CN108445393B - A kind of permanent magnet synchronous motor fault detection method and system - Google Patents

Abstract

The invention discloses a permanent magnet synchronous motor fault detection method and system. The disclosed method includes the following steps: Step S100: collecting N series of time-series original image data of the normal and faulty thermal change process of the permanent magnet synchronous motor, j=1, j≤N; Step S200: Segment the jth raw image data into pixel areas with higher surface temperature of the permanent magnet synchronous motor; Step S300: Perform feature extraction on the jth image processed image data; Step S400: Determine the state category of the motor according to the feature extracted for the jth time; Step S500: Determine whether j>N is true, if not, j=j+1, return to step S200, otherwise enter step S600; step S600: pass N times of thermal change process Extract features and their corresponding motor state categories, and establish a support vector machine model for motor fault diagnosis; Step S700: use the trained model to perform fault discrimination on the field collected features. Without direct contact with the motor, it can automatically and accurately judge the fault of the motor under continuous duty.

Description

A fault detection method and system for a permanent magnet synchronous motor

technical field

The invention relates to the technical field of permanent magnet synchronous motors, in particular to a fault detection method and system for permanent magnet synchronous motors.

Background technique

In recent years, with the rapid development of modern science and technology, the performance and process of electromagnetic materials, especially rare earth electromagnetic materials, have been gradually improved and improved. Coupled with the rapid development of power electronics, power transmission technology, and automatic control technology, the permanent magnet synchronous motor Performance just keeps getting better and better. Furthermore, permanent magnet synchronous motors have the advantages of light weight, simple structure, small size, good characteristics, and high power density. Many scientific research institutions and enterprises are actively developing permanent magnet synchronous motors, and their application fields will continue to expand. However, usually the motor is prone to failure due to poor working conditions, severe vibration, and high temperature in the working environment. Therefore, the fault diagnosis of permanent magnet synchronous motors is also an important research field.

In the process of motor fault diagnosis, the characteristics of common motor faults are most obviously manifested in the frequency of the vibration signal and the current of the stator. Although the fault detection through the frequency and current has achieved a certain reliability, these two methods require high cost Advanced testing equipment and a lot of time, while testing will interrupt production. And because most of the judgments are made through the accumulation of manual experience at present, most of the subjective factors are likely to cause misjudgment, and there is a lack of a fully automatic scientific diagnosis and judgment method.

Therefore, how to fully automatically and accurately judge the failure of the permanent magnet synchronous motor under continuous duty without directly contacting the motor is an urgent problem to be solved by those skilled in the art.

Contents of the invention

The purpose of the present invention is to provide a permanent magnet synchronous motor fault detection method and system, which can fully automatically and accurately judge the permanent magnet synchronous motor fault under continuous duty without directly contacting the motor.

In order to solve the above technical problems, the present invention provides a permanent magnet synchronous motor fault detection method, the method includes the following steps:

Step S100: collecting N series of time-series original image data of the normal and faulty thermal change process of the permanent magnet synchronous motor, j=1, j≤N;

Step S200: Segment a series of time-series original image data of the j-th thermal change process into pixel areas with higher surface temperature of the permanent magnet synchronous motor by image processing technology;

Step S300: Perform feature extraction on a series of time-series image data of the j-th thermal change process after image processing, including temperature change curve features, inter-frame variance vibration features, and histogram-related features of pixel areas with higher surface temperatures of the motor ;

Step S400: Determine the motor state category according to the features extracted from the jth thermal change process;

Step S500: Determine whether j>N is true, if not, j=j+1, return to step S200, if true, enter step S600;

Step S600: establishing a permanent magnet synchronous motor fault diagnosis support vector machine model through the features extracted from the thermal change process of the motor for N times and the corresponding motor state category;

Step S700: Use the trained model to perform fault discrimination on the field permanent magnet synchronous motor.

Preferably, in step S100, an infrared camera is used to collect original image data, and the original image data is grayscale image data.

Preferably, the series of time-series original image data of the thermal change process described in step S100 specifically refers to when the temperature of the permanent magnet synchronous motor rises to establish a new thermal balance after a rated step load is applied to the permanent magnet synchronous motor generally starting from a cold state A series of time-series original images, the infrared thermal imager is set to collect a series of time-series original image data of the thermal change process of the permanent magnet synchronous motor at preset time intervals.

