Motor health assessment method based on multi-dimensional parameters

Abstract

The present invention provides a motor health assessment method based on multidimensional parameters, comprising the following steps: step S1, collecting multidimensional parameters of the motor during operation through sensors; step S2, cleaning, denoising and normalizing the collected data to obtain fusion state parameters; step S3, marking fault samples according to fault type, degree and location information, and dividing healthy samples and fault samples; step S4, constructing a health assessment function based on healthy samples; step S5, solving the assessment function to obtain a health assessment value; step S6, assessing the motor health status according to the assessment value. The method optimizes data input by fusing multidimensional parameter information and weighted calculation, and quantifies the health status in combination with the Euclidean distance function, so that the assessment result is more accurate. The method can effectively identify the health level and fault risk of the motor, and provide an intelligent and scientific health assessment method for the maintenance and operation of the motor system.

Description

Technical Field

[0001] The present invention relates to the technical field of motor system fault monitoring, and in particular to a motor health assessment method based on multi-dimensional parameters. Background Art

[0002] Motors are key equipment widely used in industrial production, and their health status is crucial to the stability of the production line and product quality. However, since motors operate under high loads and complex environments for a long time, they are prone to a variety of problems such as wear, overheating, and insulation aging, which can lead to failures or even downtime. In order to ensure the continuity of production and reduce maintenance costs and downtime, motor health assessment has become an important means of industrial maintenance.

[0003] Traditional motor health assessment methods are mainly based on the monitoring of single parameters, such as current, vibration, temperature and noise, but a single parameter is often difficult to fully reflect the actual health status of the motor. Motor failure types are diverse and the causes of failures are complex. It is difficult to accurately judge and predict failure risks through a single parameter. For example, temperature abnormalities may be caused by a variety of factors, including overload, insufficient heat dissipation, or electrical failures inside the motor. It is difficult to effectively distinguish the fault type by monitoring the temperature alone.

[0004] With the advancement of sensor technology and data processing capabilities, multi-dimensional parameter health assessment methods have gradually attracted attention. By collecting and analyzing multiple parameters such as temperature, vibration, speed, current, etc., and combining advanced data processing methods such as data fusion, feature extraction and machine learning, it is possible to more accurately identify the characteristics of potential faults and improve the accuracy of fault prediction. This method can not only reflect the overall health status of the motor, but also locate the specific location of the fault, which helps to improve the maintenance efficiency and equipment utilization of the motor.

[0005] Therefore, the research on motor health assessment technology based on multi-dimensional parameters provides new ideas for predictive maintenance of industrial equipment, which can effectively reduce the occurrence rate of failures and improve production safety and efficiency. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention uses current sample feature classification analysis to perform fault characterization and diagnosis of the motor system. Detection and diagnosis can also be performed when the motor is in working state, which is very convenient and low-cost.

[0007] To achieve the above object, the present invention provides the following technical solution: a motor health assessment method based on multi-dimensional parameters, comprising the following steps:

[0008] Step S1, collecting multi-dimensional parameters of the motor during operation through sensors, wherein the multi-dimensional parameters at least include: current, voltage, equivalent impedance, and temperature;

[0009] Step S2, preprocessing the collected data to obtain fusion state parameters, wherein the preprocessing at least includes data cleaning, denoising and normalization;

[0010] Step S3, marking the fault samples according to the fault type, degree and location information, and determining the fault samples and healthy samples for the health of the motor;

[0011] Step S4: establishing a health evaluation function based on the health sample;

[0012] Step S5, solving the health evaluation function to obtain a health evaluation value;

[0013] Step S6: Evaluate the health of the motor based on the health evaluation value.

