A motor NVH data intelligent testing method and system

Abstract

The present application discloses an intelligent testing method and system for motor NVH data. The steps of the present application are: when the front end cover of the motor is removed, the rotating shaft is fixed by a test fixture, and the first inclination angle of the rotating shaft relative to the central axis of the motor is measured to measure the high-order harmonic data during the rotation process; the first inclination angle of the rotating shaft and the corresponding high-order harmonic data are used to train the machine learning model; the rotating shaft is limited to have half a degree of freedom in the vertical direction by the test fixture, and the speed data and the amplitude data in the vertical direction during the rotation process are measured to obtain a data table of the first inclination angle, speed and amplitude; the motor drive module is controlled according to the measured amplitude data, and the effectiveness of vibration suppression is tested. The present application uses the test fixture to put the motor shaft in different working states, simulate the wear of the motor after long-term use, and use the machine learning model to calculate the inclination angle of the motor shaft, and collect multi-dimensional data for control and test effectiveness verification.

Description

Technical Field

[0001] The present application relates to the technical field of motor testing, and in particular to a method and system for intelligent testing of motor NVH data. Background Art

[0002] Small-power high-speed motors are increasingly used in automobiles, home appliances, medical devices, and robotics, and the requirements for noise, vibration, and harshness (NVH) are becoming increasingly higher. Under the same or different speed conditions, the coaxiality of the motor's stator and rotor is a key factor. On the other hand, in order to reduce motor vibration, it is also necessary to suppress high-order harmonics during motor operation.

[0003] Manufacturers usually conduct various tests on motors, including NVH tests. However, in conventional tests, only new motors are tested, and problems that only appear after long-term use of the motor cannot be detected in a short period of time. In fact, even accelerated fatigue tests are difficult to detect hidden problems, and these problems only appear after actual users use the motor for a long time. Summary of the invention

[0004] The present application mainly provides a method and system for intelligent testing of motor NVH data to solve the problem of difficulty in obtaining actual motor NVH data after long-term use during motor testing.

[0005] In order to solve the above technical problems, a technical solution adopted in this application is to provide a motor NVH data intelligent testing method, comprising the following steps:

[0006] When the front end cover of the motor is removed, the rotating shaft is fixed by a test fixture, and the first inclination angle of the rotating shaft relative to the central axis of the motor is measured to measure the high-order harmonic data during the rotation process;

[0007] Use the first inclination angle of the rotating shaft and the corresponding high-order harmonic data to train the machine learning model;

[0008] The test fixture is used to limit the rotating shaft to have half a degree of freedom in the vertical direction, and the rotation speed data and the amplitude data in the vertical direction during the rotation process are measured to obtain a data table of the first inclination angle, rotation speed and amplitude;

[0009] The motor drive module is controlled based on the measured amplitude data and the effectiveness of vibration suppression is tested.

[0010] In some embodiments, when the motor is removed from the front end cover, the step of measuring the high-order harmonic data during the rotation process by fixing the rotating shaft with the test fixture at a first inclination angle relative to the central axis of the motor is specifically as follows:

[0011] When the front end cover of the motor is removed, a first inclination angle of the rotating shaft relative to the central axis of the motor is fixed by a test fixture;

[0012] Within the first inclination setting range, high-order harmonic data in the rotation process corresponding to a plurality of first inclination angle values ​​are measured.

[0013] In some embodiments, the step of using the first inclination angle of the rotating shaft and the corresponding high-order harmonic data to train the machine learning model is specifically:

[0014] Get the first inclination angle of the rotating shaft and the corresponding high-order harmonic data as the sample set S and determine the process error σ P ;

[0015] According to the process error σ P Calculate the insensitive parameter ε and penalty coefficient C of the support vector regression machine with the experimental sample set S;

[0016] Determine the range of the kernel function parameter σ, and traverse the range to find the parameter σ value that minimizes the generalization error upper bound P(σ);

[0017] The support vector regression machine fitting equation is calculated according to the sample set S, parameter σ, insensitive parameter ε and penalty coefficient C, the fitting error and generalization error are calculated, and the set of support vectors in the sample set is obtained.

[0018] In some embodiments, the kernel function is a Gaussian radial basis kernel.