Preferably, in the step 200, it is specifically:

Step S210: converting a series of time-series original grayscale images collected by the infrared thermal imager to the jth thermal change process into real temperature images;

Step S220: Perform Gaussian kernel filtering on a series of time-series temperature images of the jth thermal change process;

Step S230: Randomly extract a frame of image in the thermal steady-state time period from a series of time-series temperature images of the jth thermal change process after the filtering process, calculate the threshold for segmenting the extracted frame image, and extract the low-level image in the extracted frame image The pixel at the threshold is assigned a value of 0, and the other pixel values remain unchanged;

Step S240: Calculate the average value μ ₁ and standard deviation σ ₁ of the pixels of the extracted frame image described in step S230, assign the pixel points lower than μ ₁ -σ ₁ in the extracted frame image to 0, and keep other pixels unchanged, so that Segment the pixel area with higher surface temperature of the permanent magnet synchronous motor.

Preferably, the step S300 is specifically:

Step S310: Determine the time-varying curve of the maximum temperature value of the pixel according to a series of time-series temperature images of the jth thermal change process described in step S210;

Step S320: Determine the minimum value of the edge of the area according to the pixel area with a higher surface temperature of the permanent magnet synchronous motor described in step S240;

Step S330: Take the maximum temperature value of the thermal steady state constant pixel and the minimum value of the region edge as the upper and lower intervals of the feature histogram, divide the interval into 10 equal parts on average, and count the permanent magnet synchronous motors with higher surface temperatures described in step S240 Pixel area temperature histogram;

Step S340: Calculate the mean value, standard deviation, skewness, kurtosis and entropy of the temperature histogram of the pixel region with higher surface temperature of the permanent magnet synchronous motor;

Step S350: extracting the time-varying curve characteristics of the pixel maximum temperature value described in step S310;

Step S360: Calculating the thermal steady-state inter-frame variance vibration characteristics of a series of time-series temperature images of the j-th thermal change process that has been filtered in step S220.

Preferably, the step S400 is specifically:

Step S410: forming a feature vector from the extracted features of the jth thermal change process;

Step S420: Determine the motor state category according to the feature vector;

Preferably, the step S600 is specifically:

Step S610: establishing a training set of a classifier through the feature vectors extracted from the thermal change process of the motor for N times and the corresponding motor state categories;

Step S620: The classifier adopts a support vector machine, the kernel function adopts a radial basis kernel function, and the classified training set takes the extracted feature vector as input and the motor state category as output, trains the support vector machine, and obtains the fault diagnosis of the permanent magnet synchronous motor support vector machine model.

The present invention also provides a permanent magnet synchronous motor fault detection system, including an image acquisition module, an image preprocessing module, a feature extraction module, a model building module and a fault detection module, wherein:

The image acquisition module is used to collect N series of time-series original image data of the normal and faulty thermal change process of the permanent magnet synchronous motor, j=1, j≤N, and send it to the image preprocessing module;

The image preprocessing module is used to use image processing technology to segment a series of time series original image data of the jth thermal change process sent by the image acquisition module into the pixel area with a higher surface temperature of the permanent magnet synchronous motor, and send it to the feature extraction module ;

The feature extraction module is used to extract features from a series of time-series image data of the jth thermal change process sent by the image preprocessing module, including temperature change curve features, inter-frame variance vibration features and motor surface temperature Higher pixel area histogram related features are sent to the model building module;

The model building module is used to determine the motor state category according to the extracted features sent by the feature extraction module; judge whether j>N is established, if not established, j=j+1, return to the image preprocessing module, and if established, go through N times of motor Feature extraction of thermal change process and determination of motor state category, establishment of permanent magnet synchronous motor fault diagnosis support vector machine model, and sending to fault detection module;

The fault detection module is used to judge the fault of the permanent magnet synchronous motor by using the model trained by the model building module.

Preferably, an infrared thermal imager is used in the image acquisition module to acquire raw image data.

