High-precision identification method and system for noise source of asynchronous motor

Abstract

This invention discloses a high-precision identification method and system for noise sources in asynchronous motors. The invention relates to the field of asynchronous motor detection technology and includes the following steps: constructing a finite element model of the asynchronous motor to be analyzed, dividing the surface into multiple detection sub-regions, performing finite element motor operation simulation, analyzing vibration velocity and sound pressure signal data of each sub-region, synchronously acquiring actual signals, calculating the relative difference between the actual and simulated data, evaluating detection priority based on historical fault counts, forming a detection sequence according to the priority, sequentially performing acoustic-vibration coherence analysis, calculating the coherence function, classifying the noise frequency band into structural noise and air noise, performing order tracking analysis on the classified structural noise, calculating the amplitude stability index and amplitude change gradient, and achieving automatic classification and identification of electromagnetic noise and mechanical noise based on a preset threshold, significantly improving the efficiency and accuracy of fault handling.

Description

Technical Field

[0001] This invention relates to the field of asynchronous motor detection technology, specifically to a high-precision identification method and system for asynchronous motor noise sources. Background Technology

[0002] With the continuous advancement of industrialization, electric motors, as indispensable power equipment in industrial production, have a crucial impact on the efficiency and stability of the entire production system due to their performance and reliability. However, in the actual operation of electric motors, vibration and noise have become important factors affecting the quality of motor operation, especially asynchronous motors. As they are widely used in various industrial scenarios, the vibration and noise problems generated during their operation are more obvious. Therefore, how to effectively identify the noise sources and causes of asynchronous motors, and further distinguish the types of noise, is a technical problem that urgently needs to be solved.

[0003] Currently, the analysis and identification of motor noise sources mainly rely on traditional acoustic diagnostic methods and mechanical vibration analysis techniques. While these methods can reveal the sources of noise during motor operation to some extent, the complex physical field coupling effects during motor operation, such as the interaction between electromagnetic fields, mechanical vibration fields, and sound fields, make it difficult to achieve accurate classification and efficient identification of noise sources by relying solely on traditional methods. Furthermore, the causes of noise during motor operation may involve multiple types, such as electromagnetic noise, mechanical noise, and air noise. These noises have significantly different frequency characteristics, generation mechanisms, and degrees of influence, further increasing the technical difficulty of noise source identification.

[0004] In existing technologies, the identification of motor noise sources typically involves comparing and analyzing the structural parameters of various components in conjunction with the vibration conditions during motor operation. CN119106384A discloses a high-precision identification method, system, and storage medium for asynchronous motor noise sources. This method collects relevant parameters of the inner ring, outer ring, cage, and rolling elements of the motor's bearings, and performs abnormal frequency analysis on each of these components. It outputs abnormal frequency indices for the bearing inner ring, outer ring, rolling elements, and stiffness, reflecting the abnormal frequencies of these components respectively. Simultaneously, it collects vibration signals from the motor, processes these signals to generate frequency domain signals, and obtains... The spectrum, by analyzing the relationship between frequency amplitude and combining it with abnormal frequency indices, generates rolling element anomaly assessment indices, stiffness anomaly assessment indices, bearing inner ring assessment indices, and bearing outer ring anomaly assessment indices, respectively reflecting the degree of anomalies in different parts. This helps operators determine the specific location of bearing faults. However, this approach mainly focuses on the abnormal frequency analysis of components such as the inner ring, outer ring, cage, and rolling elements of the bearing, neglecting the influence of other important components in the asynchronous motor on the overall noise source, leading to a one-sided approach to motor noise source identification. Furthermore, it fails to consider the mutual influence and coupling effects between electromagnetic noise, mechanical noise, and air noise during motor operation, lacking a comprehensive perspective on noise source identification. Therefore, the accuracy and effectiveness of the identification system are reduced.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a high-precision identification method and system for asynchronous motor noise sources, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A high-precision identification method for noise sources in asynchronous motors, comprising the following steps:

[0009] A finite element model of the asynchronous motor to be analyzed is constructed. The surface of the finite element model of the asynchronous motor to be analyzed is divided into several detection sub-regions. Finite element motor operation simulation is performed, and vibration velocity signal and sound pressure signal data of each detection sub-region during operation are analyzed.

[0010] The actual vibration velocity signal and sound pressure signal characteristic parameters of each sub-region are collected synchronously, the relative difference with the finite element simulation data is calculated, and the detection priority score of each sub-region is calculated by weighted fusion algorithm based on the historical failure number of the associated components of each sub-region.

[0011] Based on the detection priority score of each detection sub-region, a detection sequence is formed. Based on the order of the detection sequence, acoustic-vibration coherence analysis is performed on each detection sub-region in turn, the digital coherence function of the acoustic-vibration signal is calculated, and the noise frequency band of each detection sub-region is specifically classified into structural noise and air noise according to the preset coherence threshold.

[0012] The classification of structural noise frequency bands is analyzed by order tracking, and the amplitude stability index and amplitude change gradient of each order line are calculated. Based on the preset decision threshold, the automatic classification and identification of electromagnetic noise and mechanical noise are realized.

[0013] Furthermore, the specific method for constructing the finite element model of the asynchronous motor to be analyzed is as follows: Based on the design parameters of the asynchronous motor, specifically including the dimensions and shapes of the stator, rotor, shaft, and end cover, the geometric model is determined. CAD software is used for modeling, and the physical properties of the materials of each component, including the stator, rotor, core, and insulation materials, are determined. These physical properties include density, elastic modulus, Poisson's ratio, and conductivity. In the finite element software, the material properties are assigned to the geometric model to form the finite element model.

[0014] The logic behind dividing the surface of the asynchronous motor finite element model to be analyzed into several detection sub-regions based on the finite element model is as follows: the surface of the finite element model is divided into several detection sub-regions with equal area.