[0019] In some embodiments, the step of limiting the rotating shaft to have half a degree of freedom in the vertical direction by a test fixture, measuring the rotation speed data and the amplitude data in the vertical direction during the rotation process, and obtaining a data table of the first inclination angle, the rotation speed and the amplitude is specifically as follows:

[0020] The test fixture is used to limit the rotating shaft to have half a degree of freedom in the vertical direction, and when the motor is stationary, the angle of the rotating shaft relative to the central axis of the motor is a first inclination angle;

[0021] Within the first inclination setting range, measuring the vertical amplitude data corresponding to the plurality of rotation speed data corresponding to the plurality of first inclination angle values;

[0022] A data table reflecting the relationship between the first inclination angle, the rotation speed and the amplitude data is generated according to the measured data.

[0023] In some implementations, the step of controlling the motor drive module according to the measured amplitude data and testing the effectiveness of vibration suppression is specifically as follows:

[0024] The measured amplitude data is converted into the vibration amplitude of the first inclination angle and input into the trained machine learning model to obtain the harmonic forward compensation amount;

[0025] According to the harmonic forward compensation amount, the harmonic reverse compensation amount of the motor drive module is reduced to obtain a corresponding command voltage;

[0026] Control the PWM inverter according to the corresponding command voltage and obtain real-time amplitude data;

[0027] Calculate the difference between the measured amplitude data and the real-time amplitude data, and judge the effectiveness of vibration suppression based on the difference.

[0028] In some embodiments, the first tilt angle setting range is [0,5°].

[0029] In order to solve the above technical problems, another technical solution adopted by the present application is to provide a motor NVH data intelligent testing system, including:

[0030] A test fixture, used for fixing the rotating shaft to form a first inclination angle relative to the central axis of the motor or limiting the rotating shaft to have half a degree of freedom in the vertical direction when the front end cover of the motor is removed;

[0031] The control module is used to measure the high-order harmonic data during the rotation process, and use the first inclination angle of the shaft and the corresponding high-order harmonic data to train the machine learning model; it is also used to measure the speed data and the vertical amplitude data during the rotation process, obtain the data table of the first inclination angle, speed and amplitude, control the motor drive module according to the measured amplitude data, and test the effectiveness of vibration suppression.

[0032] To solve the above technical problems, another technical solution adopted in the present application is: to provide a computer device, which includes a processor and a memory, wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the operations performed by the above-mentioned motor NVH data intelligent testing method.

[0033] In order to solve the above technical problems, another technical solution adopted in the present application is: providing a readable storage medium, in which at least one instruction is stored, and the instruction is loaded and executed by the processor to implement the operations performed by the above-mentioned motor NVH data intelligent testing method.

[0034] The beneficial effects of the present application are as follows: Different from the prior art, the present application discloses an intelligent test method and system for motor NVH data. The present application uses a test fixture to put the motor shaft in different working states, simulate the wear of the motor after long-term use, and calculate the inclination angle of the motor shaft based on a machine learning model, while collecting multi-dimensional data for control and test effectiveness verification. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0036] Figure 1 It is a schematic diagram of the single-side force of the motor shaft;

[0037] Figure 2 This is a schematic diagram of the shaft position after the front cover of the motor has been worn for a long time;

[0038] Figure 3 This is the front view of the structure of the motor front cover after long-term wear;

[0039] Figure 4 It is a flowchart of the steps of a specific embodiment method provided by the present application;

[0040] Figure 5 yes Figure 4 Specific step flow chart of step S1;

[0041] Figure 6 yes Figure 4 Specific step flow chart of step S2;

[0042] Figure 7 yes Figure 4 Specific step flow chart of step S3;

[0043] Figure 8 yes Figure 4 Specific step flow chart of step S4;

[0044] Fig. 9 is a schematic diagram of motor drive control in the first specific embodiment of the present application;

[0045] Fig.10 This is a schematic diagram of motor drive control of the second specific embodiment of the present application;

[0046] Fig.11 This is a schematic diagram of the third specific embodiment of the present application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0048] The terms "first", "second", "third" in the embodiments of the present application are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first", "second", "third" can expressly or implicitly include at least one of the features. In the description of the present application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0049] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0050] Manufacturers usually conduct various tests on motors, including NVH tests. However, conventional tests only test new motors and cannot detect problems that only appear after long-term use of the motor in a short period of time. Even accelerated fatigue tests are difficult to detect hidden problems, which only appear after actual users use the motor for a long time.

