Intelligent electrical equipment maintenance task decision support system and method

Abstract

The invention discloses an intelligent electrical equipment maintenance task decision support system and method, particularly relates to the field of motor maintenance, is used for solving the problem of running-in mismatching of new and old parts, and comprises the following steps: collecting multi-dimensional performance parameters of the new and old parts, constructing a performance feature database, and detailedly describing key parameter differences; a personalized running-in curve is generated by using a support vector machine regression algorithm, the load change rate, the rotating speed adjustment step length and the running-in time are accurately set, and vibration and current fluctuation caused by initial mismatching of new and old parts are avoided; by monitoring the temperature, the vibration spectrum and the current fluctuation of the equipment in real time, the load and the rotating speed are automatically adjusted, and the running-in process is ensured to be stable; the frequency spectrum sharpness index and the time sequence entropy difference index are combined, running-in state classification analysis is carried out, and potential anomalies are recognized in time; in an ambiguous state, running-in is optimized by adjusting PID parameters, it is ensured that new and old parts are gradually matched, the operation stability of equipment is improved, and faults and damage caused by poor running-in for a long time are avoided.

Description

Technical Field

[0001] The present invention relates to the field of motor maintenance, and more specifically, to an intelligent electrical equipment maintenance task decision support system and method. Background Art

[0002] With the widespread application of electrical equipment in various industrial production and infrastructure, the stable operation of the equipment has an important impact on the reliability of the overall system. In the fields of electricity, manufacturing, transportation, etc., a large number of electrical equipment need to operate continuously for a long time, and their key components will inevitably wear, age and deteriorate after experiencing high loads and frequent starts and stops. In order to ensure the normal operation of the equipment and extend its service life, regular maintenance, inspection and troubleshooting have become the core links in equipment management. With the increase in the complexity of electrical equipment, maintenance work is not only limited to traditional fault repair, but also covers preventive maintenance and predictive maintenance. Through precise detection and component replacement, the impact of sudden failures on the overall system can be avoided.

[0003] As the core device in electrical equipment, the stable operation of the motor is crucial to the reliability of the entire system. However, after long-term use, the aging of motor components and performance degradation are particularly prominent. Traditional maintenance methods are usually fault-repair-oriented, restoring the equipment to short-term operation by replacing aging or damaged components in the motor. However, existing maintenance technologies often ignore the running-in problem between new and old components. Due to the improved performance of new components, they may initially have performance differences with other old components in the motor, especially in key parameters such as speed, torque, and electrical impedance. The new and old components fail to fully match and run-in, resulting in increased vibration and current fluctuations during equipment operation, and even aggravated wear of mechanical components.

[0004] This imbalance between new and old parts not only affects the short-term performance of the motor, but is also likely to lead to the gradual accumulation of operating hazards. The existing technology lacks real-time monitoring and parameter adjustment methods for the running-in period of new and old parts, which affects the stability of the equipment after maintenance and has an adverse impact on long-term reliability.

[0005] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0006] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide an intelligent electrical equipment maintenance task decision support system and method, and propose an effective solution for the running-in process of new and old parts of the motor. First, by collecting multi-dimensional performance parameters of new and old parts and building a performance feature database, the key parameter differences of different parts can be described in detail, especially the factors affecting the running-in effect such as speed, torque and electrical impedance. Secondly, the support vector machine regression algorithm is used to generate a personalized running-in curve, and the load change rate, speed adjustment step and running-in time are accurately set according to the actual operation conditions, avoiding the vibration and current fluctuation problems caused by the initial mismatch of new and old parts. In addition, the temperature, vibration spectrum and current fluctuation of the equipment are monitored in real time to ensure that when any parameter exceeds the threshold during the running-in process, the adaptive control algorithm can automatically adjust the load and speed to ensure a smooth and orderly running-in process. By extracting frequency domain and time domain features, calculating the spectrum sharpness index and the time series entropy difference index, and combining the time-frequency gain fusion index, the running-in state is classified and analyzed, which can effectively identify the potential abnormalities of the equipment after the new parts are replaced, and present abnormal, ambiguous or normal maintenance status. Especially in the ambiguous state, by adjusting the PID parameters of the equipment controller, the running-in process of the new and old parts is further optimized to ensure that the performance of the new and old parts gradually matches, avoiding long-term performance degradation and equipment stability problems caused by insufficient running-in. This technical solution not only realizes real-time monitoring and dynamic adjustment of the running-in period, solving the deficiency of ignoring the running-in period of new and old parts in the prior art, but also improves the operating stability and long-term reliability of the motor equipment by optimizing the running-in process, and prevents secondary failures and mechanical damage caused by improper running-in, so as to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] S1, collect performance parameters of new and old components and build a multi-dimensional performance characteristic database;

[0009] S2, uses the support vector machine regression algorithm to process multi-dimensional performance characteristic data, generates a personalized running-in curve, and sets the load change rate, speed adjustment step size, and operating time during the running-in period;

[0010] S3, real-time monitoring of the equipment’s temperature, vibration spectrum, and current fluctuations, and using control algorithms to dynamically adjust load and speed when parameters exceed thresholds;

[0011] S4, based on the frequency domain and time domain feature extraction of the vibration signal during the running-in process of the new and old parts, the spectrum sharpness index and the time series entropy difference index are calculated, and the potential running-in anomaly after the replacement of the new parts of the equipment is identified through the time-frequency gain fusion index, and the maintenance abnormality state, ambiguous state or normal maintenance state are displayed;

[0012] S5, if a maintenance ambiguous state is obtained, the PID parameters of the equipment controller are adjusted to optimize the operating parameters during the running-in process.

[0013] In a preferred embodiment, step S2 includes the following contents:

[0014] First, the multi-dimensional performance data collected in step S1 is standardized;

[0015] The standardized multi-dimensional performance data is used as the training data set, the target output value is the running-in state of the equipment, and the mean square error is used as the loss function for model training to optimize the parameters and penalty factors of the SVR model.

[0016] After training is completed, the SVR model predicts the running-in status under different working conditions based on the characteristic data of new and old components; through the predicted running-in status, the corresponding running-in curve is generated, which reflects the relationship between load change, speed adjustment and running time.

[0017] In a preferred embodiment, step S3 includes the following contents:

[0018] A set of threshold ranges are preset according to the operating characteristics of the equipment. The thresholds include the upper temperature limit, the upper vibration frequency limit and the safe range of current fluctuations. Once any parameter exceeds the preset threshold, the adaptive control logic is entered, and the load and speed are dynamically adjusted according to real-time data. The adjustment strategy is executed according to the type and degree of the parameter exceeding the threshold.