[0015] The specific method used to perform finite element motor operation simulation and analyze the vibration velocity and sound pressure signals during the operation of each detection sub-region is as follows: the motor is operated at constant torque and speed, and the vibration velocity and sound pressure signals during the operation of each detection sub-region are collected. The constant torque and speed are specifically set based on the working torque and speed of the asynchronous motor to be analyzed.

[0016] Furthermore, the characteristic parameters of the actual vibration velocity signal and sound pressure signal of each sub-region include the average value and corresponding peak value of vibration velocity and sound pressure level within the same rotational speed range;

[0017] The logic behind calculating the relative difference between the data and the finite element simulation data is as follows: A relative difference coefficient is calculated by comparing the average value, corresponding peak value, and integral value within different speed ranges. This coefficient is then used to characterize the relative differences between different detection sub-regions. The specific formula for calculating the relative difference coefficient is as follows:

[0018]

[0019] In the formula, Let be the relative difference coefficient of the i-th detection sub-region. The relative difference in the mean vibration velocity within the i-th detection sub-region. The relative difference in the mean sound pressure level within the i-th detection sub-region. Let be the difference between the actual vibration velocity and the finite element simulation vibration velocity within the i-th detection sub-region at time t. Let be the difference between the actual sound pressure level and the finite element simulation sound pressure level at time t within the i-th detection sub-region; where i is the index of the detection sub-region and t is the time variable of the finite element motor operation simulation process. This represents the starting moment of the finite element motor operation simulation process. This is the termination time of the finite element motor operation simulation process;

[0020] Where calculation and The specific formula used is as follows:

[0021]

[0022] In the formula, Let be the average actual vibration velocity of the i-th detection sub-region. Let be the average vibration velocity from the finite element simulation of the i-th detection sub-region. Let be the average actual sound pressure level of the i-th detection sub-region. Let be the average sound pressure level of the finite element simulation of the i-th detection sub-region. The maximum vibration velocity in the i-th detection sub-region is... The maximum sound pressure level in the i-th detection sub-region;

[0023] The maximum vibration velocity of the i-th detection sub-region and the maximum sound pressure level of the i-th detection sub-region Specifically, it refers to the maximum value of the peak values ​​corresponding to vibration velocity and sound pressure level in finite element simulation and actual detection;

[0024] Where calculation and The specific formula used is as follows:

[0025]

[0026] In the formula, and Let be the actual vibration velocity and the finite element simulation vibration velocity of the i-th detection sub-region at time t, respectively. and Let be the actual sound pressure level and the finite element simulated sound pressure level of the i-th detection sub-region at time t, respectively. and t represents the maximum vibration velocity and sound pressure level at time t, respectively.

[0027] Furthermore, the logic behind calculating the detection priority score for each sub-region using the weighted fusion algorithm is as follows: The number of historical faults collected is normalized, and the detection priority score for each sub-region is calculated based on the processed data and the relative difference coefficient of the detection sub-regions. The specific formula for calculating the detection priority score is as follows:

[0028]

[0029] In the formula, The detection priority of the i-th detection sub-region is scored. This is the normalized value of the number of faults in the i-th detection sub-region. and These are the weighting coefficients for the relative difference coefficient and the number of failures, respectively. and and All are greater than 0;

[0030] Based on the detection priority score of each detection sub-region, the detection sub-regions are arranged in descending order according to the detection priority score to form a detection sequence.

[0031] Furthermore, the logic behind performing acoustic-vibration coherence analysis on each detection sub-region sequentially based on the detection sequence is as follows: Frequency sweep analysis is performed on each detection sub-region sequentially according to the detection sequence. Specifically, the motor is operated under constant torque, linearly increasing from the lowest stable speed to the highest speed. Time-series data of vibration velocity and sound pressure signals during the operation of each detection sub-region are collected. The collected time-series data is divided into multiple data blocks according to different speed ranges. The acoustic-vibration coherence within each data block is analyzed. Specifically, Fourier transforms are performed on the time-series data of sound pressure and vibration velocity signals, and the corresponding cross-power spectral density and self-power spectral density are calculated to obtain the acoustic-vibration constant coherence coefficient. This constant coherence coefficient characterizes the acoustic-vibration coherence. The formula used to calculate the constant coherence coefficient is as follows:

[0032]

[0033] In the formula, In frequency The acoustic vibration constant coherence coefficient at that location, For sound pressure signals and vibration velocity signals at different frequencies Cross-power spectral density at , For sound pressure signals at frequency The self-power spectral density at that location, For vibration velocity at frequency The self-power spectral density at a given location, where f is the frequency variable in the frequency domain;

[0034] Based on a preset coherence threshold, the noise frequency bands of each detection sub-region are specifically classified into structural noise and airborne noise. The specific logic is as follows: calculate the acoustic constant coherence coefficient at different frequencies within each data block, compare the acoustic constant coherence coefficient with the preset coherence threshold, and determine the noise type based on the comparison result. Specifically:

[0035] If it exists If the detection sub-region is determined to have structural noise, the corresponding noise frequency band is the structural noise.

[0036] If the data blocks corresponding to the detected sub-region all satisfy the following conditions: If so, it is determined that there is air noise in the detection sub-region, and the corresponding noise frequency band is air noise;

[0037] In the formula, The structural coherence threshold, This is the air coherence threshold.

[0038] Furthermore, the logic behind the order tracking analysis of the classified structural noise frequency band is as follows: perform order tracking analysis on the structural noise frequency band, generate an order waterfall plot, extract all continuous order lines in the order waterfall plot, and construct a dataset of order-rotation speed-sound pressure level.