[0014] Furthermore, in step S1, the collection of multi-dimensional parameters is specifically as follows:

[0015] S11, connecting a plurality of sensors to a driver of the motor;

[0016] S12, using a sampling circuit and a sampling chip to obtain original parameters of current, voltage and temperature signals through a power supply line;

[0017] S13, performing weighted calculation on the original parameters to obtain weighted parameters; the weighted formula satisfies:

[0018] ;

[0019] Among them, the subscript is the parameter number, , Indicates the total number of multidimensional parameters of the sampling points; subscript is the sampling point number, , Indicates the total number of sampling points; Indicates The parameters in The original multidimensional parameter matrix obtained by sampling at sampling points; represents the weighted multidimensional parameter matrix obtained at all sampling points, to is the weighting parameter; Shidi The parameters in The weighting coefficient of the sampling performed at each sampling point.

[0020] Further, in step S13,

[0021] ;

[0022] in, For the The parameters in The first sampling point The original weighting coefficients of the samples satisfy:

[0023] ;

[0024] in, Indicates The parameters in The error limit of sampling at sampling points is is a small positive number that avoids zero denominator.

[0025] Furthermore, the data preprocessing in step S2 is multi-dimensional parameter fusion, fusion state parameters Satisfies the following formula:

[0026] ;

[0027] in, represents the fusion state parameter; is the fusion coefficient matrix, to are the fusion coefficients, is the transpose of the normalized weighted multidimensional parameter matrix; to They represent the normalized weighting parameters respectively.

[0028] Furthermore, the normalized value is calculated according to the following formula:

[0029] ;

[0030] in, and Respectively represent The maximum and minimum values ​​of the parameter at all sampling points.

[0031] Furthermore, in step S4, establishing a health evaluation function based on the health sample specifically includes:

[0032] S41. Retrieve the fusion state parameters of the healthy sample , set its fusion coefficient matrix to 1 matrix, satisfying:

[0033]

[0034] S42. Construct a health evaluation function to satisfy:

[0035] ;

[0036] Among them, the superscript is the serial number of the healthy sample, , is the total number of healthy samples, Indicates the serial number is Fusion state parameters of healthy samples; Represents the fusion state parameter of the sample to be evaluated; is the health evaluation function.

[0037] Furthermore, in step S5, establishing a health evaluation function based on the health sample specifically includes:

[0038] ;

[0039] in, represents the minimization solution, which means solving the problem that minimizes the health evaluation function. The value of represents the minimum fusion coefficient matrix, to are the minimum fusion coefficients respectively.

[0040] Furthermore, in step S5, obtaining the health assessment value specifically refers to:

[0041] Based on the minimum fusion coefficient matrix Establish a Euclidean distance function with a distance basis of 1 matrix, satisfying:

[0042] ;

[0043] in, Represents the minimum fusion coefficient matrix Euclidean distance from 1 matrix.

[0044] Furthermore, in step S5, the physical meaning of the health assessment value is the Euclidean distance between the sample to be evaluated and the healthy sample cluster. When the Euclidean distance is 0, the sample to be evaluated is considered to be equal to the healthy sample. The larger the Euclidean distance, the farther the sample to be evaluated is from the healthy sample cluster, and the lower the health of the evaluated sample.

[0045] Further, in step S6, evaluating the health of the motor based on the health evaluation value means: When the motor reaches the theoretical optimal health, The larger the value, the lower the healthiness. When the value is greater than the preset threshold, it is determined that the motor is faulty.

[0046] Beneficial Effects

[0047] Compared with the prior art, the present invention provides the following beneficial effects:

[0048] 1. Multi-dimensional parameter collection ensures comprehensive data: By collecting multi-dimensional parameters such as current, voltage, equivalent impedance, temperature, etc., the operating status of the motor is fully captured, avoiding insufficient information caused by a single parameter, and can more accurately reflect the true health status of the motor.

[0049] 2. Data preprocessing improves data quality: Through data cleaning, denoising and normalization operations, the noise interference in the original signal is reduced, the numerical scale of each parameter is unified, and the quality of input data is ensured, thereby improving the accuracy and stability of health assessment.

[0050] 3. Weighted calculation to optimize parameter impact: In the process of multi-dimensional parameter fusion, weighted calculation is used to perform differentiated processing on different sampling points, and weights are assigned according to the importance of each parameter, so that key parameters have a greater impact on the evaluation results, thereby optimizing the accuracy of health assessment.