[0051] For example, the motor shaft is usually provided with or connected to a transmission device such as a gear or a belt, such as Figure 1 As shown, the motor shaft 2 is subjected to unilateral force, and the motor front cover 1 is worn after long-term use. In particular, the front cover of some small motors only has a shaft hole and no bearing, so long-term wear will affect the position of the shaft, such as Figure 2 As shown, the motor shaft axis CC' deviates from the motor axis BB' to form an inclination angle θ, causing the motor to be subjected to uneven force and causing vibration noise; Figure 3As shown, on the front end cover 1 of the motor, the position of the center of circle 3 is the initial position of the motor shaft 2, and under the condition of long-term unilateral force and wear, the motor shaft hole changes from the initial round hole to the oblong hole 4 as shown in the figure. Under normal circumstances, the motor shaft deviates from the axis by a small angle, usually within 0.5°, due to reasons such as the tilted installation of the motor, gravity, and force on the shaft end. The motor shaft with the front end cover worn above may even deviate from the axis by more than 5°, exacerbating the vibration caused by the air gap harmonics of the motor. In addition, long-term physical wear or electrical corrosion of the bearings may also cause the motor shaft to deviate from the axis by a large angle, and short-term accelerated fatigue testing cannot simulate the working conditions of the motor after long-term wear.

[0052] In view of the existing problems in the prior art, refer to Figure 4 The first specific embodiment of the present application provides a method for intelligent testing of motor NVH data, comprising the following steps:

[0053] S1. With the front end cover of the motor removed, the rotating shaft is fixed by a test fixture, and the first inclination angle of the rotating shaft relative to the central axis of the motor is measured to measure the high-order harmonic data during the rotation process;

[0054] When testing a brand new motor, usually only the installation deviation of the rotating shaft can be tested, and the wear of the motor after long-term use cannot be simulated. Therefore, in this embodiment, the front end cover of the motor is removed, and the motor body is fixed on the test fixture, that is, the motor stator is fixed on one hand, and the rotating shaft is fixed by the test fixture, and the first inclination angle θ of the rotating shaft relative to the central axis of the motor, that is, the deflection angle of the rotating shaft is fixed on the other hand, to simulate the wear of the front end cover of the motor after long-term use. In actual testing, the first inclination angle θ can be measured and recorded by an angle sensor set on the test fixture.

[0055] S2, using the first inclination angle of the rotating shaft and the corresponding high-order harmonic data to train the machine learning model;

[0056] The high-order harmonic data during the rotation process is measured for a fixed first inclination angle θ. Usually, the 5th and 7th harmonic data are selected for measurement. Since the shaft has a first inclination angle θ, the air gap between the rotor and the stator is not uniform, so the inclination state can be reflected by the high-order harmonic data.

[0057] S3, limiting the rotating shaft to have half a degree of freedom in the vertical direction through a test fixture, measuring the rotation speed data and the amplitude data in the vertical direction during the rotation process, and obtaining a data table of the first inclination angle, rotation speed and amplitude;

[0058] When the first inclination angle θ of the rotating shaft causes the air gap between the rotor and the stator to be uneven, the change of high-order harmonics will also cause the vibration of the motor. Since the motor shaft hole changes from a circle to an oblong shape due to long-term wear, the motor shaft may vibrate (or jump) in the oblong shaft hole. On the one hand, this vibration aggravates the impact and wear between the rotating shaft and the edge of the shaft hole. On the other hand, the high-frequency change of the air gap between the rotor and the stator increases the degree of change of high-order harmonics. In order to simulate the working condition of the motor shaft vibrating in the oblong shaft hole, the test fixture is used to limit the rotating shaft to have half a degree of freedom in the vertical direction. For example, the motor shaft limit is set in the oblong hole in the vertical direction, and the first inclination angle θ of the motor shaft in the static state is recorded. Different rotation speeds ω in the rotating state may bring different amplitude data V in the vertical direction. Since the rotating shaft is limited in the oblong hole, there is a maximum inclination angle when the rotating shaft vibrates, that is, the first inclination angle θ of the motor rotating shaft in the static state. The rotating shaft may vibrate. When the rotating shaft vibrates, the inclination angle of the rotating shaft is θ-Δθ. When the maximum amplitude data is Vmax, the corresponding vibration change angle is Δθmax. Therefore, the actual inclination angle of the rotating shaft when the rotating shaft vibrates varies between θ-Δθmax and θ. During the test, it should be noted that the length of the oblong hole is greater than the maximum value of Vmax or an open groove is used to limit the position, so that the rotating shaft has a support position when vibrating downward in the vertical direction, and the vibration amplitude in the vertical direction will not be restricted at the same time, that is, the aforementioned test fixture limits the rotating shaft to half a degree of freedom in the vertical direction. In addition, it is worth noting that in order to make the test results reflect the actual working condition data, the plane where the oblong hole is set should be in the same plane as the position of the front end cover of the motor before it is removed. How the specific test fixture should be set belongs to the conventional technology in the field of motor testing and does not need to be elaborated here.