[0019] In a preferred embodiment, step S4 includes the following contents:

[0020] During the operation of the motor, mechanical vibration data is collected in real time and the vibration waveform of the equipment under different load conditions is recorded; these signals are stored in the form of time domain data, indicating the change of vibration intensity of the equipment over time; let the vibration signal be x(t), where t is time and x is the amplitude of vibration; after each load adjustment, the vibration signal needs to be recollected to ensure the continuity of the data; after the old parts are replaced, the frequency sharpness index and time series entropy difference index are obtained based on the vibration signal generated by the installation of the new parts.

[0021] In a preferred embodiment, the acquisition logic of the spectrum sharpness index is:

[0022] First, the vibration signal x(t) in the time domain is converted to the frequency domain to obtain the spectral components of the vibration signal. For the vibration signal, the time domain signal x(t) is defined as: Among them, A is the amplitude of the main vibration signal, f0 is the main frequency, A n is the amplitude of other minor frequencies, f nare the secondary frequencies, φ and φ n is the phase angle;

[0023] The spectrum components are obtained by frequency domain transformation, and the spectrum is S(f), where f is the frequency and S(f) is the vibration intensity at the frequency f;

[0024] The distribution characteristics of the spectrum are quantified and the frequency distribution function F(f) is constructed, which reflects the relative distribution of each frequency component; it is defined as follows: Among them, f max is the highest frequency range of the spectrum, |S(f)| 2 is the energy intensity at frequency f; normalized to ensure that the total energy is distributed between 0 and f max between;

[0025] The spectrum sharpness increment function is calculated to reflect the sudden change of local frequency energy in the spectrum; the local energy change rate based on frequency is expressed as follows: Where Δf is the frequency interval, which represents the rate of change of local energy within the frequency range;

[0026] Taking all the sharpness increments at all frequencies into consideration, the spectrum sharpness index SPI is calculated. By integrating the frequency sharpness increment and taking the curvature of the local sharpness increment as a correction term into the calculation, we get: in, It is the second-order derivative at frequency f, reflecting the change in curvature of the frequency energy distribution.

[0027] In a preferred embodiment, the logic for obtaining the time series entropy difference index is:

[0028] Convert the continuous vibration signal x(t) into a discrete symbol sequence;

[0029] The amplitude of the vibration signal is divided into M intervals according to a predetermined threshold and mapped into a symbol set {s1, s2, …, s M}:s k =Φ(x(t i ))=s k , when τ k-1 <x(t i )≤τ k ; where τ0,τ1,…,τ M is the partition threshold, satisfying τ0<τ1<…<τ M ,k=1,2,…,M;

[0030] According to the statistical characteristics of the vibration signal, the threshold τ is set k :τ k =τ0+k·Δτ; where, Δτ=(τ M-τ0) / M, to ensure that the symbol intervals are of equal width or adjusted according to signal characteristics;

[0031] Using symbol sequences, the transition probabilities between symbols are calculated to reflect the dynamic changes of time series;

[0032] Calculation symbol s i Transfer to symbol s j The number of Where δ is the indicator function, which takes 1 when the condition is met and takes 0 otherwise, and T is the length of the symbol sequence;

[0033] Calculate the transition probability matrix Ensure that the transition probability of each symbol satisfies the normalization condition:

[0034] Through the transfer probability matrix, the entropy difference between the current vibration signal and the normal state signal is calculated to obtain the time series entropy difference index; the current transfer entropy is calculated

[0035] Using the vibration signal under normal operation, the corresponding transfer entropy H is calculated in the same way. normal ; Calculate the time series entropy difference index TSEDI: TSEDI = |HH normal |.

[0036] In a preferred embodiment, step S4 includes the following contents:

[0037] Two gain functions are calculated for the spectrum sharpness index and the time series entropy difference index, respectively, to indicate their sensitivity to potential anomalies in different dimensions: Combine the time-frequency gain functions into a time-frequency gain fusion index

[0038] The time-frequency gain fusion index is compared with the upper and lower thresholds; if the time-frequency gain fusion index is greater than or equal to the upper threshold, it indicates that there are obvious abnormalities in the running-in process, indicating that the newly replaced components are obviously not suitable for the current use status of the equipment, and a maintenance abnormality status is displayed; if the time-frequency gain fusion index is greater than or equal to the lower threshold and less than the upper threshold, further analysis of the running-in situation is required, and the operating parameters are adjusted to ensure safety, and a maintenance ambiguous status is displayed; if the time-frequency gain fusion index is less than the lower threshold, it indicates that the running-in process of the new and old components is stable, and a maintenance normal status is displayed.

[0039] In a preferred embodiment, step S5 includes the following contents:

[0040] Obtain the proportional, integral and differential parameters of the current controller, monitor the speed, torque and vibration signal amplitude of the equipment in real time, and analyze the response curve during load changes as a basis for subsequent PID parameter adjustments;

[0041] To address the response speed issue during equipment running-in, the time-frequency gain fusion index of the vibration signal is adjusted in real time. The adjustment method is to multiply the original proportional coefficient by a factor related to the time-frequency gain fusion index. The size of the factor depends on the position of the index between the upper and lower thresholds.

[0042] In view of the phenomenon of equipment accumulating errors, the integral coefficient is adjusted to eliminate the steady-state error. The adjustment method is to make an additive adjustment to the original integral coefficient based on the total accumulated errors of the equipment during operation.

[0043] In response to the sudden vibration or fluctuation of the equipment in the short term, the instantaneous overshoot is suppressed by adjusting the differential coefficient. The adjustment method is to adjust the differential coefficient according to the instantaneous change rate of the time-frequency gain fusion index.

[0044] Intelligent electrical equipment maintenance task decision support system, including: performance acquisition module, algorithm processing module, monitoring and adjustment module, signal analysis module and parameter optimization module;

[0045] Performance acquisition module: collects the performance parameters of new and old parts, builds a multi-dimensional performance characteristic database, outputs detailed performance characteristic data of new and old parts, and passes it to the algorithm processing module;

[0046] Algorithm processing module: Based on the collected multi-dimensional performance characteristic data, the support vector machine regression algorithm is used to generate a personalized running-in curve, set the load change rate, speed adjustment step and running time during the running-in period, output the running-in curve and load and speed adjustment plan, and pass it to the monitoring and adjustment module;

[0047] Monitoring and adjustment module: monitors the temperature, vibration spectrum and current fluctuation of the equipment in real time, uses control algorithms to dynamically adjust the load and speed when the parameters exceed the preset thresholds, outputs the real-time monitoring data of the equipment and the adjusted operating status, and transmits it to the signal analysis module;

[0048] Signal analysis module: Based on the vibration signal in the monitoring data, it extracts frequency domain and time domain features, calculates the spectrum sharpness index and time series entropy difference index, combines the time-frequency gain fusion index to identify the abnormal state of equipment running-in, outputs the abnormal classification, and passes it to the parameter optimization module;

[0049] Parameter Optimization Module: If an ambiguous state is obtained, the operating parameters during the running-in process are optimized by adjusting the PID parameters of the equipment controller, and the optimized PID parameter settings are output to ensure that the performance of the new and old parts gradually matches.

