard in the initial 1000 samples is 950. The signal-to-noise ratio is 10 × log10(500 / 5) = 10 × 2 = 20 dB. The effective operating signal recognition rate is (950 / 1000) × 100% = 95%. Experimental data shows that when the signal-to-noise ratio is <15 dB, the error in extracting fault feature frequencies increases exponentially. When the waveform integrity is <0.85, normal harmonics are easily misjudged as fault features. Therefore, the signal quality standard range is set as signal-to-noise ratio >15 dB and waveform integrity >0.85, as shown in Table 1 below. A signal-to-noise ratio of 15 dB and a waveform integrity of 0.85 are the critical inflection points to ensure an accuracy of over 94%.

[0066] Table 1: Experimental data for determining signal quality thresholds

[0067]

[0068] As shown in Table 1, the data from Experiment 2 indicate that a signal-to-noise ratio of 15 dB and a waveform integrity of 0.85 are the critical inflection points that ensure an accuracy of over 94%. This experimental data strongly supports the rationality of the threshold setting.

[0069] Please see Figure 3 The specific steps for determining the stability value of the motor's operating state are as follows:

[0070] S201: Based on the effective operation signal identification ratio, the key feature vectors of asynchronous motors in the operation monitoring signals are extracted using fast Fourier transform and wavelet packet decomposition. Signal features with feature energy density lower than the preset level or with cross-correlation coefficients with the health standard model are selected. The distribution pattern and matching degree of feature vectors within a predetermined time window are compared to obtain an abnormal feature vector record list.

[0071] Key feature vectors include stator current characteristic frequency components, rotor bar breakage characteristic frequency offset, and bearing vibration envelope characteristics;

[0072] The deep feature extraction process is initiated based on the effective operating signal. Fast Fourier Transform (FFT) is used to convert the time-domain stator current and vibration signal into a frequency-domain spectrum to locate the fundamental frequency and its higher harmonic components. For non-stationary impact signals, wavelet packet decomposition is used for time-frequency analysis. The wavelet packet decomposition is set to a 3-layer decomposition structure, using Daubechies4 as the base wavelet, dividing the signal frequency band into 8 independent frequency band nodes. The signal energy at each frequency band node is calculated, i.e., the sum of squares of wavelet coefficients within that frequency band, to construct a feature energy density vector. A pre-trained health standard model is called to perform feature comparison. This model is based on an autoencoder deep neural network, including an input layer, encoder, bottleneck layer, decoder, and output layer. The input layer has 64 neurons. The encoder has 32 neurons in the first hidden layer and 16 neurons in the second hidden layer using the ReLU activation function. The bottleneck layer has 8 neurons to extract core low-dimensional manifold features. The decoder has a symmetrical structure. The output layer has 64 neurons. The input data is reconstructed using the Sigmoid activation function. The real-time extracted feature vectors are input into a health standard model, and the cross-correlation coefficient between the input features and the model reconstruction output is calculated. If the cross-correlation coefficient is lower than a preset level or the feature energy density shows abnormal accumulation in a specific fault frequency band, the feature vector is determined to be abnormal. The distribution pattern of the feature vectors within a predetermined time window is compared; if the variance of the feature value fluctuation exceeds a stable threshold, an anomaly is recorded. For example, in the wavelet packet decomposition of the stator current signal, the sum of squares of the coefficients in the third frequency band is calculated to be 20, the sum of squares of the coefficients in all eight frequency bands is 100, the real-time feature vector modulus is 10, the reconstructed vector modulus is 9.8, and the dot product of the two is 95. Dividing the energy of the third frequency band (20) by the total energy (100) yields a feature energy density of 0.2. Dividing the dot product (95) by the product of the moduli (98) yields a cross-correlation coefficient of 0.969, indicating that the waveform structure conforms to health characteristics. If the energy density is 0.2 but the cross-correlation coefficient drops to 0.7, an anomaly is indicated, and the abnormal feature vector record list needs to be stored.