[0039] Based on the order-speed-sound pressure level dataset, the amplitude stability index and amplitude variation gradient of each order line are calculated. The specific formula used to calculate the amplitude stability index is as follows:

[0040]

[0041] In the formula, The magnitude stability index of the k-th order line. This is the sequence of sound pressure level amplitudes at all rotational speeds for the k-th order line. for standard deviation for The arithmetic mean of , where k is the index of the order line;

[0042] The specific formula used to calculate the gradient of amplitude change is as follows:

[0043]

[0044] In the formula, Let the gradient be the magnitude change of the k-th order. The sound pressure level data for the k-th order line at the j-th selected point is... This represents the sound pressure level data of the k-th order line at the (j+1)th selected point. The rotational speed data of the k-th order line at the j-th selected point. Let j be the rotational speed data of the k-th order line at the (j+1)-th selected point, where j is the index of the randomly selected point on the order line. This represents the total number of randomly selected points.

[0045] Furthermore, based on a preset decision threshold, the logic underlying the automatic classification and identification of electromagnetic noise and mechanical noise is as follows:

[0046] like and If so, then the noise corresponding to the k-th order line is determined to be electromagnetic noise;

[0047] like and If so, then the noise corresponding to the k-th order line is determined to be mechanical noise;

[0048] In the formula, As the stability threshold, This is the gradient threshold.

[0049] The present invention also provides a high-precision identification system for asynchronous motor noise sources. This system is used to execute the aforementioned high-precision identification method for asynchronous motor noise sources, and includes:

[0050] The standard signal acquisition module is used to construct the finite element model of the asynchronous motor to be analyzed, divide the surface of the finite element model of the asynchronous motor to be analyzed into several detection sub-regions, perform finite element motor operation simulation, and analyze the vibration velocity signal and sound pressure signal data of each detection sub-region during operation.

[0051] The detection priority analysis module is used to synchronously collect the actual vibration velocity signal and sound pressure signal characteristic parameters of each sub-region, calculate the relative difference with the finite element simulation data, and calculate the detection priority score of each sub-region based on the historical failure number of the associated components of each sub-region through a weighted fusion algorithm.

[0052] The environmental interference elimination module is used to score the detection priority of each detection sub-region, form a detection sequence, and perform acoustic-vibration coherence analysis on each detection sub-region in sequence based on the order of the detection sequence. It calculates the digital coherence function of the acoustic-vibration signal and classifies the noise frequency band of each detection sub-region into structural noise and air noise according to the preset coherence threshold.

[0053] The structural noise classification module is used to perform order tracking analysis on the classified structural noise frequency bands, calculate the amplitude stability index and amplitude change gradient of each order line, and realize the automatic classification and identification of electromagnetic noise and mechanical noise based on the preset decision threshold.

[0054] Compared with the prior art, the beneficial effects of the present invention are:

[0055] By constructing a detailed finite element model and dividing it into several detection sub-regions, the scheme can dynamically capture vibration and sound pressure signals in each region during the simulation of motor operation, improving the timeliness and accuracy of signal acquisition. It also makes the simulation of motor operation status more realistic. By comparing the relative differences between simulated data and actual data, abnormal noise can be effectively identified and its source determined, effectively improving the accuracy of noise source identification.

[0056] Secondly, a weighted fusion algorithm is introduced to calculate the detection priority score of each sub-region, making the detection process more targeted. By considering the historical number of faults in each sub-region, priority can be given to areas with higher fault risk during detection, thereby effectively improving detection efficiency, ensuring timely monitoring of key components, and providing important support for motor fault prevention and maintenance.

[0057] Furthermore, the solution achieves accurate classification of noise frequency bands through acoustic-vibration coherence analysis and digital coherence function calculation, effectively distinguishing structural noise from air noise, reducing the misdiagnosis rate, and optimizing maintenance strategies. Especially in complex noise environments, it can quickly identify and locate problems, significantly improving the efficiency and accuracy of fault handling. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the overall method flow of the present invention;

[0059] Figure 2 A fitted curve showing the relative difference between the actual average sound pressure level and the average sound pressure level.

[0060] Figure 3 This is a graph showing the relative difference between the mean vibration velocity and the relative difference coefficient.

[0061] Figure 4 A bar chart showing the statistical mapping of relative difference coefficients;

[0062] Figure 5 This is a schematic diagram of the overall system structure of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0064] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0065] Example:

[0066] Please see Figures 1-4 The present invention provides a technical solution:

[0067] A high-precision identification method for noise sources in asynchronous motors, comprising the following steps:

[0068] Step 1: Construct a finite element model of the asynchronous motor to be analyzed, divide the surface of the finite element model of the asynchronous motor to be analyzed into several detection sub-regions, perform finite element motor operation simulation, and analyze the vibration velocity signal and sound pressure signal data of each detection sub-region during operation.

[0069] The specific method for constructing the finite element model of the asynchronous motor to be analyzed is as follows: Based on the design parameters of the asynchronous motor, specifically the dimensions and shapes of the stator, rotor, shaft, and end caps, the geometric model is determined. CAD software is used for modeling, and the physical properties of the materials of each component, including the stator, rotor, core, and insulation materials, are determined. These physical properties include density, elastic modulus, Poisson's ratio, and conductivity. In the finite element software, corresponding material properties are assigned to each geometry. The specific steps are as follows: Select geometry: Select the component to be assigned material properties in the software; Define material: Define the corresponding material for the selected geometry, inputting the previously collected physical property parameters; Material library: If the software provides a material library, predefined material properties can be directly selected from the library to ensure they match the actual situation. The material properties are then assigned to the geometric model to form the finite element model.

[0070] The logic behind dividing the surface of the asynchronous motor's finite element model into several detection sub-regions based on the finite element model is as follows: The finite element model surface is divided into several detection sub-regions with equal area. This includes a comprehensive analysis of the asynchronous motor's finite element model to understand its shape, size, and characteristics, including the position and geometric characteristics of key components such as the stator, rotor, end covers, and bearings. The surface area of ​​relevant motor components is calculated. Based on the complexity of the motor and the required signal acquisition accuracy, the number of sub-regions is determined. The number of sub-regions should be moderate; too many will increase processing complexity, while too few will fail to fully reflect the characteristics of each motor component. Each detection sub-region should allow for convenient placement of sensors and acquisition devices to ensure real-time and effective acquisition of vibration velocity and sound pressure signals. Considering the sensor placement and orientation, the system ensures that the characteristic signals of each region are captured to the maximum extent possible. It also ensures that the signals acquired from each sub-region are independent during subsequent processing, facilitating subsequent noise source analysis and comparison.