[0050] The technical effects and advantages of the intelligent electrical equipment maintenance task decision support system and method of the present invention are as follows:

[0051] The present invention proposes an effective solution for the running-in process of new and old parts of the motor. First, by collecting multi-dimensional performance parameters of new and old parts and building a performance feature database, the key parameter differences of different parts can be described in detail, especially the factors affecting the running-in effect such as speed, torque and electrical impedance. Secondly, a personalized running-in curve is generated by using a support vector machine regression algorithm, and the load change rate, speed adjustment step and running-in time are accurately set according to the actual operating conditions, avoiding the vibration and current fluctuation problems caused by the initial mismatch between the new and old parts. In addition, the temperature, vibration spectrum and current fluctuation of the equipment are monitored in real time to ensure that when any parameter exceeds the threshold during the running-in process, the adaptive control algorithm can automatically adjust the load and speed to ensure a smooth and orderly running-in process. By extracting frequency domain and time domain features, calculating the spectrum sharpness index and the time series entropy difference index, and combining the time-frequency gain fusion index, the running-in status is classified and analyzed, which can effectively identify the potential abnormalities of the equipment after the new parts are replaced, and present abnormal, ambiguous or normal maintenance status. Especially in the ambiguous state, by adjusting the PID parameters of the equipment controller, the running-in process of new and old parts is further optimized to ensure that the performance of new and old parts gradually matches, avoiding long-term performance degradation and equipment stability problems caused by insufficient running-in. This technical solution not only realizes real-time monitoring and dynamic adjustment of the running-in period, solving the problem of ignoring the running-in period of new and old parts in the existing technology, but also improves the operating stability and long-term reliability of motor equipment by optimizing the running-in process, and prevents secondary failures and mechanical damage caused by improper running-in. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flow chart of the intelligent electrical equipment maintenance task decision support method of the present invention.

[0053] Figure 2 It is a flow chart of the intelligent electrical equipment maintenance task decision support system of the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] Embodiment 1: Figure 1 The present invention provides an intelligent electrical equipment maintenance task decision support method, comprising:

[0056] S1, collect performance parameters of new and old components and build a multi-dimensional performance characteristic database;

[0057] S2, uses the support vector machine regression algorithm to process multi-dimensional performance characteristic data, generates a personalized running-in curve, and sets the load change rate, speed adjustment step size, and operating time during the running-in period;

[0058] S3, real-time monitoring of the equipment’s temperature, vibration spectrum, and current fluctuations, and using control algorithms to dynamically adjust load and speed when parameters exceed thresholds;

[0059] S4, based on the frequency domain and time domain feature extraction of the vibration signal during the running-in process of the new and old parts, the spectrum sharpness index and the time series entropy difference index are calculated, and the potential running-in anomaly after the replacement of the new parts of the equipment is identified through the time-frequency gain fusion index, and the maintenance abnormality state, ambiguous state or normal maintenance state are displayed;

[0060] S5, if a maintenance ambiguous state is obtained, the PID parameters of the equipment controller are adjusted to optimize the operating parameters during the running-in process.

[0061] After a long period of operation, key components of the motor, such as the rotor, stator, and bearings, are prone to wear or aging. In order to maintain the normal operation of the motor, these components are usually replaced. However, the performance difference between the new and old components may cause the equipment to experience increased vibration, unbalanced mechanical load, and other problems after replacement, affecting the overall operating stability of the motor. Therefore, after replacing the components, it is necessary to perform running-in management of the new and old components. In order to achieve accurate management of running-in, it is first necessary to collect the key performance parameters of the new and old components and build a data model containing multi-dimensional performance characteristics as the basis for subsequent running-in control and adjustment.

[0062] Step S1 includes the following contents:

[0063] Install strain gauges, accelerometers, temperature sensors, vibration sensors and current sensors on key components of the motor. Strain gauges are mainly installed on the motor bearings and rotor parts to measure torque and mechanical strain. Accelerometers are used to monitor mechanical vibrations and are installed on the motor housing and its fixed parts. Temperature sensors need to be placed around the stator windings, rotors and bearings to record temperature changes of various components. Current sensors are placed at the motor power input end to monitor motor current fluctuations in real time. Vibration sensors are used to monitor the overall vibration spectrum and abnormal vibration behavior of the motor.

[0064] Select appropriate acquisition hardware, including multi-channel data acquisition cards and signal conditioning equipment. The acquisition card must have high-speed sampling capabilities (at least 10kHz) to capture the dynamic performance changes of the motor when it is running at high speed. At the same time, select a signal conditioning module that supports the conversion of analog signals to digital signals, and convert the analog signals collected by each sensor into processable digital signals to facilitate subsequent data processing and storage.

[0065] Before starting the motor, ensure that all sensors are installed and connected to the data acquisition system. Start the motor and gradually load different operating conditions, including no-load, half-load and full-load states, and continuously collect performance parameters of new and old parts under different loads. The collected parameters include but are not limited to speed, torque, vibration frequency, temperature changes and current fluctuations. During the collection process, ensure that the running time of each state is long enough to ensure that sufficiently stable and representative parameter data is collected.

[0066] The collected performance data is transmitted to the central database in real time. The collected data is preprocessed, including data cleaning and outlier removal. Filtering technology (such as Kalman filtering) is used to remove noise signals and retain effective performance parameters.

[0067] The processed performance data is classified according to different dimensions to form a multi-dimensional performance characteristic database of new and old parts. The classification dimensions include mechanical characteristics (such as vibration frequency, torque), electrical characteristics (such as current fluctuation, impedance), thermal characteristics (such as temperature distribution), and operating characteristics under various load conditions. These multi-dimensional characteristics are stored using database technology (such as MySQL or MongoDB) for subsequent call and analysis.