[0073] S202: Based on the list of abnormal feature vector records, identify the degree of dispersion of the motor operating state per unit time. By calculating the Euclidean distance between the real-time feature vector and the health standard model, determine the degree of deviation between the motor operating state and the standard state, and generate a stability measure of the motor operating state.

[0074] The quantification of motor operating state dispersion based on anomaly feature vector record list hinges on calculating the Euclidean distance between the real-time feature vector and the baseline center of the health standard model. The baseline center of the health standard model is the centroid formed by clustering all healthy samples in the high-dimensional feature space. The feature vectors of real-time acquired multi-dimensional data, including stator current frequency components and bearing vibration envelope features, are mapped to the same high-dimensional space. Distance calculation is performed to obtain the coordinate values ​​of the real-time feature vector and the coordinate values ​​of the health baseline center. For each dimension, the difference between the two coordinate values ​​is calculated, and the difference is squared. The square root of the sum of the squared differences across all dimensions is then used to obtain the Euclidean distance. The stability metric of motor operation is calculated using numerical methods. An inverse proportional logic is employed, dividing a preset constant by 1 and adding the sum of Euclidean distances. For example, simplifying to a two-dimensional feature space, the real-time feature vector coordinates are 5 and 8, the health baseline center coordinates are 2 and 4, and the preset constant is set to 100. Taking the square root of (5-2^2)+(8-4^2)=25, we obtain an Euclidean distance of 5. Subtracting 100 / (1+5)=16.67, we arrive at a motor operation stability metric of approximately 16.67. This calculation indicates a significant deviation between the current state and the health baseline, resulting in a substantial decrease in the stability metric, suggesting that the motor is in an unstable operating region.

[0075] S203: Based on the stability measurement of the motor operating state and combined with the preset motor operating stability grading standard, perform stability correlation analysis, and calculate the stability deviation comparison value after comparing the real-time operating state stability measurement with the preset stability threshold to obtain the motor operating state stability value.

[0076] Based on the motor's operational stability measurement and a preset grading standard, a correlation analysis is performed. The preset grading standard divides the stability measurement into four levels: stable, sub-healthy, warning, and dangerous, and defines a critical score for each level. The stability deviation is calculated by comparing the real-time state with the set standard. The current real-time stability measurement is subtracted from the corresponding rated stability threshold. A negative result indicates insufficient stability. The deviation rate is then calculated by dividing the absolute value of this difference by the rated threshold. The final motor operational stability value is obtained by combining the deviation rate and the real-time measurement value through a weighted comprehensive score. The calculation logic is: real-time stability measurement multiplied by a weighting coefficient, minus the deviation rate multiplied by 100, and then multiplied by a penalty weighting coefficient. For example... The real-time stability metric is 16.67. The current operating condition requires a rated stability threshold of 80. The weighting coefficient A is set to 0.7, and the penalty weighting coefficient B is set to 0.3. 16.67-80=-63.33. Taking |-63.33| / 80, we get an offset rate of 0.7916. 16.67*0.7=11.669. The offset rate 0.7916*100*0.3=23.748. 11.669-23.748=-12.079, resulting in a final stability value of -12.079. This negative value strongly indicates that the motor's operating condition is severely below the standard and is heavily penalized due to the high offset rate, directly mapping to the highest risk level.

[0077] Please see Figure 4 The specific steps for creating the fault monitoring priority index table are as follows:

[0078] S301: Based on the stability value of motor operation status, extract the real-time risk level score and monitoring priority dynamic configuration rules of asynchronous motor, perform data matching by motor equipment ID, filter motor records with risk level scores higher than the current monitoring priority, and obtain the risk motor statistics;