[0059] S4, controlling the motor drive module according to the measured amplitude data, and testing the effectiveness of vibration suppression;

[0060] Reference Figure 5 As a further preferred embodiment, when the front end cover of the motor is removed, the step of measuring the high-order harmonic data during the rotation process by testing the first inclination angle of the fixed rotating shaft of the tool relative to the central axis of the motor is specifically as follows:

[0061] S11, with the front end cover of the motor removed, fixing a first inclination angle of the rotating shaft relative to the central axis of the motor by means of a test fixture;

[0062] S12. Within a first inclination setting range, measure higher harmonic data of a rotation process corresponding to a plurality of first inclination angle values.

[0063] Since the wear of the front cover during actual use of the motor will cause the motor shaft inclination to change continuously, it is also necessary to test and measure the corresponding high-order harmonic data for different first inclination angle values ​​during the test process. For example, starting from 0°, the corresponding high-order harmonic data is measured every 0.05° angle value.

[0064] Reference Figure 6 As a further preferred implementation, the step of using the first inclination angle of the rotating shaft and the corresponding high-order harmonic data to train the machine learning model is specifically as follows:

[0065] S21, obtaining the first inclination angle of the rotating shaft and the corresponding high-order harmonic data as a sample set S and determining the process error σ P ;

[0066] The first inclination angle and the corresponding high-order harmonic data are taken as the sample set S, where

[0067] S={(x 1 ,y 1 ),(x 2 ,y 2 ,……,(x n ,y n )}

[0068] Process error σ P That is, the standard deviation of the process noise, which can be obtained using the sample variance As the variance of the process noise Estimates:

[0069]

[0070] in,

[0071] for An unbiased estimator of ;

[0072] m is the estimation accuracy adjustment parameter, and m=5 is usually selected.

[0073] S22, according to the process error σ P Calculate the insensitive parameter ε and penalty coefficient C of the support vector regression machine with the experimental sample set S;

[0074] The insensitive parameter ε can be determined using the empirical formula:

[0075]

[0076] The constant τ is usually chosen to be 3, which performs well for different data sets, noise levels, and objective functions.

[0077] The penalty coefficient C can be determined by the following formula:

[0078] C=max(|μ y +3σ y |,|μ y -3σ y |)

[0079] in, σ y for y i The standard deviation of .

[0080] S23, determining a range interval of the kernel function parameter σ, and traversing the range to find the parameter σ value that minimizes the generalization error upper bound P(σ);

[0081] The range of the kernel function parameter σ is generally determined as:

[0082]

[0083] Among them, w min and w max are the far end points x of a given point x i The lower and upper bounds of the weights are affected, and l is the dimension of the input vector space. For complex processes (such as the influence of different inclination angles on the vibration of air gap harmonics), the corresponding output usually varies greatly within the feasible domain, so the far end point x from the given point x i The impact on the kernel function's regression equation needs to be as small as possible in order to output a more accurate support vector regression machine fitting equation. and All are definite constants;

[0084] Kernel Function ||x i -x|| means x and x in vector space i The Euclidean distance, σ is the parameter σ of the kernel function that needs to be determined;

[0085] in is the sum of the fitting errors.

[0086] S24. Calculate the support vector regression machine fitting equation according to the sample set S, parameter σ, insensitive parameter ε and penalty coefficient C, calculate the fitting error and generalization error, and obtain the set of support vectors in the sample set; the above steps are the basic algorithm for support vector regression machine modeling, and there is no need to repeat them here.

[0087] As a further preferred implementation, the kernel function is a Gaussian radial basis kernel.