Preferably, when the infrared thermal imager collects the original image data, the infrared thermal imager is placed at the center of the side of the permanent magnet synchronous motor, the infrared thermal imager and the permanent magnet synchronous motor are located at the same level, and the distance between the two The selection of the horizontal distance should make the image collected by the infrared thermal imager include the entire permanent magnet synchronous motor, and the relative position of the two is fixed each time it is collected.

Without direct contact with the motor, it can automatically and accurately judge permanent magnet synchronous motor faults under continuous duty, including stator turn-to-turn faults, bearing faults, heat dissipation faults, and demagnetization faults.

Description of drawings

Fig. 1 is the flow chart of a kind of permanent magnet synchronous motor fault detection method that the first embodiment provides;

Fig. 2 is the flow chart of a kind of permanent magnet synchronous motor fault detection method that the second embodiment provides;

Fig. 3 is a curve diagram of the maximum temperature change of a normal motor and a faulty motor;

Fig. 4 is the structural block diagram of a kind of permanent magnet synchronous motor fault detection system provided by the present invention;

Fig. 5 is a schematic front view of the placement position of the permanent magnet synchronous motor and the infrared thermal imaging camera provided by the present invention;

Fig. 6 is a schematic top view of the permanent magnet synchronous motor and the thermal imaging camera provided by the present invention.

Detailed ways

In order to enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings.

Referring to FIG. 1 , FIG. 1 is a flowchart of a fault detection method for a permanent magnet synchronous motor provided in the first embodiment.

A method for fault detection of a permanent magnet synchronous motor, the method comprising the following steps:

Step S100: Collect N series of time-series original image data of the normal and faulty thermal change process of the permanent magnet synchronous motor, j=1, j≤N, N is determined according to actual needs and experience;

Step S200: Segment a series of time-series original image data of the j-th thermal change process into pixel areas with higher surface temperature of the permanent magnet synchronous motor by image processing technology;

Step S300: Perform feature extraction on a series of time-series image data of the j-th thermal change process after image processing, including temperature change curve features, inter-frame variance vibration features, and histogram-related features of pixel areas with higher surface temperatures of the motor ;

Step S400: Determine the motor state category according to the features extracted from the jth thermal change process;

Step S500: Determine whether j>N is established, if not, j=j+1, return to step S200, if established, enter step S600;

Step S600: establishing a permanent magnet synchronous motor fault diagnosis support vector machine model through the features extracted from the thermal change process of the motor for N times and the corresponding motor state category;

Step S700: Use the trained model to perform fault discrimination on the field permanent magnet synchronous motor.

A series of time-series original image data of the thermal change process of the normal permanent magnet synchronous motor and several faulty permanent magnet synchronous motors were collected respectively. Preferably, an infrared thermal imager is used to separately collect a series of time-series original gray images of the permanent magnet synchronous motor thermal change process including the normal operation of the permanent magnet synchronous motor and various faults such as stator turn-to-turn faults, bearing faults, eccentric faults, and demagnetization faults. image data. A series of time-series original image data of the thermal change process specifically refers to a series of time-series original images when the temperature of the permanent magnet synchronous motor rises to a new thermal balance after the rated step load is applied to the permanent magnet synchronous motor from the cold state. The thermal imaging camera is set to collect a series of time series original image data of the thermal change process of the permanent magnet synchronous motor at preset time intervals. The preset time interval is set according to experience.

A series of time-series original image data of a certain thermal change process is segmented by image processing technology to obtain pixel areas with higher surface temperature of the permanent magnet synchronous motor. Feature extraction is performed on a series of time-series image data that has undergone image processing. Extraction includes temperature variation curve features, inter-frame variance vibration features and histogram-related features of the pixel area with high surface temperature of the motor. Determine the motor state category according to the features extracted from this thermal change process. According to the method from step S200 to step S400, a series of time-series original image data collected from the normal and different fault types of the permanent magnet synchronous motor thermal change process are subjected to image processing and feature extraction, and the motor status category is determined according to the extracted features. A support vector machine model for fault diagnosis of permanent magnet synchronous motors is established through the features extracted from all thermal change processes and their corresponding motor state categories. Use the trained model to identify the faults of the permanent magnet synchronous motors collected in the field, judge the operating status of the motors, and if there is a fault, judge which permanent magnet synchronous motor fault belongs to, and give an alarm.