[0071] The specific method for conducting finite element motor operation simulation and analyzing the vibration velocity and sound pressure signal data of each detection sub-region during operation is as follows: Under constant torque and speed, the vibration velocity and sound pressure signals of each detection sub-region during operation are collected. The constant torque and speed are specifically set according to the working torque and speed of the asynchronous motor to be analyzed. Generally, the commonly used working torque and speed of the asynchronous motor to be analyzed are selected.

[0072] When analyzing vibration velocity and sound pressure signal data during the operation of each detection sub-region, the key is to select appropriate signal acquisition equipment, determine the acquisition scheme, and select data processing methods. The commonly used type is the piezoelectric accelerometer, which is suitable for acquiring low-frequency to high-frequency signals. Electromagnetic or photoelectric velocity sensors can be selected, which are suitable for acquiring vibration velocity signals. A suitable acoustic microphone should be selected, usually a high-sensitivity dynamic or condenser microphone, to capture weak sound wave signals. The microphone should be placed close to the noise source, ensuring that it is pointed at the noise source and avoiding background noise interference. According to the operating frequency and noise characteristics of the motor, an appropriate sampling frequency should be set. Nyquist's law is usually adopted, and the sampling frequency should be more than twice the highest signal frequency.

[0073] Connect the vibration sensor and sound pressure sensor to the data acquisition system, such as a data acquisition card or portable recorder, through a suitable interface. Archive and save the acquired vibration velocity signal and sound pressure signal data, usually in file formats such as CSV or MAT, for subsequent analysis.

[0074] Step 2: Synchronously collect the actual vibration velocity signal and sound pressure signal characteristic parameters of each sub-region, calculate the relative difference with the finite element simulation data, and calculate the detection priority score of each sub-region based on the historical failure number of the associated components of each sub-region through a weighted fusion algorithm.

[0075] The characteristic parameters of the actual vibration velocity signal and sound pressure signal of each sub-region include the average value and corresponding peak value of vibration velocity and sound pressure level within the same rotational speed range;

[0076] The logic behind calculating the relative difference between the data and the finite element simulation data is as follows: A relative difference coefficient is calculated by comparing the average value, corresponding peak value, and integral value within different speed ranges. This coefficient is then used to characterize the relative differences between different detection sub-regions. The specific formula for calculating the relative difference coefficient is as follows:

[0077]

[0078] In the formula, Let be the relative difference coefficient of the i-th detection sub-region. The relative difference in the mean vibration velocity within the i-th detection sub-region. The relative difference in the mean sound pressure level within the i-th detection sub-region. Let be the difference between the actual vibration velocity and the finite element simulation vibration velocity within the i-th detection sub-region at time t. Let be the difference between the actual sound pressure level and the finite element simulation sound pressure level at time t within the i-th detection sub-region; where i is the index of the detection sub-region and t is the time variable of the finite element motor operation simulation process. This represents the starting moment of the finite element motor operation simulation process. This is the termination time of the finite element motor operation simulation process;

[0079] It should be noted that the relative difference coefficient of the i-th detection sub-region The relative coefficient of difference reflects the degree of difference between actual measurement data and finite element simulation data. The larger the value, the greater the difference between the actual vibration and sound pressure level and the simulated data, and the higher the probability of motor failure.

[0080] By calculating the relative differences in the means, the overall trend and shift can be effectively captured, which is crucial for judging the accuracy of the model. The relative difference in the mean vibration velocity within each detection sub-region This value can intuitively reflect the difference between the actual vibration velocity and the simulation results in the overall trend, by introducing instantaneous differences. and This allows for real-time monitoring of the narrow differences between actual and simulated data at different points in time, reflecting potential instantaneous fluctuations, sudden events, or nonlinear effects during motor operation. (Integral term) This represents the cumulative effect of these differences throughout the entire operation. This setting can reflect differences over time, rather than just the instantaneous performance at a single moment. It also reflects the continuous deviations that may exist during operation, which is especially important for evaluating motors that have been running for a long time.

[0081] Where calculation and The specific formula used is as follows:

[0082]

[0083] In the formula, Let be the average actual vibration velocity of the i-th detection sub-region. Let be the average vibration velocity from the finite element simulation of the i-th detection sub-region. Let be the average actual sound pressure level of the i-th detection sub-region. Let be the average sound pressure level of the finite element simulation of the i-th detection sub-region. The maximum vibration velocity in the i-th detection sub-region is... The maximum sound pressure level in the i-th detection sub-region;

[0084] It should be noted that the first The relative difference in the mean vibration velocity within each detection sub-region represents the ratio of the absolute difference between the actual measured mean vibration velocity and the mean vibration velocity simulated by the finite element method to the maximum vibration velocity within that sub-region. The absolute difference is used... It can directly measure the difference between actual and simulated data, unaffected by the sign of the data. This processing ensures the positive direction of the difference, facilitating subsequent comparative analysis. Normalizing the absolute difference and maximum value allows data from different sub-regions to be compared on the same dimension. By using relative values, it avoids the difficulty of comparison caused by different units, and can more intuitively reflect the degree of deviation between the actual state and the theoretical prediction, especially when the signal characteristics of different detection sub-regions may differ significantly. The setup method is the same.

[0085] The maximum vibration velocity of the i-th detection sub-region and the maximum sound pressure level of the i-th detection sub-region Specifically, it refers to the maximum value of the peak values ​​corresponding to vibration velocity and sound pressure level in finite element simulation and actual detection;

[0086] Where calculation and The specific formula used is as follows:

[0087]

[0088] In the formula, and Let be the actual vibration velocity and the finite element simulation vibration velocity of the i-th detection sub-region at time t, respectively. and Let be the actual sound pressure level and the finite element simulated sound pressure level of the i-th detection sub-region at time t, respectively. and t represents the maximum vibration velocity and sound pressure level at time t, respectively.