[0051] 4. Adaptive adjustment of health assessment function: By constructing a health assessment function, the health status of the sample to be evaluated can be quantified. The weight parameters in the assessment function can be adaptively adjusted to make the assessment result more in line with the actual operating status of the motor.

[0052] 5. Euclidean distance to evaluate health status deviation: The health evaluation value is calculated based on the Euclidean distance between the minimum fusion coefficient matrix and the healthy sample, which can intuitively reflect the difference between the sample to be evaluated and the healthy sample. When the evaluation value is zero, it means that the sample is close to the optimal health state, and the larger the evaluation value, the worse the health of the motor, which is convenient for quickly identifying fault risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0054] Figure 2 A schematic diagram of a fault diagnosis model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to better understand the purpose, structure and function of the present invention, the motor system fault qualitative diagnosis method based on current sample feature classification analysis of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0056] As a non-invasive diagnostic method, current signature analysis can effectively detect some common faults of the motor, such as bearing damage, stator winding inter-turn short circuit, rotor bar breakage, etc., by analyzing the current signal during the operation of the motor.

[0057] Example 1

[0058] refer to Figure 1This embodiment provides a motor health assessment method based on multi-dimensional parameters, comprising the following steps:

[0059] Step S1, collecting multi-dimensional parameters of the motor during operation through sensors, wherein the multi-dimensional parameters at least include: current, voltage, equivalent impedance, and temperature;

[0060] Step S2, preprocessing the collected data to obtain fusion state parameters, wherein the preprocessing at least includes data cleaning, denoising and normalization;

[0061] Step S3, marking the fault samples according to the fault type, degree and location information, and determining the fault samples and healthy samples for the health of the motor;

[0062] Step S4: establishing a health evaluation function based on the health sample;

[0063] Step S5, solving the health evaluation function to obtain a health evaluation value;

[0064] Step S6: Evaluate the health of the motor based on the health evaluation value.

[0065] In step S1, the collection of multi-dimensional parameters is specifically as follows:

[0066] S11, connecting a plurality of sensors to a driver of the motor;

[0067] S12, using a sampling circuit and a sampling chip to obtain original parameters of current, voltage and temperature signals through a power supply line;

[0068] S13, performing weighted calculation on the original parameters to obtain weighted parameters; the weighted formula satisfies:

[0069] ;

[0070] Among them, the subscript is the parameter number, , Indicates the total number of multidimensional parameters of the sampling points; subscript is the sampling point number, , Indicates the total number of sampling points; Indicates The parameters in The original multidimensional parameter matrix obtained by sampling at sampling points; represents the weighted multidimensional parameter matrix obtained at all sampling points, to is the weighting parameter; Shidi The parameters in The weighting coefficient of the sampling performed at each sampling point.

[0071] In step S13,

[0072] ;

[0073] in, For the The parameters in The first sampling point The original weighting coefficients of the samples satisfy:

[0074] ;

[0075] in, Indicates The parameters in The error limit of sampling at sampling points is is a small positive number that avoids zero denominator.

[0076] The data preprocessing in step S2 is multi-dimensional parameter fusion, fusion state parameters Satisfies the following formula:

[0077] ;

[0078] in, represents the fusion state parameter; is the fusion coefficient matrix, to are the fusion coefficients, is the transpose of the normalized weighted multidimensional parameter matrix; to They represent the normalized weighting parameters respectively.

[0079] The normalized value is calculated according to the following formula:

[0080] ;

[0081] in, and Respectively represent The maximum and minimum values ​​of the parameter at all sampling points.

[0082] In step S4, establishing a health evaluation function based on the health sample specifically includes:

[0083] S41. Retrieve the fusion state parameters of the healthy sample , set its fusion coefficient matrix to 1 matrix, satisfying:

[0084]

[0085] S42. Construct a health evaluation function to satisfy:

[0086] ;

[0087] Among them, the superscript is the serial number of the healthy sample, , is the total number of healthy samples, Indicates the serial number is Fusion state parameters of healthy samples; Represents the fusion state parameter of the sample to be evaluated; is the health evaluation function.