[0079] Based on the stability value of the motor's operating status, a reverse mapping algorithm is used to extract the real-time risk level score of the asynchronous motor. The risk level score is set to a closed range of 0 to 100, with a higher score indicating greater risk. The conversion logic is to subtract the normalized stability value from 100; if the stability value is negative, it is reset to zero, making the risk score 100. The system then calls the monitoring priority dynamic configuration rules stored in the management system, which define the corresponding monitoring frequency and data transmission priority for different risk score ranges. All online motors are traversed by motor device ID to perform data matching, comparing the calculated real-time risk level score of each motor with the current system-set monitoring priority. If a motor's risk level score maps to a required priority... Motors with a monitoring priority higher than the current actual configuration are marked as monitoring lagging. All motor records meeting this condition are statistically analyzed to generate a risk motor statistic. For example, a motor has a normalized stability value of 20, and its current monitoring priority is level 2 (low-frequency monitoring). The rule setting is that a score greater than 70 corresponds to priority 5 (high-frequency real-time monitoring). Subtracting 20 from 100 yields a risk level score of 80, which corresponds to a priority of level 5. Comparison shows that the expected priority of level 5 is greater than the current priority of level 2, satisfying the condition that the risk level score is higher than the current monitoring priority. This calculation indicates that the motor has a high risk but is only under low-frequency monitoring; the monitoring level must be immediately upgraded, and this record is included in the risk motor statistic.

[0080] S302: Based on the risk motor statistics, call the preset monitoring priority mapping algorithm to identify motor records with priority lag or mismatch, extract the corresponding motor equipment number and the working condition block number, and generate a monitoring priority mapping adjustment sequence.

[0081] Based on the statistics of high-risk motors, a preset monitoring priority mapping algorithm is invoked to resolve resource allocation conflicts and ensure that high-risk motors receive preferential access to computing and bandwidth resources. The algorithm identifies motor records where there is priority lag (i.e., a motor with a priority higher than the current priority) or mismatch (i.e., a motor with a priority lower than the current priority), resulting in wasted resources. For each identified record, its unique motor device number and the corresponding operating block number are extracted to generate a sequence containing the device ID, target priority, and adjustment time window instruction set. When generating the sequence, the network bandwidth margin of the operating block is checked. If there are too many high-priority motors in a block, resulting in insufficient bandwidth, a dynamic balancing strategy is activated to temporarily reduce the priority of low-risk motors in the same block to free up data channels. For example, if the bandwidth occupancy rate of block A reaches 95% and three new high-risk motors need to be upgraded to priority 5, the algorithm automatically scans the existing priority 5 motors in block A, selects motors whose risk scores have dropped below 30, and lowers their priority to 2, releasing bandwidth resources for the new high-risk motors to ensure real-time transmission of critical data. Finally, a monitoring priority mapping adjustment sequence containing specific adjustment instructions is output.

[0082] S303: Call the motor equipment number and operating condition block number in the monitoring priority mapping adjustment sequence, replace the corresponding monitoring priority mapping field in the database, identify the mapping relationship between the motor equipment number and the risk level score, and generate a fault monitoring priority index table;

[0083] The monitoring priority mapping adjustment sequence is invoked to perform the final database update operation. The transaction processing mechanism is used to locate the corresponding record in the device management database, lock the relevant row, and replace the original monitoring priority mapping field with the calculated target priority. The trigger source of this change, i.e., the risk level score, is recorded in the log table. After the update, the index relationship between the motor device number and the risk level score is re-established, generating the latest fault monitoring priority index table. This index table resides in the memory database for the acquisition module to query in real time to determine the sampling frequency for each motor in the next second. For example, for motor device number 1001, the monitoring frequency field in the database is updated from once per hour to 10 times per second, and the risk level field is updated to 80. After rebuilding the index, the acquisition module reads the change and immediately adjusts the data acquisition strategy for that motor.

[0084] Please see Figure 5 The specific steps for tracking and handling potential motor faults are as follows:

[0085] S401: Call the motor ID marked in the fault monitoring priority index table, locate the original operation record data in the distributed storage, extract the feature spectrum of the motor in the target frequency range, and identify the frequency position and amplitude deviation of the abnormal features by comparing the difference between the abnormal frequency components in the feature spectrum and the health standard model, and obtain the abnormal feature sequence position information.