[0089] It should be noted that the difference calculation uses absolute values. and The purpose is to eliminate the influence of sign, reflecting the magnitude of the difference regardless of whether the deviation is positive or negative. Absolute value calculation is highly applicable to measuring instantaneous differences, especially in dynamic signals where data may fluctuate between positive and negative at certain moments. and As a normalization benchmark, the instantaneous differences can be normalized to a proportional value. The relative differences between instantaneous vibration velocity and sound pressure level can accurately reflect the dynamic deviation between actual data and finite element simulation data during motor operation, providing real-time information for performance evaluation.

[0090] Step 3: Based on the detection priority score of each detection sub-region, a detection sequence is formed. Based on the order of the detection sequence, acoustic-vibration coherence analysis is performed on each detection sub-region in turn. The digital coherence function of the acoustic-vibration signal is calculated. According to the preset coherence threshold, the noise frequency band of each detection sub-region is specifically classified into structural noise and air noise.

[0091] The logic behind calculating the detection priority score for each sub-region using the weighted fusion algorithm is as follows: The number of historical faults collected is normalized; based on the processed data and the relative difference coefficient of the detection sub-regions, the detection priority score for each sub-region is calculated. The specific formula for calculating the detection priority score is as follows:

[0092]

[0093] In the formula, The detection priority of the i-th detection sub-region is scored. This is the normalized value of the number of faults in the i-th detection sub-region. and These are the weighting coefficients for the relative difference coefficient and the number of failures, respectively. and and All are greater than 0;

[0094] It should be noted that the detection priority score for the i-th detection sub-region Used to comprehensively characterize the detection priority of the detection sub-region The larger the value, the higher the priority of the detection sub-region, and the more likely it should be detected.

[0095] Among them, number of failures Reflects the first The frequency of failures in a detection sub-region during its operational history is an important indicator of the region's fault susceptibility. A high number of failures in a sub-region usually indicates that the region has been under high stress or in a critical operating position for a long time; that there are design flaws or insufficient reliability; or that it is susceptible to external environmental influences. In actual engineering, resources for regular inspection and maintenance are often limited, so it is necessary to prioritize monitoring areas with high failure frequencies. The number of failures can serve as a comprehensive reflection of historical data to guide resource allocation and the selection of key detection areas; therefore, the number of failures is directly proportional to the detection priority score.

[0096] The number of failures may vary by a large order of magnitude. For example, a certain area may have a high number of failures, while other areas may have almost no failures. Directly using the number of failures to calculate the score may result in large data bias and make it impossible to compare under a unified scale.

[0097] The relative difference coefficient It provides real-time performance evaluation results for the detected sub-region, reflecting the current state of the region more directly, such as the deviation of vibration velocity and sound pressure level from simulated values. Real-time status data is typically more indicative, indicating the number of faults. It is a statistical result of historical data. Although it cannot directly reflect real-time problems, it can provide an indication of long-term problem trends. Historical fault records are a key indicator of sub-region reliability and cannot be ignored in priority scoring, hence the setting. and and All are greater than 0;

[0098] Based on the detection priority score of each detection sub-region, the detection sub-regions are arranged in descending order according to the detection priority score to form a detection sequence.

[0099] The specific logic for performing acoustic-vibration coherence analysis on each detection sub-region in the order of the detection sequence is as follows: Frequency sweep analysis is performed on each detection sub-region sequentially according to the detection sequence. Specifically, the motor is operated under constant torque, linearly increasing from the lowest stable speed to the highest speed. Time-series data of vibration velocity and sound pressure signals are collected during the operation of each detection sub-region. The collected time-series data is divided into multiple data blocks according to different speed ranges. The acoustic-vibration coherence within each data block is analyzed. Specifically, Fourier transforms are performed on the time-series data of sound pressure and vibration velocity signals, and the corresponding cross-power spectral density and auto-power spectral density are calculated to obtain the acoustic-vibration constant coherence coefficient. This constant coherence coefficient characterizes the acoustic-vibration coherence. The specific formula for calculating the acoustic-vibration constant coherence coefficient is as follows:

[0100]

[0101] In the formula, In frequency The acoustic vibration constant coherence coefficient at that location, For sound pressure signals and vibration velocity signals at different frequencies Cross-power spectral density at , For sound pressure signals at frequency The self-power spectral density at that location, For vibration velocity at frequency The self-power spectral density at a given location, where f is the frequency variable in the frequency domain;

[0102] It should be noted that, in terms of frequency The constant coherence coefficient of acoustic vibration at the location It is a standardized indicator with a value range of 100%. arrive ,when When, it indicates that the sound pressure signal and the vibration velocity signal are at the same frequency. The parts are completely related; when When, it means that the two are completely unrelated;

[0103] Power spectral density (PSD) is a core indicator in signal frequency domain analysis, characterizing the energy distribution of a signal at different frequencies. It is a crucial tool for analyzing signal characteristics. For acoustic-vibration coherence problems, PSD can reveal the energy correlation between two signals in the frequency domain, thus quantifying their coupling relationship. Cross-power spectral density (CPSD) is an indicator of the interaction characteristics between sound pressure and vibration velocity signals, describing the frequency distribution of the two signals. The degree of energy correlation and common influence at a certain frequency range is considered. If two signals have strong common characteristics within a certain frequency range, the cross power spectral density value will be high. The self power spectral density reflects the energy distribution of the sound pressure signal and the vibration velocity signal, respectively. As a normalization benchmark, it ensures that the calculation results have consistency and physical meaning.