[0088] Reference Figure 7As a further preferred embodiment, the step of limiting the rotating shaft to have half a degree of freedom in the vertical direction by means of a test fixture, measuring the rotation speed data and the amplitude data in the vertical direction during the rotation process, and obtaining a data table of the first inclination angle, rotation speed and amplitude is specifically as follows:

[0089] S31, limiting the rotation shaft to have half a degree of freedom in the vertical direction through a test fixture, and the angle of the rotation shaft relative to the central axis of the motor when the motor is stationary is a first inclination angle;

[0090] S32, within the first inclination setting range, measuring the vertical amplitude data corresponding to the plurality of rotation speed data corresponding to the plurality of first inclination angle values:

[0091] It is worth noting that in order for the test results to reflect the actual working conditions, the vertical amplitude data should also be the amplitude of the measured shaft on the plane of the motor front end cover before it is removed. For example, a laser rangefinder or other non-contact method can be used to measure the vertical amplitude on the plane of the motor front end cover before it is removed. How the specific test tooling should be set up is a conventional technology in the field of motor testing and there is no need to elaborate here.

[0092] S33, generating a data table reflecting the relationship between the first inclination angle, the rotation speed and the amplitude data according to the measured data:

[0093] Since the amplitude is affected by the first inclination angle and the rotation speed, it is necessary to measure multiple first inclination angle values ​​and vertical amplitude data corresponding to multiple rotation speed data. This data table can be used to achieve precise feedback control in the subsequent motor control process.

[0094] Normally, in order to reduce the impact of high-order harmonics, harmonic reverse compensation is used in motor drive control to reduce motor vibration, that is, injecting high-order harmonics that are opposite to the high-order harmonics to eliminate the impact. However, this is the case when the motor shaft deflection angle is zero or very small.

[0095] When the front cover of the motor is severely worn, the movable space of the shaft in the shaft hole becomes larger. When the deflection angle of the shaft becomes larger, the rotor is very likely to produce excessive amplitude at high speeds, which not only intensifies the high-frequency vibration impact and wear between the shaft and the edge of the shaft hole, but is also an important source of noise.

[0096] However, the technicians found that if an additional harmonic forward compensation is added at this time, it can also be understood as appropriately reducing the original harmonic reverse compensation, rather than trying to inject a completely opposite amount of high-order harmonics into the motor drive control to eliminate the impact as in the conventional solution. The additional harmonic forward compensation is equivalent to giving the shaft Figure 1A small force is applied in the direction of the arrow to suppress the vibration of the shaft in the opposite direction of the arrow. Therefore, the harmonic positive compensation reduces the vibration noise of the motor and also reduces the high-frequency vibration impact and wear of the shaft and the edge of the shaft hole, thus extending the service life of the motor. Figure 8 As a further preferred embodiment, the step of controlling the motor drive module according to the measured amplitude data and testing the effectiveness of vibration suppression is specifically as follows:

[0097] S41, converting the measured amplitude data into the vibration amplitude of the first inclination angle and inputting it into the trained machine learning model to obtain the harmonic forward compensation amount;

[0098] like Fig. 9 As shown, the measured amplitude data V is converted into the vibration amplitude Δθ of the first inclination angle and input into the trained machine learning model to obtain the harmonic forward compensation value i' dq5h and i' dq7h .

[0099] S42, reducing the harmonic reverse compensation amount of the motor drive module according to the harmonic forward compensation amount to obtain a corresponding command voltage;

[0100] The conventional harmonic suppression scheme is to input the three-phase input current i a 、i b 、i c Extract high-order harmonic current to obtain i dq5h and i dq7h ; and the high-order harmonic current i dq5h and i dq7h As the harmonic reverse compensation amount, it performs harmonic current control and voltage coordinate conversion in turn;

[0101] In this scheme, the three-phase input current i of the input motor is first a 、i b 、i c Extract high-order harmonic current to obtain i dq5h and i dq7h , high-order harmonic current i dq5h and i dq7h Subtract the harmonic forward compensation i' dq5h and i' dq7h Then, harmonic current control and voltage coordinate transformation are performed in sequence to obtain the voltage compensation value u * dh and u * qh , and then according to the voltage command value u output by the current control module * d and u * q Subtract the voltage compensation value u* dh and u * qh Get the final command voltage.

[0102] S43, control the PWM inverter according to the corresponding command voltage and obtain real-time amplitude data; that is, the command voltage is subjected to Park inverse transformation and SVPWM modulation in sequence to obtain a PWM control signal and input it to the inverter, thereby obtaining the input voltage of the motor to control the rotation of the motor, and measuring the inclination change of the motor shaft through the angle sensor on the test fixture and converting it into real-time amplitude data.