[0068] During the operation of the motor, after the parts are replaced, the performance difference between the new and old parts will affect the overall operation of the motor, especially the key parameters such as torque, vibration frequency, speed, etc. may not perform the same under different loads. In order to ensure that the motor can smoothly pass the running-in period of the new and old parts, it is very important to generate a personalized running-in curve. By processing the multi-dimensional performance characteristic data, the running-in process of the new and old parts can be predicted, and a reasonable load change rate, speed adjustment step and running-in time can be provided for different operating conditions. The key to this step is to use the support vector machine regression algorithm (SVR) to generate a running-in curve that adapts to the performance difference between the new and old parts by processing the collected multi-dimensional performance data.

[0069] Step S2 includes the following contents:

[0070] First, the multi-dimensional performance data collected in step S1 is standardized to ensure that data of different dimensions have the same dimension and avoid calculation deviation caused by dimensional difference.

[0071] The standardized multi-dimensional performance data is used as the training data set, where each data point contains performance parameters such as torque, vibration frequency, and temperature under different load and speed conditions. The target output value is the running-in state of the equipment (such as vibration intensity, mechanical stress, temperature rise rate, etc.), and these output values ​​are used to judge the degree of running-in. The mean square error (MSE) is used as the loss function for model training to optimize the parameters and penalty factors of the SVR model to ensure that the model has good generalization ability for the prediction of the running-in process.

[0072] After training, the SVR model predicts the running-in status under different working conditions based on the characteristic data of new and old components. The corresponding running-in curve is generated through the predicted running-in status, which reflects the relationship between load change, speed adjustment and running time. The specific generation method is as follows: for different load conditions, the SVR model outputs the corresponding running-in status values ​​(vibration, temperature rise, stress), and according to these outputs, the load change rate and speed adjustment step are set to make the running-in process proceed smoothly under different load conditions. For example, assuming that under a certain load, the model predicts that the vibration is too large, then the load change rate under the load is set to a smaller value, and the speed step is adjusted to reduce the impact of vibration on the components.

[0073] In the generated running-in curve, the load change rate and speed adjustment step size under each operating condition are optimized based on the performance prediction results. For example, the calculation formula for the load change rate ΔP is:

[0074] Among them, P final and P initial are the final load and initial load respectively, and T is the preset running-in period. The speed adjustment step Δn is calculated according to the vibration frequency change predicted by the model and is calculated as follows:

[0075] Among them, n final and n initial These are the final and initial speeds respectively. Through this step size, the speed is gradually adjusted to make the running-in process of new and old parts smoother.

[0076] The running-in time is determined based on the running-in state change curve of the equipment. The running-in time predicted by the SVR model is corrected in combination with historical data to ensure that the running-in time under each load condition is long enough to achieve a good match of the components. The running-in time calculation formula is:

[0077] Among them, error is the difference between the current running-in state and the ideal state, α and β are constants adjusted according to historical data, and K is the maximum running-in time. This formula can be used to dynamically adjust the running-in time under different load conditions to ensure that the components are fully run-in.

[0078] A personalized running-in curve is generated based on the performance data of new and old components and the SVR model, setting the load change rate, speed adjustment step and operating time to provide precise control parameters for the subsequent running-in process.

[0079] During the running-in process of the new and old parts of the motor, as the equipment load, speed and other operating conditions change, the performance differences between the parts may cause problems such as temperature rise, increased vibration or current fluctuations. If these parameters exceed the predetermined safety threshold, it may accelerate the wear of the components and even cause mechanical failure. Therefore, it is particularly important to monitor the operating status of the motor in real time and respond to parameter changes in a timely manner. Through the adaptive control algorithm, abnormal fluctuations in the operation of the equipment can be adjusted in real time to ensure that the load and speed are optimized in time during the running-in process to avoid damage to other parts of the equipment.

[0080] Step S3 includes the following contents:

[0081] During the running-in process, sensors installed on key components of the equipment monitor operating parameters such as temperature, vibration and current in real time. Temperature sensors are arranged on windings, bearings and rotors, vibration sensors are installed on the motor housing and support structure, and current sensors are located at the power input. All sensors transmit data to the control system wirelessly or wired. In order to ensure the accuracy of the data, a time synchronization protocol (such as IEEE 1588PTP) is used to synchronously collect data from different sensors to ensure that the correlation between various parameters is accurately reflected.

[0082] Before the data collected by the sensor enters the control system, it is necessary to filter out noise and outliers. Use Kalman filter to filter temperature, vibration and current signals in real time, remove external noise or instantaneous interference signals, and retain valid operation data. For current fluctuation signals, use Fourier transform to convert current signals from time domain to frequency domain, and identify abnormal harmonics or interference signals by analyzing their spectral components.

[0083] After filtering and preprocessing, the real-time sensor data enters the adaptive control algorithm for threshold judgment. First, a set of threshold ranges are preset according to the operating characteristics of the equipment. The thresholds include the upper temperature limit, the upper vibration frequency limit, and the safety range of current fluctuations. The specific definitions are as follows:

[0084] Temperature threshold: T max , whose value is determined by the material properties of the equipment and the operating load. For example, the safety temperature threshold of the winding is set at 130°C.

[0085] Vibration frequency threshold: f max , according to the mechanical structure of the motor, for example, the vibration frequency should not exceed 100Hz.

[0086] Current fluctuation threshold: THD, which is monitored by the ratio of the peak amplitude to the fundamental amplitude of the current.

[0087] Once any parameter exceeds the preset threshold, the adaptive control logic is entered, and the load and speed are dynamically adjusted according to real-time data. The adjustment strategy is executed according to the type and degree of the parameter exceeding the threshold:

[0088] Temperature exceeds threshold: If the temperature parameter T is greater than the temperature threshold, reduce the load change rate. The adjustment formula is: Among them, α T is the temperature adjustment coefficient.

[0089] Vibration frequency exceeds the threshold: If the vibration frequency parameter f is greater than the vibration frequency threshold, adjust the speed adjustment step size to return the vibration frequency to a safe range. The adjustment formula is: Among them, α f is the vibration frequency adjustment coefficient.

[0090] Current fluctuation exceeds the threshold: When the current fluctuation exceeds the safety range (for example, THD is greater than 20%), reduce the load change rate and appropriately reduce the speed. The adjustment strategy is similar to the treatment of temperature and vibration.

[0091] After each parameter adjustment, the control system will compare the adjusted load and speed with the actual operating status of the equipment and update the algorithm adjustment coefficient (such as α T , α f ) to optimize the accuracy of subsequent adjustments. The Q-learning algorithm in reinforcement learning is used to update the optimal adjustment strategy under different load conditions according to the device response after each adjustment, so that the control system can continuously optimize the adjustment logic in long-term operation.