[0086] The system retrieves the high-priority motor IDs from the fault monitoring priority index table, locates and reads the original operating data of these motors from the distributed file system within the past monitoring cycle, and uses Fast Fourier Transform to convert the time-domain data into a frequency-domain feature spectrum. It focuses on preset target frequency bands, specifically targeting the stator current fundamental frequency bands for squirrel-cage induction motors. By comparing the amplitude differences between the generated feature spectrum and the baseline spectrum of the health standard model at the same frequency, a differential operation is performed. If the amplitude deviation at a certain frequency point exceeds a specific multiple of the background noise level, an anomaly is determined at that frequency point. The system records the specific Hertz values ​​of all abnormal frequency points and the amplitude values ​​exceeding the baseline. The deviation is used to obtain the location information of the abnormal feature sequence through structured combination. For example, a 4-pole motor has a power supply frequency of 50 Hz, a synchronous speed of 1500 rpm, and an actual measured speed of 1485 rpm. In the spectrum, the measured amplitude at this frequency point is -40 dB, while the healthy reference is -80 dB. By subtracting the actual speed from the synchronous speed and then dividing by the synchronous speed, the slip rate is obtained as 0.01. Substituting this into the formula, the characteristic frequency position is calculated to be 49 Hz. Subtracting the reference amplitude of -80 dB from the measured amplitude of -40 dB gives a deviation of 40 dB. This calculation result shows that there is an abnormal gain of 40 dB at 49 Hz, which matches the characteristics of rotor bar breakage faults and accurately identifies the location of the abnormal feature.

[0087] S402: Based on the location information of the abnormal feature sequence, sort the feature spectrum by motor equipment ID in time series, identify the cumulative number of times the abnormal features appear in the continuous monitoring period, compare it with the preset abnormal frequency threshold, filter motor equipment with the cumulative number exceeding the threshold, extract the associated operating feature status information, remove data that does not meet the continuous abnormal conditions, and generate a continuous abnormal motor status sequence.

[0088] Based on the location information of abnormal feature sequences, a time dimension is introduced for longitudinal analysis. All feature spectra generated in the past 24 hours are sorted by motor device ID and time series. For each identified abnormal feature frequency, its occurrence within a continuous monitoring period is tracked, and the cumulative occurrence count is calculated. An abnormal frequency threshold logic is set, and the actual cumulative count is compared with this threshold. If the condition is met, it indicates that the abnormality is a persistent physical defect rather than transient interference. After filtering out motor devices that meet the conditions, the associated operating feature status information is extracted. Data that does not meet the continuous abnormality condition due to power grid fluctuations or instantaneous overload is removed, generating a continuous abnormal motor status sequence. For example, the monitoring window is set to 100 samples, and the abnormal frequency threshold is 80 times. For motor device ID 2002, its 49 Hz feature frequency appears 85 times in the last 100 samples, and the associated load data shows no drastic fluctuations. Comparing 85 times with the threshold of 80 times, 85 is greater than 80, meeting the condition. This judgment process confirms that device 2002 has a persistent rotor bar breakage risk and includes it in the continuous abnormal motor status sequence to prevent false alarms caused by occasional interference.

[0089] S403: Call the continuous abnormal motor state sequence, match the structured information of the original signal acquisition, write all motor equipment information identified as continuously abnormal into the fault hazard tracking block, collect the motor equipment ID and the corresponding abnormal feature description, and obtain the motor fault management hazard tracking and handling list.