[0103] S44. Calculate the difference between the measured amplitude data and the real-time amplitude data, and judge the effectiveness of vibration suppression based on the difference; the measured amplitude data reflects the amplitude data when the harmonic forward compensation is not used for vibration suppression control, and the real-time amplitude data is the amplitude data after the harmonic forward compensation is used for vibration suppression control. In theory, the real-time amplitude data should be smaller than the measured amplitude data.

[0104] The specific effectiveness judgment method can adopt conventional judgment methods, such as using MSE mean square error calculation. When the mean square error is less than the set value, it can be judged that the vibration suppression is effective. Or as a simplified calculation method, when the mean of the difference between the measured amplitude data and the real-time amplitude data is less than the set threshold, it can be judged that the vibration suppression is effective.

[0105] As a further preferred embodiment, the first inclination angle setting range is [0,5°].

[0106] Further as a preferred implementation, since the first inclination angle setting range is [0,5°] and the angle value is very small, the amplitude data V=Ltanθ≈Lθ can be approximately considered within this setting range, where L is the axial length of the motor shaft inside the motor, that is, the distance between the front and rear end covers, which can simplify the calculation amount and save the computing resources of the motor controller.

[0107] Reference Fig.10 The second specific embodiment specifically describes the motor drive control scheme in actual use after obtaining corresponding data after testing this scheme:

[0108] A. Get the three-phase input current i of the input motor a 、i b 、i c Clark transformation and Park transformation are performed in sequence to obtain the feedback current i in the rotor dq coordinate system. d 、i q ;

[0109] B. Obtain the command current i in the rotor dq coordinate system * d 、i* q , according to the feedback current i d 、i q and command current i * d 、i * q Perform current control to obtain the first voltage vector u in the rotor dq coordinate system * d 、u * q ;

[0110] C. The three-phase input current i of the input motor a 、i b 、i c Extract high-order harmonic current to obtain i dq5h 、i qq7h ;

[0111] Normally, in order to reduce the impact of high-order harmonics, harmonic reverse compensation is used in motor drive control to reduce motor vibration, that is, injecting high-order harmonics that are opposite to the high-order harmonics. dq5h 、i qq7h To eliminate the influence, but this is the case when the motor shaft deflection angle is zero or very small.

[0112] When the front cover of the motor is severely worn, the movable space of the shaft in the shaft hole becomes larger. When the deflection angle of the shaft becomes larger, the rotor is very likely to produce excessive amplitude at high speeds, which not only intensifies the high-frequency vibration impact and wear between the shaft and the edge of the shaft hole, but is also an important source of noise.

[0113] However, the technicians found that if an additional harmonic forward compensation is added at this time, it can also be understood as appropriately reducing the original harmonic reverse compensation, rather than trying to inject a completely opposite amount of high-order harmonics into the motor drive control to eliminate the impact as in the conventional solution. The additional harmonic forward compensation is equivalent to giving the shaft Figure 1 A smaller force is applied in the direction of the arrow to suppress the vibration of the shaft in the opposite direction of the arrow. Therefore, the harmonic positive compensation reduces the vibration noise of the motor, and at the same time reduces the high-frequency vibration impact and wear of the shaft and the edge of the shaft hole, thereby extending the service life of the motor.

[0114] In order to calculate the harmonic forward compensation amount, it is first necessary to predict the first inclination angle data θ of the rotating shaft during the operation of the motor.

[0115] D. Input the high-order harmonic currents extracted from the most recent N sampling time points into the trained machine learning model to obtain N first inclination angle data θ of the rotating shaft;

[0116] Since angle sensors are generally not installed on the motor shaft, especially for miniaturized motors, it is necessary to calculate the first inclination data θ through high-order harmonic data to reflect its inclination state. Due to the complex characteristics of motor vibration, in order to avoid single measurement data errors, the average value of multiple measurements is used as the basis for calculation.

[0117] E. When the mean of the first inclination angle data of N shafts When it is greater than the set threshold, it can be determined that the motor shaft is tilted at this time. Therefore, according to the average value of the first inclination angle data of N shafts And the motor speed ω calculates the harmonic forward compensation i' dq5h 、i' qq7h ;

[0118] F. According to the harmonic forward compensation amount i' dq5h 、i' qd7h and the high-order harmonic current i extracted at the current sampling time point dq5h 、i qq7h The harmonic reverse compensation amount is calculated;

[0119] G. According to the harmonic reverse compensation amount, harmonic current control and voltage coordinate conversion are performed in sequence to obtain the harmonic compensation voltage u * dh 、u * qh ;

[0120] H. According to the first voltage vector u * d 、u * q and harmonic compensation voltage u * dh 、u * qh generating a command voltage;

[0121] I. The command voltage is subjected to Park inverse transformation and SVPWM modulation in sequence to obtain a PWM control signal and input it to the inverter to obtain the input voltage of the motor to control the rotation of the motor.