[0092] By using sensors to collect the equipment's temperature, vibration spectrum, and current fluctuation data in real time, and using adaptive control algorithms to perform real-time threshold judgment and dynamic adjustment, it is ensured that all parameters of the motor remain within a safe range during the running-in process, thus avoiding abnormal wear or failure.

[0093] During the motor running-in process, the vibration signal is an important parameter that reflects the operating status of the equipment and the running-in of the components. Due to the insufficient running-in of the new and old components, abnormal changes in vibration frequency and amplitude may occur, and these abnormalities are often manifested in the two dimensions of time domain and frequency domain. It may be difficult to capture these complex dynamic signal characteristics by relying solely on time domain or frequency domain analysis, so it is necessary to combine wavelet transform and fast Fourier transform (FFT) to conduct a comprehensive analysis of the vibration signal. By extracting the time-frequency characteristics of the vibration signal and applying pattern recognition algorithms, potential abnormalities in the running-in process, such as mechanical imbalance, rotor misalignment, looseness or local damage, can be effectively detected.

[0094] Step S4 includes the following contents:

[0095] During the operation of the motor, the vibration sensor collects mechanical vibration data in real time and records the vibration waveform of the equipment under different load conditions. These signals are stored in the form of time domain data, indicating the change in vibration intensity of the equipment over time. Assume that the vibration signal is x(t), where t is time and x is the amplitude of the vibration. After each load adjustment, the vibration signal needs to be re-collected to ensure data continuity.

[0096] After the old parts are replaced, the frequency sharpness index and the time series entropy difference index are obtained based on the vibration signals generated by the operation after the new parts are installed.

[0097] Among them, the acquisition logic of the spectrum sharpness index is:

[0098] First, the vibration signal x(t) in the time domain is converted to the frequency domain to obtain the spectral components of the vibration signal. For the vibration signal, the time domain signal x(t) is defined as:

[0099] Among them, A is the amplitude of the main vibration signal, f0 is the main frequency, A n is the amplitude of other minor frequencies, f n are the secondary frequencies, φ and φ n is the phase angle.

[0100] The spectrum components are obtained through frequency domain transformation, and the spectrum is S(f), where f is the frequency and S(f) is the vibration intensity at the frequency f. At this time, the frequency component distribution of the signal is nonlinear.

[0101] Quantify the distribution characteristics of the spectrum and construct the frequency distribution function F(f), which reflects the relative distribution of each frequency component. The definition is as follows:

[0102] This formula represents the ratio of the energy at frequency f to the energy of the entire spectrum. max is the highest frequency range of the spectrum, |S(f)| 2 is the energy intensity at frequency f. By normalizing, the total energy distribution is ensured to be between 0 and f max between.

[0103] Calculate the spectrum sharpness increment function to reflect the sudden change of local frequency energy in the spectrum. The local energy change rate based on frequency is expressed as follows:

[0104] Where Δf is the frequency interval, which indicates the rate of change of local energy within the frequency range. If a certain frequency segment has a large energy increase, the sharpness increment will be large, reflecting the sharp characteristics of the spectrum in this segment.

[0105] Taking all the sharpness increments at all frequencies into consideration, the spectrum sharpness index SPI is calculated. By integrating the frequency sharpness increment and taking the curvature of the local sharpness increment as a correction term into the calculation, we get:

[0106] The formula accumulates the sharpness contribution of each frequency band by integration, and the square of the exponent is used to amplify the frequency band with higher sharpness, making the index more sensitive to locally prominent spike signals.

[0107] in, It is the second-order derivative at frequency f, reflecting the curvature change of frequency energy distribution. The introduction of this correction term can significantly amplify the frequency band with drastic energy changes in the vibration signal (such as the harmonic components caused by faults), thereby improving the ability to capture abnormal signals.

[0108] After the new component is replaced, the spectrum sharpness index is used to analyze the vibration signals generated by the new and old components during the running-in process, quantify the concentration and sharpness of its spectrum distribution, and evaluate the running-in effect between components. This index can reflect whether the replaced new component is mismatched with the existing old component or has local abnormalities. The larger the index, the more energy there is in a specific frequency band, which may be caused by local abnormalities caused by the incompatibility between the mechanical characteristics of the new component and the old component, such as mechanical imbalance, loose structure or excessive local friction, indicating potential failure risks. The smaller the index, the smoother the vibration signals of the new and old components, the more evenly distributed the energy, the more stable the running-in process, and no significant abnormal vibration occurs, indicating that the replaced new component matches the old component well.

[0109] Among them, the logic for obtaining the time series entropy difference index is:

[0110] The continuous vibration signal x(t) is converted into a discrete symbol sequence in order to calculate the dynamic characteristics of the symbol sequence.

[0111] The amplitude of the vibration signal is divided into M intervals according to a predetermined threshold and mapped into a symbol set {s1, s2, …, s M}:s k =Φ(x(t i ))=s k , when τ k-1 <x(t i )≤τ k ; where τ0,τ1,…,τ M is the partition threshold, satisfying τ0<τ1<…<τ M,k=1,2,…,M。

[0112] According to the statistical characteristics of the vibration signal, the threshold τ is set k :τ k =τ0+k·Δτ; where, Δτ=(τ M -τ0) / M, ensuring that the symbol intervals are of equal width or adjusted according to signal characteristics.

[0113] Using the symbol sequence, the transition probability between symbols is calculated to reflect the dynamic changes of the time series.

[0114] Calculation symbol s i Transfer to symbol s j The number of Where δ is the indicator function, which takes 1 when the condition is met and takes 0 otherwise, and T is the length of the symbol sequence.

[0115] Calculate the transition probability matrix Ensure that the transition probability of each symbol satisfies the normalization condition:

[0116] Through the transfer probability matrix, the entropy difference between the current vibration signal and the normal state signal is calculated to obtain the time series entropy difference index. Calculate the current transfer entropy

[0117] Using the vibration signal under normal operation, the corresponding transfer entropy H is calculated in the same way. normal .

[0118] Calculate the time series entropy difference index TSEDI: TSEDI = |HH normal |.

[0119] The entropy difference index reflects the difference between the current signal complexity and the normal state.

[0120] The time series entropy difference index is used to quantify the change in the complexity of the vibration signal during the running-in process of new and old parts, reflecting the difference in complexity between the current vibration signal and the signal under normal operating conditions. The larger the index, the greater the difference in complexity between the current signal and the normal state, which may mean that the instability during the running-in process is aggravated, and there are potential abnormal running-in problems between components, such as mechanical imbalance, uneven friction or local damage. The smaller the index, the closer the complexity of the current signal is to the normal state, indicating that the new and old parts are well matched during the running-in process, the operating state is stable, and no significant abnormal vibration characteristics appear.