[0090] The system executes a closed-loop operation for hazard management by invoking a sequence of continuously abnormal motor states. The motor equipment ID, specific abnormality description, occurrence time span, and risk level are uniformly encapsulated and written into the fault hazard tracking block. This block is a dedicated storage area for long-term tracking of persistent faults. Information is collected, organized, deduplicated, and merged, containing detailed information on all high-risk motors awaiting processing, and directly pushed to maintenance personnel's terminals as the basis for generating work orders. For example, equipment ID 2002, abnormality descriptions such as rotor bar breakage sidefrequency amplitude exceeding the standard, high risk level, and earliest discovery time of 08:00 on the same day are written into the tracking block. When generating the list, the system checks if there are any unclosed old work orders for the equipment. If so, the fault description is merged and updated; otherwise, a new entry is added. This process ensures that every continuous abnormal record is documented, forming a traceable list of motor fault management hazard tracking and handling.

[0091] Please see Figure 6 The specific steps for assessing the overall health status set of an asynchronous motor are as follows:

[0092] S501: Call the original operation record of the corresponding motor equipment in the motor fault management hidden danger tracking and handling list, extract the motor operation feature vector, identify the numerical correlation between the current monitoring result and the original record, organize the correspondence between the operation feature, monitoring judgment result and the original record in chronological order, and generate a motor status association binding table;

[0093] The system retrieves key motors from the motor fault management hazard tracking and handling list, extracts their original lifecycle operation records from cold storage, and extracts the motor's operation feature vectors at different historical stages. It then identifies the numerical correlation between current monitoring results and historical records. The system chronologically maps and concatenates the operation features, the monitoring judgment results at the time, and the original maintenance log records, providing a complete data context for trend analysis. For example, for a motor with abnormal vibration, the system retrieves all vibration data and maintenance records from the past three years since its commissioning. It correlates the current vibration value of 5 mm / s with the vibration value of 2 mm / s recorded after maintenance one year ago, aligning it with the load current data from the same period on the timeline. Through this binding operation, a comprehensive data table is constructed, including time, vibration amplitude, maintenance events, and load status, revealing the trajectory of vibration evolution over time and maintenance events, and generating a motor status association binding table.

[0094] S502: Based on the motor status association binding table, compare the real-time operating characteristics with the health benchmark information in the original records, check whether the feature offset between the two exceeds the preset safety evolution range, mark the abnormal records that exceed the range to the hidden danger section data field, and obtain the motor status difference marking results.

[0095] Evolution trend analysis is performed based on the motor status association binding table. Real-time operating characteristics are compared with the health baseline information in the original records, i.e., the fingerprint data when the motor was first put into operation, to check whether the characteristic offset exceeds the preset safe evolution range. The rate of change of the offset over time, i.e., the slope, is calculated. If the rate of change is significantly higher than the historical average or the preset aging curve within a certain period, it indicates that the motor is experiencing accelerated deterioration. Time periods exceeding the range and abnormal records are precisely marked in the hidden danger zone data field to obtain the motor status difference marking results. For example, the initial vibration value of a motor bearing is 1 mm / s, and after 5000 hours of operation, it reaches 5 mm / s. The preset safe aging rate is an increase of 0.5 mm / s per 1000 hours. Subtracting the initial value of 1 from the current value of 5 gives a total offset of 4 mm / s. Dividing the total offset of 4 by the operating time of 5 thousand hours gives an actual deterioration rate of 0.8 mm / s per thousand hours. Comparing the actual deterioration rate of 0.8 with the safe aging rate of 0.5, 0.8 is greater than 0.5. This calculation result indicates that the bearing wear rate is 60% faster than expected, belonging to abnormal accelerated aging, and the hidden danger zone is marked at the 5000-hour time node.