[0122] As a further preferred embodiment, when the average of the first inclination angle data of the N rotating shafts is greater than the set threshold, the step of calculating the harmonic forward compensation amount according to the average of the first inclination angle data of the N rotating shafts and the motor speed is specifically as follows:

[0123] E1. When the average value of the first inclination angle data of N rotating shafts is greater than the set threshold, And the motor speed ω query data table to obtain the amplitude data V;

[0124] E2. Input the amplitude data V into the harmonic forward compensation module to obtain the harmonic forward compensation amount.

[0125] As a further preferred embodiment, when the average of the first inclination angle data of N shafts is greater than the set warning value, an alarm prompt or an error is issued, which indicates that the shaft inclination angle may be too large, affecting normal operation or causing serious failure.

[0126] As a further preferred implementation, the step of inputting the amplitude data into the harmonic forward compensation module to obtain the harmonic forward compensation amount is specifically as follows:

[0127] E21, calculating the vibration amplitude Δθ of the first inclination angle according to the amplitude data V;

[0128] E22, input the vibration amplitude Δθ of the first inclination angle into the trained machine learning model to obtain the harmonic forward compensation value i' dq5h 、i' dq7h .

[0129] In the above-mentioned first specific embodiment, the machine learning model is used in both steps D and E. On the one hand, the harmonic value is input and the angle value is output, and on the other hand, the angle value is input and the harmonic value is output. Therefore, the training of the machine learning model can be to train two machine learning sub-models separately, the first machine learning sub-model is used to input the harmonic value and output the angle value, and the second machine learning sub-model is used to input the angle value and output the harmonic value; or the training of one of the machine learning models that inputs the angle value and outputs the harmonic value, and when the harmonic value is input, the support vector regression machine fitting equation is used to perform a reverse solution to obtain the output angle value.

[0130] In the above-mentioned first specific embodiment and the second specific embodiment, the difference between the motor drive control is that in the motor test phase, the amplitude data V can be accurately obtained through actual measurement, while in the actual operation phase of the motor, the amplitude data V changes rapidly, and usually high-precision sensors are not installed on the motor to increase the cost of miniaturized motors. Therefore, the speed is obtained through the speed acquisition module, and the average value of the first inclination data is used. The lowest cost and most convenient way is to query the data table to obtain the amplitude data V with the motor speed ω, because there are many mature solutions to choose from in the motor control module, and the implementation is not difficult.

[0131] Reference Fig.11 In view of the problems existing in the prior art, the third specific embodiment of the present application provides a motor NVH data intelligent testing system, including:

[0132] A test fixture, used for fixing the rotating shaft to form a first inclination angle relative to the central axis of the motor or limiting the rotating shaft to have half a degree of freedom in the vertical direction when the front end cover of the motor is removed;

[0133] The control module is used to measure the high-order harmonic data during the rotation process, and use the first inclination angle of the shaft and the corresponding high-order harmonic data to train the machine learning model; it is also used to measure the speed data and the vertical amplitude data during the rotation process, obtain the data table of the first inclination angle, speed and amplitude, control the motor drive module according to the measured amplitude data, and test the effectiveness of vibration suppression.

[0134] In response to the problems existing in the prior art, the fourth specific embodiment of the present application provides a computer device, which includes a processor and a memory, wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the operations performed by the above-mentioned motor NVH data intelligent testing method.

[0135] In response to the problems existing in the prior art, the fifth specific embodiment of the present application provides a readable storage medium, in which at least one instruction is stored. The instruction is loaded and executed by a processor to implement the operations performed by the above-mentioned motor NVH data intelligent testing method.

[0136] The above descriptions are merely embodiments of the present application and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A motor NVH data intelligent testing method, characterized in that: The following steps are involved: When the front end cover of the motor is removed, the rotating shaft is fixed by a test fixture, and the first inclination angle of the rotating shaft relative to the central axis of the motor is measured to measure the high-order harmonic data during the rotation process; Use the first inclination angle of the rotating shaft and the corresponding high-order harmonic data to train the machine learning model; The test fixture is used to limit the rotating shaft to have half a degree of freedom in the vertical direction, and the rotation speed data and the amplitude data in the vertical direction during the rotation process are measured to obtain a data table of the first inclination angle, rotation speed and amplitude; The motor drive module is controlled based on the measured amplitude data and the effectiveness of vibration suppression is tested.