In a further solution, a series of time-series original image data of the thermal change process of the permanent magnet synchronous motor with the same type of normal and different fault types can be collected, and then through image processing and feature extraction, the motor status category can be determined according to the extracted features Finally, the support vector machine model for fault diagnosis of this type of permanent magnet synchronous motor is established. For different types of permanent magnet synchronous motors, the corresponding permanent magnet synchronous motor fault diagnosis support vector machine models are established. When making fault judgments for different types of permanent magnet synchronous motors, you can input the motor model in advance, and then use the fault diagnosis support vector machine model of the permanent magnet synchronous motor corresponding to the model to perform fault judgment on the permanent magnet synchronous motor.

The method can fully and accurately judge permanent magnet synchronous motor faults including stator turn-to-turn faults, bearing faults, heat dissipation faults, and demagnetization faults under continuous duty without direct contact with the motor.

Referring to FIG. 2 to FIG. 3 , FIG. 2 is a flowchart of a fault detection method for a permanent magnet synchronous motor provided in the second embodiment, and FIG. 3 is a maximum temperature change curve of a normal motor and a faulty motor.

A method for fault detection of a permanent magnet synchronous motor, the method comprising the following steps:

Step S100: The infrared thermal imager collects N series of time-series original grayscale image data of the normal and faulty thermal change processes of the permanent magnet synchronous motor, j=1, j≤N.

Step S210: Convert a series of time-series original grayscale images of the jth thermal change process collected by the infrared thermal imager into real temperature images.

Since the original image collected by the infrared thermal imager is a grayscale image, the value of each pixel of the grayscale image does not represent the real temperature. It needs to be converted according to the following formula (1) to obtain the real temperature image:

Among them, T(x, y) is the temperature value at the image (x, y), W _tot (x, y) is the total radiation at the image (x, y) collected by the infrared camera, and τ _atm is the room air σ is the Stefan-Boltzmann constant, T _refl is the reflection temperature, T _atm is the room temperature, ε _obj is the emissivity of the object.

Step S220: Perform Gaussian kernel filtering on a series of time-series temperature images of the j-th thermal change process, and use a 3×3 Gaussian kernel to remove noise in a series of time-series temperature images, wherein the filtering formula is as follows:

Where I(x, y) is the pixel value at (x, y) of the temperature image after Gaussian filtering, and Kernel(k, l) is the pixel value at (k, l) of the 3×3 Gaussian kernel function.

Step S230: Randomly extract a frame of the thermal steady-state time period from a series of time-series temperature images of the j-th thermal change process after filtering, and use the maximum inter-class variance method to calculate T _treshold as the threshold for image segmentation. The pixels below the threshold in the extracted frame image are assigned a value of 0, and the values of other pixels remain unchanged.

Step S240: Calculate the average value μ ₁ and standard deviation σ ₁ of the pixels of the extracted frame image described in step S230, assign the pixel points lower than μ ₁ -σ ₁ in the extracted frame image to 0, and keep other pixels unchanged, so that Segment the pixel area with higher surface temperature of the permanent magnet synchronous motor. The mean value μ ₁ and standard deviation σ ₁ are obtained by the following formula:

Among them, I(c) represents the temperature value of the cth pixel, and there are n pixels in total.

Step S310: Determine the time-varying curve T _max (t) of the pixel maximum temperature value T _max according to a series of time-series temperature images of the jth thermal change process described in step S210, as shown in FIG. 3 , are of the same type The temperature curve T _max (t) of the normal motor and the faulty motor T _max over time, and the corresponding temperature rise time integral area. And the area of the integral area of the motor temperature rise time can be input into the support vector machine as one of the characteristics of fault diagnosis.