[0096] S503: Based on the motor sequence of the marked hidden danger sections in the motor status difference labeling results, summarize all abnormal marking statuses by motor equipment ID, collect the associated management fields, and embed the summarized information into the structured metadata of the original operation record to obtain the overall health assessment status set of the asynchronous motor;

[0097] Based on the motor condition difference annotation results, including sudden failures, ongoing hidden dangers, and accelerated aging, the associated management fields include the responsible person, last maintenance time, and spare parts inventory status. The summarized information is embedded into the structured data header of the original operation record in the form of metadata tags, realizing the permanent binding of data and evaluation results, reflecting the current status and including historical evolution paths. For example, all annotations of equipment ID3005 are summarized, including the hidden danger of accelerated bearing aging and the record of occasional stator temperature exceeding the limit, and associated with information such as the responsible person Zhang San and sufficient spare parts inventory. The information is packaged and written into the header metadata area of ​​the original data file of the equipment. Finally, an overall health assessment status set of asynchronous motor is generated, which not only shows that the current health of equipment 3005 is 60 points, but also provides a degradation trend chart and spare parts suggestions that can be viewed by clicking, providing high-quality annotation samples for predictive maintenance models.

[0098] Please see Figure 7 An AI-based asynchronous motor fault management system includes:

[0099] The signal verification module collects operation monitoring signals of the asynchronous motor from multiple sensors, filters raw data entries with abnormal signal strength fluctuations or signal-to-noise ratios below a preset range, determines whether the signal meets the integrity requirements, removes signal segments that do not meet the requirements, and obtains the effective operation signal recognition ratio.

[0100] The feature analysis module, based on the effective operating signal recognition ratio, uses time-frequency analysis to extract key feature vectors of asynchronous motors from the operating monitoring signals, statistically analyzes the matching degree between feature energy distribution and health standard model, identifies sparse signal segments with feature distribution, analyzes the stability of motor operating status, and establishes motor operating status stability values.

[0101] The priority dynamic adjustment module compares the motor risk level score with the monitoring priority configuration rules based on the motor's operating status stability value, filters motor equipment records with risk levels higher than the current priority, adjusts the monitoring mapping relationship value, writes the updated results into the monitoring priority field, and builds a fault monitoring priority index table.

[0102] The hazard identification and tracking module, based on the fault monitoring priority index table, locates the risk motor ID, extracts the associated original operation records and feature spectra, compares real-time features with the health standard model, identifies abnormal states within the continuous monitoring period, filters motor equipment with an abnormal cumulative number of times greater than a preset threshold, writes them into the fault hazard tracking block, and generates a motor fault management hazard tracking and handling list.

[0103] The trend assessment and labeling module extracts the associated monitoring results and original records based on the motor fault management hidden danger tracking and handling list, compares the trend difference between the motor operating status and the original baseline, marks the location of the hidden danger status in the data structure, and obtains the overall health assessment status set of the asynchronous motor.

[0104] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the protection scope defined by the technical solution of the present invention.

Claims

1. An asynchronous motor fault management method based on artificial intelligence, characterized in that, Includes the following steps: S1: Based on the vibration sensor, collect the operation monitoring signal of the asynchronous motor, filter the original data items with abnormal signal strength fluctuation or signal-to-noise ratio below the preset range, determine whether the signal meets the integrity and quality requirements, remove the signal segments that do not meet the requirements, and obtain the effective operation signal recognition ratio. S2: Based on the effective operating signal identification ratio, extract the key feature vectors of motor faults from the operating monitoring signals, screen the signal features with sparse feature energy distribution and low matching degree with the health standard model, analyze the consistency of key features in the frequency domain and time domain, and obtain the stability value of motor operating status. S3: Based on the stability value of the motor operating status, extract the risk level score and fault monitoring priority configuration rules, filter out devices with risk levels higher than the set threshold, adjust the monitoring priority mapping relationship, and generate a fault monitoring priority index table; S4: Based on the motor ID marked in the fault monitoring priority index table, extract the original operation record data, filter motor equipment with a number of consecutive abnormal feature manifestations greater than a set value, write the motor equipment information into the fault hazard tracking block, and obtain the motor fault management hazard tracking and handling list. The specific steps for establishing the motor fault management hidden danger tracking and handling checklist are as follows: S401: Call the motor ID marked in the fault monitoring priority index table, locate the original operation record data in the distributed storage, extract the feature spectrum of the motor in the target frequency band, and identify the frequency position and amplitude deviation of the abnormal features by comparing the difference between the abnormal frequency components in the feature spectrum and the health standard model to obtain the abnormal feature sequence position information. S402: Based on the abnormal feature sequence location information, sort the feature spectrum by motor equipment ID in time series, identify the cumulative number of times abnormal features appear in the continuous monitoring period, compare it with the preset abnormal frequency threshold, filter motor equipment with the cumulative number of times exceeding the threshold, extract the associated operating feature status information, remove data that does not meet the continuous abnormal conditions, and generate a continuous abnormal motor status sequence. S403: Call the continuous abnormal motor state sequence, match the structured information of the original signal acquisition, write all motor equipment information identified as continuously abnormal into the fault hazard tracking block, collect the motor equipment ID and the corresponding abnormal feature description, and obtain the motor fault management hazard tracking and handling list. S5: Based on the operation records of the corresponding motor equipment in the motor fault management hidden danger tracking and handling list, identify the logical correlation between the current monitoring results and the original records, determine the difference in the evolution trend between the motor operation status and the original health record, mark the difference results to the hidden danger section, and output the overall health assessment status set of the asynchronous motor. The overall health assessment status set of the asynchronous motor includes fault evolution trend difference identifiers, abnormal motor mark numbers, and monitoring result comparison index items.