[0104] Acoustic-vibration coherence analysis aims to study the physical coupling relationship between sound pressure signals and vibration velocity signals. By analyzing the constant coherence coefficient of acoustic-vibration, it is possible to reveal the degree of contribution of vibration signals to sound pressure signals, the transmission path and coupling mechanism between acoustic-vibration signals, and the acoustic-vibration coupling strength in different frequency ranges.

[0105] The power spectral density is obtained by Fourier transform, which can convert the time domain signal into the frequency domain signal, making it convenient to analyze the characteristics of the signal at different frequencies. Acoustic-vibration coherence analysis mainly focuses on the correlation between sound pressure signal and vibration velocity signal within a specific frequency range. Therefore, analysis based on power spectral density is the most suitable.

[0106] Based on a preset coherence threshold, the noise frequency bands of each detection sub-region are specifically classified into structural noise and airborne noise. The specific logic is as follows: calculate the acoustic constant coherence coefficient at different frequencies within each data block, compare the acoustic constant coherence coefficient with the preset coherence threshold, and determine the noise type based on the comparison result. Specifically:

[0107] If it exists If the detection sub-region is determined to have structural noise, the corresponding noise frequency band is the structural noise.

[0108] If the data blocks corresponding to the detected sub-region all satisfy the following conditions: If so, it is determined that there is air noise in the detection sub-region, and the corresponding noise frequency band is air noise;

[0109] In the formula, The structural coherence threshold, This is the air coherence threshold, which can be set according to the specific motor model and expert experience.

[0110] Step 4: Perform order tracking analysis on the classified structural noise frequency bands, calculate the amplitude stability index and amplitude change gradient of each order line, and realize automatic classification and identification of electromagnetic noise and mechanical noise based on the preset decision threshold.

[0111] The logic behind performing order tracking analysis on the classified structural noise frequency band is as follows: perform order tracking analysis on the structural noise frequency band, generate an order waterfall plot, extract all continuous order lines from the order waterfall plot, and construct a dataset of order-rotation speed-sound pressure level.

[0112] Based on the order-speed-sound pressure level dataset, the amplitude stability index and amplitude variation gradient of each order line are calculated. The specific formula used to calculate the amplitude stability index is as follows:

[0113]

[0114] In the formula, The magnitude stability index of the k-th order line. This is the sequence of sound pressure level amplitudes at all rotational speeds for the k-th order line. for standard deviation for The arithmetic mean of , where k is the index of the order line;

[0115] It should be noted that the first The amplitude stability index of each order line reflects the stability of the sound pressure level of that order. The index is close to 1, indicating that the amplitude is more stable; it is close to 0, indicating that the amplitude fluctuates more.

[0116] Standard deviation is a commonly used statistic for assessing the dispersion of data. It can effectively reflect the variation of sound pressure level amplitude at different rotational speeds. When the standard deviation of sound pressure level is large, it indicates that there are large fluctuations in the rotational speed, which directly affects the stability of the amplitude. The introduction of the mean value allows the size of the standard deviation to be evaluated relative to the overall level of sound pressure level amplitude, thus reflecting stability more accurately.

[0117] The specific formula used to calculate the gradient of amplitude change is as follows:

[0118]

[0119] In the formula, Let the gradient be the magnitude change of the k-th order. The sound pressure level data for the k-th order line at the j-th selected point is... This represents the sound pressure level data of the k-th order line at the (j+1)th selected point. The rotational speed data of the k-th order line at the j-th selected point. Let j be the rotational speed data of the k-th order line at the (j+1)-th selected point, where j is the index of the randomly selected point on the order line. This represents the total number of randomly selected points.

[0120] It should be noted that the first Gradient of amplitude change at each order This reflects the sensitivity of the sound pressure level of this order to changes in rotational speed. The larger the value, the more obvious the response of the sound pressure level to changes in rotational speed, and the amplitude shows a clear increasing trend with rotational speed.

[0121] Difference form This formula is used to calculate the change in sound pressure level between two rotational speed points, directly reflecting the change in sound pressure level with rotational speed. This calculation method can capture the instantaneous rate of change of sound pressure level during the rotational speed change process, making it easy to analyze its sensitivity. By averaging the change gradients of multiple selected points, the random error of a single data point can be reduced, improving the reliability and stability of the results. The amplitude change gradient is used to quantify the degree of response of sound pressure level to changes in rotational speed, and can evaluate the change characteristics of acoustic properties of a specific order under different rotational speed conditions. Essentially, this formula calculates the slope of sound pressure level with respect to rotational speed.

[0122] The logic underlying the automatic classification and identification of electromagnetic noise and mechanical noise based on a preset decision threshold is as follows:

[0123] like and If so, then the noise corresponding to the k-th order line is determined to be electromagnetic noise;

[0124] like and If so, then the noise corresponding to the k-th order line is determined to be mechanical noise;

[0125] In the formula, As the stability threshold, This is the gradient threshold.

[0126] It should be noted that electromagnetic noise is generated by electromagnetic force excitation, and its amplitude usually remains stable during frequency sweep. This is because the excitation source of electromagnetic force, such as current and magnetic field, is constant and periodic and does not change significantly with rotational speed. Therefore, the fluctuation of electromagnetic noise sound pressure level is small, and its amplitude stability index value is close to 1, indicating that the sound pressure level is stable.

[0127] The sound pressure level of electromagnetic noise typically does not change significantly with rotational speed, and its amplitude gradient is close to zero, reflecting a weak trend in sound pressure level change with rotational speed. Mechanically forced noise is caused by the movement or vibration of mechanical components. Its excitation source usually changes significantly with speed during operation. Therefore, the sound pressure level of mechanically forced noise fluctuates greatly, and its amplitude stability index is low, indicating that the sound pressure level is relatively unstable. Mechanically forced noise usually shows a monotonic trend with rotational speed, and the sound pressure level amplitude increases significantly with increasing rotational speed. This is because mechanical impacts, such as imbalances and resonances, are amplified with changes in rotational speed. This characteristic makes the amplitude gradient of mechanical noise significantly positive, reflecting the sensitivity of the sound pressure level to changes in rotational speed.