[0121] Two gain functions are calculated for the spectrum sharpness index and the time series entropy difference index, respectively, to indicate their sensitivity to potential anomalies in different dimensions:

[0122] The gain function enhances the abnormal state response of both and makes the abnormal peak more prominent through the nonlinear form of the function.

[0123] Combine the time-frequency gain functions into a time-frequency gain fusion index

[0124] By combining the gain effects of the sum and square root terms, the potential abnormal signal changes in the time-frequency domain are amplified, which is particularly suitable for the analysis of vibration signals with concentrated frequencies and high time complexity.

[0125] The time-frequency gain fusion index is mainly used to evaluate the vibration signal after the replacement of new parts. By analyzing the gain effects in the frequency domain and time domain, it can identify and detect potential abnormalities during the running-in process. The index rises rapidly when the frequency concentration or time complexity of the vibration signal changes significantly, reflecting possible problems such as mismatch between the new and old parts, mechanical imbalance or local wear. The larger the time-frequency gain fusion index, the greater the abnormality of the frequency domain and time domain characteristics of the vibration signal during the running-in of the new and old parts, the frequency concentration or the significant increase in time complexity, which may reflect problems such as mechanical imbalance, excessive friction or local structural looseness between the parts. The smaller the index, the more stable the vibration signal is in the two dimensions of time and frequency, the uniform energy distribution, the good running-in effect between the parts, the stable operation of the equipment, and no significant abnormalities. By paying attention to the two dimensions of time and frequency at the same time, the index can comprehensively evaluate the operating status of the equipment, help identify potential vibration anomalies, and provide strong decision support for maintenance personnel to guide subsequent running-in adjustments or maintenance work.

[0126] Compare the time-frequency gain fusion index with the upper and lower thresholds; if the time-frequency gain fusion index is greater than or equal to the upper threshold, it means that there are obvious abnormalities in the running-in process, which may be due to the mismatch between the new and old parts, causing mechanical imbalance, structural looseness or excessive friction, indicating that the newly replaced parts are obviously not suitable for the current use status of the equipment, and the maintenance abnormality status is displayed; if the time-frequency gain fusion index is greater than or equal to the lower threshold and less than the upper threshold, although the vibration signal has no significant abnormality, there may be potential hidden dangers in the running-in process, and the running-in situation needs to be further analyzed and the operating parameters adjusted to ensure safety, showing the maintenance ambiguous status; if the time-frequency gain fusion index is less than the lower threshold, it means that the running-in process of the new and old parts is stable, the vibration signal is stable, the equipment is in good operating condition, there is no obvious abnormality, the running-in effect is ideal, and the maintenance is normal.

[0127] If the maintenance ambiguous state is obtained, it means that there are potential risks in the running-in process of the new and old parts of the motor. Although the vibration signal does not show clear abnormalities, it may cause running-in problems or equipment failures in long-term operation. At this time, it is necessary to further adjust the control strategy of the equipment, especially the PID parameters of the controller, to ensure that the running-in process of the new and old parts is smoother by optimizing the response characteristics of the equipment to vibration and load changes, and avoid long-term wear or failure caused by local mismatch. At this time, step S5 is executed, and the specific processing process is as follows:

[0128] Step S5 includes the following contents:

[0129] Obtain the proportional, integral and differential parameters of the current controller, monitor the speed, torque and vibration signal amplitude of the equipment in real time, and analyze the response curve during load changes as a basis for subsequent PID parameter adjustments.

[0130] In order to solve the response speed problem in the equipment running-in process, the time-frequency gain fusion index of the vibration signal is adjusted in real time. The adjustment method is to multiply the original proportional coefficient by a factor related to the time-frequency gain fusion index. The size of the factor depends on the position of the index between the upper and lower thresholds. The role of this factor is to increase the proportional coefficient when the time-frequency gain fusion index is close to the upper limit, thereby improving the system's sensitivity to changes; when the index is close to the lower limit, the proportional coefficient is reduced to make the system more stable. The formula is: The device's sensitivity to load changes is dynamically corrected by adjusting the proportional coefficient.

[0131] K p 'Adjusted proportional coefficient. Used to control the response strength of the system to the error.

[0132] K p The original proportional coefficient determines the system's sensitivity to the error signal. The larger the coefficient, the faster the system responds.

[0133] The adjustment coefficient is used to control the adjustment range of the proportional coefficient. This coefficient is set according to the specific equipment characteristics and the needs of the running-in process, and is generally used to enlarge or reduce the adjustment amount.

[0134] θ low The lower threshold is used to determine whether the vibration signal is within the normal range. If the time-frequency gain fusion index is close to this value, it means that the equipment operation is stable.

[0135] θ high The upper threshold is used to determine the abnormal state of the vibration signal. When the time-frequency gain fusion index approaches or exceeds this value, it indicates that there is a significant abnormality in the system.

[0136] The normalization expression normalizes the position of the current index relative to the upper and lower thresholds to determine the magnitude of the adjustment.

[0137] In view of the phenomenon of equipment accumulating errors, the integral coefficient is adjusted to eliminate the steady-state error. The adjustment method is to make an additive adjustment to the original integral coefficient based on the total accumulated errors of the equipment during operation. The larger the accumulated error, the larger the adjustment range of the integral coefficient, thereby gradually eliminating the long-term deviation in the system, making the running-in of new and old parts smoother, and ensuring the stability of the long-term operation of the equipment. The formula is: Ensure that the system can stably return to the ideal state after long-term operation and avoid long-term fluctuations in vibration signals.

[0138] K l 'Adjusted integral coefficient. It is used to eliminate the steady-state error in the system. The error accumulated over a long period of time will be corrected by this coefficient.

[0139] K l The original integral coefficient is used to adjust the error elimination speed of the equipment during long-term operation. The larger the integral coefficient, the faster the steady-state error elimination speed.

[0140] The integral adjustment coefficient determines the adjustment range of the integral term. This parameter is used to dynamically adjust the size of the integral term according to the specific running-in status.

[0141] The error cumulative sum represents the sum of the error values ​​from the initial time τ = 0 to the current time t. The error value e(τ) is the difference between the set value and the actual value.

[0142] T adjust Adjust the time constant to determine the time scale of the integration process and control the speed and amplitude of the integral coefficient adjustment.