2. The asynchronous motor fault management method based on artificial intelligence according to claim 1, characterized in that, The effective operating signal recognition ratio includes signal amplitude compliance rate, spectrum distribution deviation, and sampling period integrity matching degree. The motor operating status stability value includes characteristic frequency distribution uniformity, fault feature point matching degree, and operating parameter recording continuity. The fault monitoring priority index table balances equipment risk level identification, monitoring frequency update item, and key motor node mark. The motor fault management hidden danger tracking and handling list includes hidden danger motor number, characteristic abnormality frequency statistics, and hidden danger block index.

3. The asynchronous motor fault management method based on artificial intelligence according to claim 1, characterized in that, The specific steps for determining the effective operating signal recognition ratio are as follows: S101: The operation monitoring signals are collected by a sensor group deployed at a predetermined position on the stator housing of the asynchronous motor, the instantaneous amplitude and background noise intensity information of each signal are extracted, the peak factor and noise interference of the signal are checked, signal segments that exceed the preset interference range are eliminated, and the data are summarized and statistically analyzed to obtain the number of signal samples that meet the quality requirements. The operation monitoring signals include stator current signal, rotor vibration signal, and winding temperature signal; S102: Based on the signal-to-noise ratio and waveform integrity information of each sample in the number of signal samples that meet the quality requirements, compare it with the preset signal quality standard range, eliminate abnormal signals outside the range, count the number of remaining signal samples and calculate its proportion to the total number of collected signal samples to obtain the effective operating signal recognition ratio.

4. The asynchronous motor fault management method based on artificial intelligence according to claim 3, characterized in that, The specific steps for determining the stability value of the motor's operating state are as follows: S201: Based on the effective operating signal identification ratio, the key feature vectors of the asynchronous motor in the operating monitoring signal are extracted using fast Fourier transform and wavelet packet decomposition. Signal features with feature energy density lower than the preset level or with cross-correlation coefficient with the health standard model are selected. The distribution pattern and matching degree of the feature vectors within the predetermined time window are compared to obtain an abnormal feature vector record list. S202: Based on the abnormal feature vector record list, identify the degree of dispersion of the motor operating state per unit time, and determine the degree of deviation between the motor operating state and the standard state by calculating the Euclidean distance between the real-time feature vector and the health standard model, and generate a motor operating state stability measure. S203: Based on the motor operating state stability metric and combined with the preset motor operating stability grading standard, perform stability correlation analysis, and calculate the stability offset comparison value after comparing the real-time operating state stability metric with the preset stability threshold to obtain the motor operating state stability value.