[0128] Please see Figure 5 The present invention also provides a high-precision identification system for asynchronous motor noise sources. This system is used to execute the aforementioned high-precision identification method for asynchronous motor noise sources, and includes:

[0129] The standard signal acquisition module is used to construct the finite element model of the asynchronous motor to be analyzed, divide the surface of the finite element model of the asynchronous motor to be analyzed into several detection sub-regions, perform finite element motor operation simulation, and analyze the vibration velocity signal and sound pressure signal data of each detection sub-region during operation.

[0130] The detection priority analysis module is used to synchronously collect the actual vibration velocity signal and sound pressure signal characteristic parameters of each sub-region, calculate the relative difference with the finite element simulation data, and calculate the detection priority score of each sub-region based on the historical failure number of the associated components of each sub-region through a weighted fusion algorithm.

[0131] The environmental interference elimination module is used to score the detection priority of each detection sub-region, form a detection sequence, and perform acoustic-vibration coherence analysis on each detection sub-region in sequence based on the order of the detection sequence. It calculates the digital coherence function of the acoustic-vibration signal and classifies the noise frequency band of each detection sub-region into structural noise and air noise according to the preset coherence threshold.

[0132] The structural noise classification module is used to perform order tracking analysis on the classified structural noise frequency bands, calculate the amplitude stability index and amplitude change gradient of each order line, and realize the automatic classification and identification of electromagnetic noise and mechanical noise based on the preset decision threshold.

[0133] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0134] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0135] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0136] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A high-precision method for identifying noise sources in asynchronous motors, characterized in that, The specific steps include: A finite element model of the asynchronous motor to be analyzed is constructed. The surface of the finite element model of the asynchronous motor to be analyzed is divided into several detection sub-regions. Finite element motor operation simulation is performed, and vibration velocity signal and sound pressure signal data of each detection sub-region during operation are analyzed. The actual vibration velocity signal and sound pressure signal characteristic parameters of each sub-region are collected synchronously, the relative difference with the finite element simulation data is calculated, and the detection priority score of each sub-region is calculated by weighted fusion algorithm based on the historical failure number of the associated components of each sub-region. Based on the detection priority score of each detection sub-region, a detection sequence is formed. Based on the order of the detection sequence, acoustic-vibration coherence analysis is performed on each detection sub-region in turn, the digital coherence function of the acoustic-vibration signal is calculated, and the noise frequency band of each detection sub-region is specifically classified into structural noise and air noise according to the preset coherence threshold. The classification of structural noise frequency bands is analyzed by order tracking, and the amplitude stability index and amplitude change gradient of each order line are calculated. Based on the preset decision threshold, the automatic classification and identification of electromagnetic noise and mechanical noise are realized. The specific method for constructing the finite element model of the asynchronous motor to be analyzed is as follows: Based on the design parameters of the asynchronous motor, specifically including the dimensions and shapes of the stator, rotor, shaft, and end cover, the geometric model is determined. CAD software is used for modeling, and the physical properties of the materials of each component, including the stator, rotor, core, and insulation materials, are determined. These physical properties include density, elastic modulus, Poisson's ratio, and conductivity. In the finite element software, the material properties are assigned to the geometric model to form the finite element model. The characteristic parameters of the actual vibration velocity signal and sound pressure signal of each sub-region include the average value and corresponding peak value of vibration velocity and sound pressure level within the same rotational speed range; The logic behind calculating the relative difference between the data and the finite element simulation data is as follows: A relative difference coefficient is calculated by comparing the average value, corresponding peak value, and integral value within different speed ranges. This coefficient is then used to characterize the relative differences between different detection sub-regions. The specific formula for calculating the relative difference coefficient is as follows: ； In the formula, Let be the relative difference coefficient of the i-th detection sub-region. The relative difference in the mean vibration velocity within the i-th detection sub-region. The relative difference in the mean sound pressure level within the i-th detection sub-region. Let be the difference between the actual vibration velocity and the finite element simulation vibration velocity within the i-th detection sub-region at time t. Let be the difference between the actual sound pressure level and the finite element simulated sound pressure level at time t within the i-th detection sub-region; where i is the index of the detection sub-region and t is the time variable of the finite element motor operation simulation process. This represents the starting moment of the finite element motor operation simulation process. This is the termination time of the finite element motor operation simulation process; The logic behind performing order tracking analysis on the classified structural noise frequency band is as follows: perform order tracking analysis on the structural noise frequency band, generate an order waterfall plot, extract all continuous order lines from the order waterfall plot, and construct a dataset of order-rotation speed-sound pressure level. Based on the order-speed-sound pressure level dataset, the amplitude stability index and amplitude variation gradient of each order line are calculated. The specific formula used to calculate the amplitude stability index is as follows: ； In the formula, The magnitude stability index of the k-th order line. This is the sequence of sound pressure level amplitudes at all rotational speeds for the k-th order line. for standard deviation for The arithmetic mean of , where k is the index of the order line; The specific formula used to calculate the gradient of amplitude change is as follows: ； In the formula, Let the gradient be the magnitude change of the k-th order. The sound pressure level data for the k-th order line at the j-th selected point is... This represents the sound pressure level data of the k-th order line at the (j+1)th selected point. The rotational speed data of the k-th order line at the j-th selected point. Let j be the rotational speed data of the k-th order line at the (j+1)-th selected point, where j is the index of the randomly selected point on the order line. This represents the total number of randomly selected points. The logic underlying the automatic classification and identification of electromagnetic noise and mechanical noise based on a preset decision threshold is as follows: like and If so, then the noise corresponding to the k-th order line is determined to be electromagnetic noise; like and If so, then the noise corresponding to the k-th order line is determined to be mechanical noise; In the formula, As the stability threshold, This is the gradient threshold.