[0143] For sudden vibration or fluctuations in the short term of the equipment, the instantaneous overshoot is suppressed by adjusting the differential coefficient. The adjustment method is to adjust the differential coefficient according to the instantaneous change rate of the time-frequency gain fusion index. When the vibration signal of the equipment changes rapidly, the differential coefficient is reduced to suppress the fluctuation amplitude, ensure the smooth operation of the equipment, and avoid the impact of short-term fluctuations on the overall running-in process. The formula is: By adjusting the differential term, the impact of short-term fluctuations on system stability can be offset.

[0144] K d 'The adjusted differential coefficient is used to suppress instantaneous fluctuations in the system and reduce overshoot.

[0145] K dThe original differential coefficient is used to control the reaction speed of the system to the instantaneous error change. The larger the differential coefficient, the stronger the system's reaction to sudden fluctuations.

[0146] The differential adjustment coefficient determines the adjustment amplitude of the differential term and controls the sensitivity of the control system to instantaneous fluctuations of the vibration signal.

[0147] The time rate of change of the time-frequency gain fusion index. It is used to detect the instantaneous change speed of the current vibration signal and reflect the violent fluctuation of the signal in a short period of time. If the value is large, it means that the system is in an unstable state and the differential coefficient needs to be reduced for smooth operation.

[0148] The present invention proposes an effective solution for the running-in process of new and old parts of the motor. First, by collecting multi-dimensional performance parameters of new and old parts and building a performance feature database, the key parameter differences of different parts can be described in detail, especially the factors affecting the running-in effect such as speed, torque and electrical impedance. Secondly, a personalized running-in curve is generated by using a support vector machine regression algorithm, and the load change rate, speed adjustment step and running-in time are accurately set according to the actual operating conditions, avoiding the vibration and current fluctuation problems caused by the initial mismatch between the new and old parts. In addition, the temperature, vibration spectrum and current fluctuation of the equipment are monitored in real time to ensure that when any parameter exceeds the threshold during the running-in process, the adaptive control algorithm can automatically adjust the load and speed to ensure a smooth and orderly running-in process. By extracting frequency domain and time domain features, calculating the spectrum sharpness index and the time series entropy difference index, and combining the time-frequency gain fusion index, the running-in status is classified and analyzed, which can effectively identify the potential abnormalities of the equipment after the new parts are replaced, and present abnormal, ambiguous or normal maintenance status. Especially in the ambiguous state, by adjusting the PID parameters of the equipment controller, the running-in process of new and old parts is further optimized to ensure that the performance of new and old parts gradually matches, avoiding long-term performance degradation and equipment stability problems caused by insufficient running-in. This technical solution not only realizes real-time monitoring and dynamic adjustment of the running-in period, solving the problem of ignoring the running-in period of new and old parts in the existing technology, but also improves the operating stability and long-term reliability of motor equipment by optimizing the running-in process, and prevents secondary failures and mechanical damage caused by improper running-in.

[0149] Embodiment 2: Figure 2 The invention provides an intelligent electrical equipment maintenance task decision support system, including: a performance acquisition module, an algorithm processing module, a monitoring and adjustment module, a signal analysis module and a parameter optimization module;

[0150] Performance acquisition module: collects the performance parameters of new and old components, builds a multi-dimensional performance characteristic database, and outputs detailed performance characteristic data of new and old components, which is passed to the algorithm processing module.

[0151] Algorithm processing module: Based on the collected multi-dimensional performance characteristic data, the support vector machine regression algorithm is used to generate a personalized running-in curve, set the load change rate, speed adjustment step and running time during the running-in period, output the running-in curve and the load and speed adjustment plan, and pass it to the monitoring and adjustment module.

[0152] Monitoring and adjustment module: monitors the temperature, vibration spectrum and current fluctuation of the equipment in real time, uses control algorithms to dynamically adjust the load and speed when the parameters exceed the preset thresholds, outputs the real-time monitoring data of the equipment and the adjusted operating status, and transmits it to the signal analysis module.

[0153] Signal analysis module: Based on the vibration signal in the monitoring data, it extracts frequency domain and time domain features, calculates the spectrum sharpness index and time series entropy difference index, combines the time-frequency gain fusion index to identify the abnormal state of equipment running-in, outputs the abnormal classification, and passes it to the parameter optimization module.

[0154] Parameter Optimization Module: If an ambiguous state is obtained, the operating parameters during the running-in process are optimized by adjusting the PID parameters of the equipment controller, and the optimized PID parameter settings are output to ensure that the performance of the new and old parts gradually matches.

[0155] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0156] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0157] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including a..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0158] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An intelligent electrical equipment maintenance task decision support method, characterized in that: Includes steps: S1, collect performance parameters of new and old components and build a multi-dimensional performance characteristic database; S2, uses the support vector machine regression algorithm to process multi-dimensional performance characteristic data, generates a personalized running-in curve, and sets the load change rate, speed adjustment step size, and operating time during the running-in period; S3, real-time monitoring of the equipment’s temperature, vibration spectrum, and current fluctuations, and using control algorithms to dynamically adjust load and speed when parameters exceed thresholds; S4, based on the frequency domain and time domain feature extraction of the vibration signal during the running-in process of the new and old parts, the spectrum sharpness index and the time series entropy difference index are calculated, and the potential running-in anomaly after the replacement of the new parts of the equipment is identified through the time-frequency gain fusion index, and the maintenance abnormality state, ambiguous state or normal maintenance state are displayed; S5, if a maintenance ambiguous state is obtained, the PID parameters of the equipment controller are adjusted to optimize the operating parameters during the running-in process.

2. The intelligent electrical equipment maintenance task decision support method according to claim 1 is characterized by: Step S2 includes the following contents: First, the multi-dimensional performance data collected in step S1 is standardized; The standardized multi-dimensional performance data is used as the training data set, the target output value is the running-in state of the equipment, and the mean square error is used as the loss function for model training to optimize the parameters and penalty factors of the SVR model. After training is completed, the SVR model predicts the running-in status under different working conditions based on the characteristic data of new and old parts; and generates the corresponding running-in curve based on the predicted running-in status.

3. The intelligent electrical equipment maintenance task decision support method according to claim 2 is characterized by: Step S3 includes the following contents: A set of threshold ranges are preset according to the operating characteristics of the equipment, including the upper temperature limit, the upper vibration frequency limit, and the safe range of current fluctuations. Once any parameter exceeds the preset threshold, the adaptive control logic is entered, and the load and speed are dynamically adjusted according to real-time data. Adjustment strategies are executed based on the type and degree of parameter exceeding the threshold.