5. The asynchronous motor fault management method based on artificial intelligence according to claim 4, characterized in that, The key feature vectors include the stator current characteristic frequency component, the rotor bar breakage characteristic frequency offset, and the bearing vibration envelope characteristics.

6. The asynchronous motor fault management method based on artificial intelligence according to claim 4, characterized in that, The steps for establishing the fault monitoring priority index table are as follows: S301: Based on the stability value of the motor's operating status, extract the real-time risk level score and monitoring priority dynamic configuration rules of the asynchronous motor, perform data matching by motor equipment ID, filter motor records with risk level scores higher than the current monitoring priority, and obtain the risk motor statistics; S302: Based on the risk motor statistics, call the preset monitoring priority mapping algorithm to identify motor records with priority lag or mismatch, extract the corresponding motor equipment number and the working condition block number, and generate a monitoring priority mapping adjustment sequence. S303: Call the motor equipment number and operating condition block number in the monitoring priority mapping adjustment sequence, replace the corresponding monitoring priority mapping field in the database, identify the mapping relationship between the motor equipment number and the risk level score, and generate a fault monitoring priority index table.

7. The asynchronous motor fault management method based on artificial intelligence according to claim 1, characterized in that, The specific steps for the overall health assessment status set of the asynchronous motor are as follows: S501: Call the original operation record of the corresponding motor equipment in the motor fault management hidden danger tracking and handling list, extract the motor operation feature vector, identify the numerical correlation between the current monitoring result and the original record, organize the correspondence between the operation feature, monitoring judgment result and the original record in chronological order, and generate a motor status association binding table; S502: Based on the motor status association binding table, compare the real-time operating characteristics with the health benchmark information in the original record, check whether the feature offset between the two exceeds the preset safety evolution range, mark the abnormal records that exceed the range to the hidden danger section data field, and obtain the motor status difference marking result. S503: Based on the motor sequence of the marked hidden danger sections in the motor status difference marking results, summarize all abnormal marking statuses by motor equipment ID, collect the associated management fields, and embed the summarized information into the structured metadata of the original operation record to obtain the overall health assessment status set of the asynchronous motor.

8. An asynchronous motor fault management system based on artificial intelligence, characterized in that, The system is used to implement the artificial intelligence-based asynchronous motor fault management method according to any one of claims 1-7, the system comprising: The signal verification module collects operation monitoring signals of the asynchronous motor from multiple sensors, filters raw data entries with abnormal signal strength fluctuations or signal-to-noise ratios below a preset range, determines whether the signal meets the integrity requirements, removes signal segments that do not meet the requirements, and obtains the effective operation signal recognition ratio. The feature analysis module, based on the effective operating signal recognition ratio, uses time-frequency analysis to extract key feature vectors of the asynchronous motor in the operating monitoring signal, calculates the matching degree between the feature energy distribution and the health standard model, identifies sparse signal segments of feature distribution, analyzes the stability of the motor operating state, and establishes the stability value of the motor operating state. The priority dynamic adjustment module, based on the motor operating status stability value, compares the motor risk level score with the monitoring priority configuration rules, filters motor equipment records with risk levels higher than the current priority, adjusts the monitoring mapping relationship value, writes the update result into the monitoring priority field, and constructs a fault monitoring priority index table. The hidden danger identification and tracking module, based on the fault monitoring priority index table, locates the risk motor ID, extracts the associated original operation records and feature spectrum, compares the real-time features with the health standard model, identifies the abnormal state within the continuous monitoring period, filters motor equipment with an abnormal cumulative number of times greater than a preset threshold, writes them into the fault hidden danger tracking block, and generates a motor fault management hidden danger tracking and handling list. The trend assessment and labeling module, based on the motor fault management hidden danger tracking and handling list, extracts the associated monitoring results and original records, compares the trend difference between the motor operating status and the original baseline, labels the hidden danger status position in the data structure, and obtains the overall health assessment status set of the asynchronous motor.

Images