Step S320: Determine the minimum value T _min of the edge of the area according to the pixel area with a higher surface temperature of the permanent magnet synchronous motor described in step S240;

Step S330: Take the maximum temperature value T _max of the thermally stable constant pixel in step S310 and the minimum value T _min of the region edge determined in step S320 as the upper and lower intervals of the feature histogram, divide the interval into 10 equal parts on average, and count the steps S240 The temperature histogram of the pixel region with higher surface temperature of the permanent magnet synchronous motor is as follows:

Among them, S( _xi ) is the number of pixels in the temperature range of _xi ; H( _xi ) is the frequency of occurrence of temperature in the temperature range of _xi .

Step S340: Calculate the mean value μ ₂ , standard deviation σ ₂ , skewness Sk, kurtosis K, and entropy E of the temperature histogram of the pixel region with a higher surface temperature of the permanent magnet synchronous motor obtained in step 330:

Step S350: Extracting the features of the time-varying curve T _max (t) of the pixel maximum temperature value T _max described in step S310:

Among them, Area is the integral of the rising temperature within the rising time, t _s refers to the rising time section required to reach the new thermal steady state from 10% to 90%, and T _max (t) is the maximum temperature of the motor at time t; T _atm is the room ambient temperature.

Step S360: Calculating the thermal steady-state inter-frame variance vibration feature V of a series of time-series temperature images of the j-th thermal change process after the filtering process described in step S220:

Wherein, V is the maximum value in the inter-frame variance of the thermal steady-state time-series temperature image, I _t (z) is the temperature value of the zth pixel of the thermal steady-state t frame, and a frame of image has n pixel values in total; is the average temperature value of the z-th pixel of the total f frames of images in the thermal steady state, and the series of time-series temperature images are f frames.

Step S410: Form the extracted features of the jth thermal change process into a feature vector, and a series of time-series temperature images of a thermal change process can obtain a corresponding 19-dimensional feature vector X=[Area, V, T _max , T _min , μ ₂ , σ ₂ , Sk, E, K, H(x ₁ ),...H( _xi ),...H(x ₁₀ ), where H( _xi ) is the _xith temperature in the hottest zone The frequency of the interval, T _max is a constant value in thermal steady state.

Step S420: Determine the motor state category Y according to the feature vector X. The determination of the category Y is determined by the state category of the test motor. For example, the fault types of the permanent magnet synchronous motor that can be detected include stator turn-to-turn fault Y ₁ , bearing fault Y ₂ , heat dissipation fault Y ₃ , demagnetization fault Y ₄ , and normal motor faults. Class Y ₀ .

Step S500: Determine whether j>N is true, if not, j=j+1, return to step S200, if true, enter step S610;

Step S610: establishing a training set of a classifier through the feature vectors extracted from the thermal change process of the motor for N times and the corresponding motor state categories;

Step S620: The classifier adopts a support vector machine, the kernel function adopts a radial basis kernel function, and the classified training set takes the extracted feature vector X as input and the motor state category Y as output, trains the support vector machine, and obtains a permanent magnet synchronous motor Fault diagnosis support vector machine model.

Step S700: Use the trained model to perform fault discrimination on the field permanent magnet synchronous motor. The infrared thermal imaging camera collects a series of time-series original grayscale image data of the on-site permanent magnet synchronous motor, and then uses steps S210 to S360 to obtain the characteristics of the on-site permanent magnet synchronous motor. The characteristics of the motor are used to judge the running state of the motor. If there is a fault, it is judged which kind of permanent magnet motor fault it belongs to, and an alarm is issued.

Referring to Figures 4 to 6, Figure 4 is a structural block diagram of a permanent magnet synchronous motor fault detection system provided by the present invention, and Figure 5 is a schematic front view of the permanent magnet synchronous motor and infrared thermal imaging camera placement provided by the present invention , FIG. 6 is a schematic top view of the permanent magnet synchronous motor and the infrared thermal imaging camera provided by the present invention.