2. The motor NVH data intelligent testing method according to claim 1 is characterized in that: The step of measuring the high-order harmonic data during the rotation process by fixing the rotating shaft with respect to the central axis of the motor by testing the first inclination angle of the test fixture with respect to the central axis of the motor when the front end cover of the motor is removed is specifically as follows: When the front end cover of the motor is removed, a first inclination angle of the rotating shaft relative to the central axis of the motor is fixed by a test fixture; Within the first inclination setting range, high-order harmonic data in the rotation process corresponding to a plurality of first inclination angle values ​​are measured.

3. The motor NVH data intelligent testing method according to claim 2 is characterized in that: The step of using the first inclination angle of the rotating shaft and the corresponding high-order harmonic data to train the machine learning model is specifically as follows: Get the first inclination angle of the rotating shaft and the corresponding high-order harmonic data as the sample set S and determine the process error σ P ; According to the process error σ P Calculate the insensitive parameter ε and penalty coefficient C of the support vector regression machine with the experimental sample set S; Determine the range of the kernel function parameter σ, and traverse the range to find the parameter σ value that minimizes the generalization error upper bound P(σ); The support vector regression machine fitting equation is calculated according to the sample set S, parameter σ, insensitive parameter ε and penalty coefficient C, the fitting error and generalization error are calculated, and the set of support vectors in the sample set is obtained.

4. The motor NVH data intelligent testing method according to claim 3 is characterized in that: The kernel function is a Gaussian radial basis kernel.

5. The motor NVH data intelligent testing method according to claim 1, characterized in that: The step of limiting the rotating shaft to have half a degree of freedom in the vertical direction by the test fixture, measuring the rotation speed data and the amplitude data in the vertical direction during the rotation process, and obtaining the data table of the first inclination angle, the rotation speed and the amplitude is specifically as follows: The test fixture is used to limit the rotating shaft to have half a degree of freedom in the vertical direction, and when the motor is stationary, the angle of the rotating shaft relative to the central axis of the motor is a first inclination angle; Within the first inclination setting range, measuring the amplitude data in the vertical direction corresponding to the plurality of rotation speed data corresponding to the plurality of first inclination angle values; A data table reflecting the relationship between the first inclination angle, the rotation speed and the amplitude data is generated according to the measured data.

6. The motor NVH data intelligent testing method according to claim 1, characterized in that: The step of controlling the motor drive module according to the measured amplitude data and testing the effectiveness of vibration suppression is specifically as follows: The measured amplitude data is converted into the vibration amplitude of the first inclination angle and input into the trained machine learning model to obtain the harmonic forward compensation amount; According to the harmonic forward compensation amount, the harmonic reverse compensation amount of the motor drive module is reduced to obtain a corresponding command voltage; Control the PWM inverter according to the corresponding command voltage and obtain real-time amplitude data; Calculate the difference between the measured amplitude data and the real-time amplitude data, and judge the effectiveness of vibration suppression based on the difference.

7. The motor NVH data intelligent testing method according to any one of claims 2 to 5, characterized in that: The first inclination angle setting range is [0,5°].

8. An intelligent test system for motor NVH data, characterized in that: Included are: A test fixture, used for fixing the rotating shaft to form a first inclination angle relative to the central axis of the motor or limiting the rotating shaft to have half a degree of freedom in the vertical direction when the front end cover of the motor is removed; The control module is used to measure the high-order harmonic data during the rotation process, and use the first inclination angle of the shaft and the corresponding high-order harmonic data to train the machine learning model; it is also used to measure the speed data and the vertical amplitude data during the rotation process, obtain the data table of the first inclination angle, speed and amplitude, control the motor drive module according to the measured amplitude data, and test the effectiveness of vibration suppression.

9. A computer device, characterized in that: The computer device includes a processor and a memory, wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the operation performed by the motor NVH data intelligent testing method according to any one of claims 1-7.

10. A readable storage medium, characterized in that: The readable storage medium stores at least one instruction, which is loaded and executed by the processor to implement the operations performed by the motor NVH data intelligent testing method as described in any one of claims 1-7.

Images