[0088] In step S5, establishing a health evaluation function based on the health sample specifically includes:

[0089] ;

[0090] in, represents the minimization solution, which means solving the problem that minimizes the health evaluation function. The value of represents the minimum fusion coefficient matrix, to are the minimum fusion coefficients respectively.

[0091] In step S5, obtaining the health assessment value specifically refers to:

[0092] Based on the minimum fusion coefficient matrix Establish a Euclidean distance function with a distance basis of 1 matrix, satisfying:

[0093] ;

[0094] in, Represents the minimum fusion coefficient matrix Euclidean distance from 1 matrix.

[0095] In step S5, the physical meaning of the health evaluation value is the Euclidean distance between the sample to be evaluated and the healthy sample cluster. When the Euclidean distance is 0, the sample to be evaluated is considered to be equal to the healthy sample. The larger the Euclidean distance, the farther the sample to be evaluated is from the healthy sample cluster, and the lower the health of the evaluation sample.

[0096] In step S6, evaluating the health of the motor based on the health evaluation value refers to: When the motor reaches the theoretical optimal health, The larger the value, the lower the healthiness. When the value is greater than the preset threshold, it is determined that the motor is faulty.

[0097] Example 2

[0098] refer to Figure 2 This embodiment provides a motor health assessment method based on multi-dimensional parameters, comprising the following steps:

[0099] Collect current, voltage, temperature, impedance, and capacitive reactance sample data during motor operation through sensors;

[0100] Perform pre-processing operations such as data cleaning, denoising and normalization on the collected current, voltage, temperature, impedance and capacitive reactance data to obtain a sample data set;

[0101] Establish a deep learning model based on convolutional neural network, and import the processed sample data set into the model for training and optimization;

[0102] Considering the fault type, fault severity and fault location information, the fault samples are labeled and the model is further trained;

[0103] A hybrid algorithm is introduced to identify the abnormal state of each characteristic under motor fault and establish an optimization model based on fault type;

[0104] The neural network algorithm is used to extract fault features from the optimized data, and a morphological graph set is established based on the features of the classified faults;

[0105] According to the extracted features, based on the trained deep learning model, the fault type of the motor signal is identified and located, and a fault type pattern library is established;

[0106] Use the classifier to perform qualitative diagnosis of motor faults, and based on the characteristics of the input signal, the model can determine whether the motor has a fault and locate the specific location of the fault.

[0107] Furthermore, the fault samples are marked, and the marking contents include: fault type, fault location and fault cause;

[0108] Specifically, motor failures can be divided into many types, including bearing failure, insulation aging, winding failure, brush wear, overload and overheating, fan failure, etc. These failures not only affect the normal operation of the motor, but may also cause more serious damage;

[0109] ‌ Among them, the hybrid algorithm adopted by the present invention includes:

[0110] Use technology to balance the data set and adjust the number of samples in each category through specific methods to achieve a balanced state to solve the problem of data imbalance;

[0111] Signal processing algorithm, using filtering technology and noise model to suppress noise and solve various noise interference problems in current data;

[0112] Model adaptive technology is introduced to achieve automatic adjustment and optimization of the model by learning adaptive strategies to adapt to changes in motor failure modes.

[0113] Specifically, the method based on current characteristic analysis of the present invention can identify characteristic information in the motor current signal and realize accurate diagnosis of motor faults. For example, mechanical faults of the motor such as eccentricity and bearing damage will cause air gap vibration. The vibration frequency is linearly related to the motor current spectrum, and monitoring can be achieved through current analysis. For electrical faults such as stator winding inter-turn short circuit and rotor bar breakage, different current characteristics will be shown in the early stage, and accurate diagnosis can be achieved by analyzing these characteristics.

[0114] The core advantage of the motor system fault qualitative diagnosis method based on current sample feature classification analysis of the present invention lies in its high efficiency and accuracy. This method can monitor the health status of the motor in real time by analyzing the current signal when the motor is running, and give an early warning before a fault occurs, thereby improving the reliability and safety of the motor operation.