2. The high-precision identification method for noise sources of asynchronous motors according to claim 1, characterized in that: The logic behind dividing the surface of the asynchronous motor finite element model to be analyzed into several detection sub-regions based on the finite element model is as follows: the surface of the finite element model is divided into several detection sub-regions with equal area. The specific method used to perform finite element motor operation simulation and analyze the vibration velocity and sound pressure signals during the operation of each detection sub-region is as follows: the motor is operated at constant torque and speed, and the vibration velocity and sound pressure signals during the operation of each detection sub-region are collected. The constant torque and speed are specifically set based on the working torque and speed of the asynchronous motor to be analyzed.

3. The high-precision identification method for asynchronous motor noise sources according to claim 2, characterized in that: Where calculation and The specific formula used is as follows: ； In the formula, Let be the average actual vibration velocity of the i-th detection sub-region. Let be the average vibration velocity from the finite element simulation of the i-th detection sub-region. Let be the average actual sound pressure level of the i-th detection sub-region. Let be the average sound pressure level of the finite element simulation of the i-th detection sub-region. The maximum vibration velocity in the i-th detection sub-region is... The maximum sound pressure level in the i-th detection sub-region; The maximum vibration velocity of the i-th detection sub-region and the maximum sound pressure level of the i-th detection sub-region Specifically, it refers to the maximum value of the peak values ​​corresponding to vibration velocity and sound pressure level in finite element simulation and actual detection; Where calculation and The specific formula used is as follows: ； In the formula, and Let be the actual vibration velocity and the finite element simulation vibration velocity of the i-th detection sub-region at time t, respectively. and Let be the actual sound pressure level and the finite element simulated sound pressure level of the i-th detection sub-region at time t, respectively. and t represents the maximum vibration velocity and sound pressure level at time t, respectively.

4. The high-precision identification method for noise sources of asynchronous motors according to claim 3, characterized in that: The logic behind calculating the detection priority score for each sub-region using the weighted fusion algorithm is as follows: The number of historical faults collected is normalized; based on the processed data and the relative difference coefficient of the detection sub-regions, the detection priority score for each sub-region is calculated. The specific formula for calculating the detection priority score is as follows: ； In the formula, The detection priority of the i-th detection sub-region is scored. This is the normalized value of the number of faults in the i-th detection sub-region. and These are the weighting coefficients for the relative difference coefficient and the number of failures, respectively. and and All are greater than 0; Based on the detection priority score of each detection sub-region, the detection sub-regions are arranged in descending order according to the detection priority score to form a detection sequence.

5. The high-precision identification method for noise sources of asynchronous motors according to claim 4, characterized in that: The specific logic for performing acoustic-vibration coherence analysis on each detection sub-region in the order of the detection sequence is as follows: Frequency sweep analysis is performed on each detection sub-region sequentially according to the detection sequence. Specifically, the motor is operated under constant torque, linearly increasing from the lowest stable speed to the highest speed. Time-series data of vibration velocity and sound pressure signals are collected during the operation of each detection sub-region. The collected time-series data is divided into multiple data blocks according to different speed ranges. The acoustic-vibration coherence within each data block is analyzed. Specifically, Fourier transforms are performed on the time-series data of sound pressure and vibration velocity signals, and the corresponding cross-power spectral density and auto-power spectral density are calculated to obtain the acoustic-vibration constant coherence coefficient. This constant coherence coefficient characterizes the acoustic-vibration coherence. The specific formula for calculating the acoustic-vibration constant coherence coefficient is as follows: ； In the formula, In frequency The acoustic vibration constant coherence coefficient at that location, For sound pressure signals and vibration velocity signals at different frequencies Cross-power spectral density at , For sound pressure signals at frequency The self-power spectral density at that location, For vibration velocity at frequency The self-power spectral density at a given location, where f is the frequency variable in the frequency domain; Based on a preset coherence threshold, the noise frequency bands of each detection sub-region are specifically classified into structural noise and airborne noise. The specific logic is as follows: calculate the acoustic constant coherence coefficient at different frequencies within each data block, compare the acoustic constant coherence coefficient with the preset coherence threshold, and determine the noise type based on the comparison result. Specifically: If it exists If the detection sub-region is determined to have structural noise, the corresponding noise frequency band is the structural noise. If the data blocks corresponding to the detected sub-region all satisfy the following conditions: If so, it is determined that there is air noise in the detection sub-region, and the corresponding noise frequency band is air noise; In the formula, The structural coherence threshold, This is the air coherence threshold.

6. A high-precision identification system for noise sources of asynchronous motors, characterized in that: The high-precision identification system for asynchronous motor noise sources is used to execute the high-precision identification method for asynchronous motor noise sources according to any one of claims 1-5, including: The standard signal acquisition module is used to construct the finite element model of the asynchronous motor to be analyzed, divide the surface of the finite element model of the asynchronous motor to be analyzed into several detection sub-regions, perform finite element motor operation simulation, and analyze the vibration velocity signal and sound pressure signal data of each detection sub-region during operation. The detection priority analysis module is used to synchronously collect the actual vibration velocity signal and sound pressure signal characteristic parameters of each sub-region, calculate the relative difference with the finite element simulation data, and calculate the detection priority score of each sub-region based on the historical failure number of the associated components of each sub-region through a weighted fusion algorithm. The environmental interference elimination module is used to score the detection priority of each detection sub-region, form a detection sequence, and perform acoustic-vibration coherence analysis on each detection sub-region in sequence based on the order of the detection sequence. It calculates the digital coherence function of the acoustic-vibration signal and classifies the noise frequency band of each detection sub-region into structural noise and air noise according to the preset coherence threshold. The structural noise classification module is used to perform order tracking analysis on the classified structural noise frequency bands, calculate the amplitude stability index and amplitude change gradient of each order line, and realize the automatic classification and identification of electromagnetic noise and mechanical noise based on the preset decision threshold.

Images