4. The intelligent electrical equipment maintenance task decision support method according to claim 3 is characterized by: Step S4 includes the following contents: During the operation of the motor, mechanical vibration data is collected in real time and the vibration waveform of the equipment under different load conditions is recorded; the vibration signal is assumed to be x(t), where t is time and x represents the amplitude of the vibration; after the old parts are replaced, the frequency sharpness index and the time series entropy difference index are obtained based on the vibration signal generated by the installation of the new parts.

5. The intelligent electrical equipment maintenance task decision support method according to claim 4 is characterized by: in, The logic for obtaining the spectrum sharpness index is: First, the vibration signal x(t) in the time domain is converted to the frequency domain to obtain the spectral components of the vibration signal. For the vibration signal, the time domain signal x(t) is defined as: Among them, A is the amplitude of the main vibration signal, f0 is the main frequency, A n is the amplitude of other minor frequencies, f n are the secondary frequencies, φ and φ n is the phase angle; The spectrum components are obtained by frequency domain transformation, and the spectrum is S(f), where f is the frequency and S(f) is the vibration intensity at frequency f: The distribution characteristics of the spectrum are quantified and the frequency distribution function F(f) is constructed, which reflects the relative distribution of each frequency component; it is defined as follows: Among them, f max is the highest frequency range of the spectrum, |S(f)1 2 is the energy intensity at frequency f; normalized to ensure that the total energy is distributed between 0 and f max between; Calculate the spectral sharpness delta function, based on the local energy change rate of frequency, expressed as follows: Where Δf is the frequency interval; Taking all the sharpness increments at all frequencies into consideration, the spectrum sharpness index SPI is calculated. By integrating the frequency sharpness increment and taking the curvature of the local sharpness increment as a correction term into the calculation, we get: in, It is the second-order derivative at frequency f, reflecting the change in curvature of the frequency energy distribution.

6. The intelligent electrical equipment maintenance task decision support method according to claim 5 is characterized by: in, The logic for obtaining the time series entropy difference index is: Convert the continuous vibration signal x(t) into a discrete symbol sequence; The amplitude of the vibration signal is divided into M intervals according to a predetermined threshold and mapped into a symbol set {s1, s2, ..., s M }:s k =Φ(x(t i ))=B k , when τ k-1 <x(t i )≤τ k ;where, τ0, τ1,…,τ M is the partition threshold, satisfying τ0<τ1<…<τ M , k = 1, 2, ..., M; According to the statistical characteristics of the vibration signal, the threshold τ is set k :τ k =τ0+k·Δτ; where, Δτ=(τ M -τ0) / M, to ensure that the symbol intervals are of equal width or adjusted according to signal characteristics; Using the symbol sequence, calculate the transition probability between symbols; Calculation symbol s i Transfer to symbol s j The number of times N ij : Where δ is the indicator function, which takes 1 when the condition is met and takes 0 otherwise, and T is the length of the symbol sequence; Calculate the transition probability matrix P: Ensure that the transition probability of each symbol satisfies the normalization condition: Through the transfer probability matrix, the entropy difference between the current vibration signal and the normal state signal is calculated to obtain the time series entropy difference index; the current transfer entropy H is calculated: Using the vibration signal under normal operation, the corresponding transfer entropy H is calculated in the same way. normal ; Calculate the time series entropy difference index TSEDI: TSEDI = |HH normal |.

7. The intelligent electrical equipment maintenance task decision support method according to claim 6 is characterized by: Step S4 includes the following contents: Two gain functions are calculated for the spectrum sharpness index and the time series entropy difference index, respectively, to indicate their sensitivity to potential anomalies in different dimensions: Combine the time-frequency gain functions into a time-frequency gain fusion index VAF: Compare the time-frequency gain fusion index with the upper and lower thresholds; If the time-frequency gain fusion index is greater than or equal to the upper threshold, a maintenance abnormality state is displayed; If the time-frequency gain fusion index is greater than or equal to the lower threshold and less than the upper threshold, a maintenance ambiguity state is displayed; if the time-frequency gain fusion index is less than the lower threshold, a maintenance normal state is displayed.

8. The intelligent electrical equipment maintenance task decision support method according to claim 7 is characterized by: Step S5 includes the following contents: Obtain the proportional, integral and differential parameters of the current controller, monitor the speed, torque and vibration signal amplitude of the equipment in real time, and analyze the response curve during load changes as a basis for PID parameter adjustment; To address the response speed issue during equipment running-in, the time-frequency gain fusion index of the vibration signal is adjusted in real time. This is to multiply the original proportional coefficient by a factor related to the time-frequency gain fusion index. The size of the factor depends on the position of the index between the upper and lower thresholds. In view of the phenomenon of equipment accumulating errors, the integral coefficient is adjusted to eliminate the steady-state error. This is to make an additive adjustment to the original integral coefficient based on the total accumulated errors of the equipment during operation. In response to sudden vibration or fluctuation of the equipment in the short term, the instantaneous overshoot is suppressed by adjusting the differential coefficient. The differential coefficient is adjusted according to the instantaneous change rate of the time-frequency gain fusion index.

9. An intelligent electrical equipment maintenance task decision support system, used to implement the intelligent electrical equipment maintenance task decision support method according to any one of claims 1 to 8, characterized in that: include: Performance acquisition module, algorithm processing module, monitoring and adjustment module, signal analysis module and parameter optimization module; Performance acquisition module: collects the performance parameters of new and old parts, builds a multi-dimensional performance characteristic database, outputs detailed performance characteristic data of new and old parts, and passes it to the algorithm processing module; Algorithm processing module: Based on the collected multi-dimensional performance characteristic data, the support vector machine regression algorithm is used to generate a personalized running-in curve, set the load change rate, speed adjustment step and running time during the running-in period, output the running-in curve and load and speed adjustment plan, and pass it to the monitoring and adjustment module; Monitoring and adjustment module: monitors the temperature, vibration spectrum and current fluctuation of the equipment in real time, uses control algorithms to dynamically adjust the load and speed when the parameters exceed the preset thresholds, outputs the real-time monitoring data of the equipment and the adjusted operating status, and transmits it to the signal analysis module; Signal analysis module: Based on the vibration signal in the monitoring data, it extracts frequency domain and time domain features, calculates the spectrum sharpness index and time series entropy difference index, combines the time-frequency gain fusion index to identify the abnormal state of equipment running-in, outputs the abnormal classification, and passes it to the parameter optimization module; Parameter Optimization Module: If an ambiguous state is obtained, the operating parameters during the running-in process are optimized by adjusting the PID parameters of the equipment controller, and the optimized PID parameter settings are output to ensure that the performance of the new and old parts gradually matches.

Images