[0115] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A motor health assessment method based on multi-dimensional parameters, characterized in that: The following steps are involved: Step S1, collecting multi-dimensional parameters of the motor during operation through sensors, wherein the multi-dimensional parameters at least include: current, voltage, equivalent impedance, and temperature; Step S2, preprocessing the collected data to obtain fusion state parameters: the preprocessing at least includes data cleaning, denoising and normalization; The fusion state parameters obtained by the preprocessing are multi-dimensional parameter fusion, and the fusion state parameters Satisfies the following formula: ； in, represents the fusion state parameter; is the fusion coefficient matrix, to are the fusion coefficients, is the transpose of the normalized weighted multidimensional parameter matrix; to denote the normalized weighting parameters respectively; Step S3, marking the fault samples according to the fault type, degree and location information, and determining the fault samples and healthy samples for the health of the motor; Step S4: Establishing a health evaluation function based on the health sample; specifically including: S41. Retrieve the fusion state parameters of the healthy sample , set its fusion coefficient matrix to 1 matrix, satisfying: ； S42. Construct a health evaluation function to satisfy: ； Among them, the superscript is the serial number of the healthy sample, , is the total number of healthy samples, Indicates the serial number is Fusion state parameters of healthy samples; Represents the fusion state parameter of the sample to be evaluated; is the health assessment function; Step S5, solving the health evaluation function to obtain a health evaluation value; specifically comprising: ； in, represents the minimization solution, which means solving the problem that minimizes the health evaluation function. The value of represents the minimum fusion coefficient matrix, to are the minimum fusion coefficients respectively; Obtaining a health assessment value specifically refers to: Based on the minimum fusion coefficient matrix Establish a Euclidean distance function with a distance basis of 1 matrix, satisfying: ； in, Represents the minimum fusion coefficient matrix Euclidean distance from 1 matrix; Step S6: Evaluate the health of the motor based on the health evaluation value.

2. The motor health assessment method based on multi-dimensional parameters according to claim 1 is characterized in that: In step S1, the collection of multi-dimensional parameters is specifically as follows: S11, connecting a plurality of sensors to a driver of the motor; S12, using a sampling circuit and a sampling chip to obtain original parameters of current, voltage and temperature signals through a power supply line; S13, performing weighted calculation on the original parameters to obtain weighted parameters; the weighted formula satisfies: ； Among them, the subscript is the parameter number, , Indicates the total number of multidimensional parameters of the sampling points; subscript is the sampling point number, , Indicates the total number of sampling points; Indicates The parameters in The original multidimensional parameter matrix obtained by sampling at sampling points; represents the weighted multidimensional parameter matrix obtained at all sampling points, to is the weighting parameter; Shidi The parameters in The weighting coefficient of the sampling performed at each sampling point.

3. The motor health assessment method based on multi-dimensional parameters according to claim 2 is characterized in that: In step S13, ； in, For the The parameters in The first sampling point The original weighting coefficients of the samples satisfy: ； in, Indicates The parameters in The error limit of sampling at sampling points is is a small positive number that avoids zero denominator.

4. The motor health assessment method based on multi-dimensional parameters according to claim 3 is characterized in that: The normalized value is calculated according to the following formula: ； in, and Respectively represent The maximum and minimum values ​​of the parameter at all sampling points.

5. The motor health assessment method based on multi-dimensional parameters according to claim 4 is characterized in that: In step S6, the physical meaning of the health evaluation value is the Euclidean distance between the sample to be evaluated and the healthy sample cluster. When the Euclidean distance is 0, the sample to be evaluated is considered to be equal to the healthy sample. The larger the Euclidean distance, the farther the sample to be evaluated is from the healthy sample cluster, and the lower the health of the evaluation sample.

6. The motor health assessment method based on multi-dimensional parameters according to claim 5 is characterized in that: In step S7, evaluating the health of the motor based on the health evaluation value means: When the motor reaches the theoretical optimal health, The larger the value, the lower the healthiness. When the value is greater than the preset threshold, it is determined that the motor is faulty.

Images