The present invention also provides a permanent magnet synchronous motor fault detection system, including an image acquisition module 1, an image preprocessing module 2, a feature extraction module 3, a model building module 4 and a fault detection module 5, wherein:

Image collection module 1, used to collect N series of time-series original image data of normal permanent magnet synchronous motor and faulty permanent magnet synchronous motor thermal change process, j=1, j≤N, N is determined according to actual needs and experience , sent to the image preprocessing module 2;

The image preprocessing module 2 is used to segment a series of time-series original image data of the j-th thermal change process sent by the image acquisition module 1 into the pixel area with a higher surface temperature of the permanent magnet synchronous motor using image processing technology, and send it to the feature extract module 3;

The feature extraction module 3 is used to extract features from a series of time-series image data of the jth thermal change process sent by the image preprocessing module 2 after image processing, including temperature change curve features, inter-frame variance vibration features and motor The relevant features of the histogram of the pixel area with higher surface temperature are sent to the model building module 4;

The model building module 4 is used to determine the motor state category according to the extracted features sent by the feature extraction module 3; judge whether j>N is established, if not established, j=j+1, return to the image preprocessing module, and if established, pass through N times The feature extraction of the thermal change process of the motor and the determination of the motor state category establish a permanent magnet synchronous motor fault diagnosis support vector machine model and send it to the fault detection module 5;

The fault detection module 5 is used to use the model trained by the model building module 4 to discriminate the fault of the on-site permanent magnet synchronous motor.

The image acquisition module 1 collects a series of time-series original image data of the normal and several faulty thermal change processes of the permanent magnet synchronous motor respectively. Preferably, an infrared thermal imager is used to collect a series of time-series original image data of the thermal change process of the permanent magnet synchronous motor including the normal operation of the permanent magnet synchronous motor and various faults such as stator turn-to-turn faults, bearing faults, eccentric faults, and demagnetization faults . A series of time-series original image data of the thermal change process specifically refers to a series of time-series original images when the temperature of the permanent magnet synchronous motor rises to a new thermal balance after the rated step load is applied to the permanent magnet synchronous motor from the cold state. The thermal imaging camera is set to collect a series of time series original image data of the thermal change process of the permanent magnet synchronous motor at preset time intervals.

The image preprocessing module 2 uses image processing technology to segment a series of time-series original image data of a certain thermal change process into pixel areas with higher surface temperature of the permanent magnet synchronous motor. The feature extraction module 3 performs feature extraction on the series of time-series image data that has undergone image processing. Extraction includes temperature variation curve features, inter-frame variance vibration features and histogram-related features of the pixel area with high surface temperature of the motor. The model building module 4 determines the state category of the motor according to the features extracted from this thermal change process. According to the method of image preprocessing module 2 and feature extraction module 3, a series of time-series original image data collected in the thermal change process of permanent magnet synchronous motors with different fault types are image processed and feature extracted, and the motor state is determined according to the extracted features category. A support vector machine model for fault diagnosis of permanent magnet synchronous motors is established through the features extracted from all thermal change processes and their corresponding motor state categories. The fault detection module 5 uses the trained model to perform fault discrimination on the field permanent magnet synchronous motor, and judges the running state of the motor. If there is a fault, it judges which permanent magnet synchronous motor fault belongs to, and gives an alarm.

In a further solution, a series of time-series original image data of the thermal change process of the permanent magnet synchronous motor of the same model and different fault types can be collected, and then through image preprocessing and feature extraction, and the state category of the motor is determined according to the extracted features. Types of Support Vector Machine Models for Fault Diagnosis of Permanent Magnet Synchronous Motors. After establishing the corresponding permanent magnet synchronous motor fault diagnosis support vector machine model for different types of permanent magnet synchronous motors, you can input the motor model in advance when making fault judgments for different types of permanent magnet synchronous motors, and then use the model corresponding The fault diagnosis support vector machine model of the permanent magnet synchronous motor is used for fault discrimination of the permanent magnet synchronous motor.

Without direct contact with the motor, it can automatically and accurately judge permanent magnet synchronous motor faults under continuous duty, including stator turn-to-turn faults, bearing faults, heat dissipation faults, and demagnetization faults.

As shown in Figures 5 and 6, when the infrared thermal imaging camera collects the original image data, an infrared thermal imaging camera is installed in front of the measured permanent magnet synchronous motor target, and the infrared thermal imaging camera has the ability to regularly capture thermal images and transmit them to The function of the computer center, the temperature measurement range is -20°C-300°C, and the temperature resolution is less than 0.1°C. The thermal imaging camera should be placed at the center of the side of the permanent magnet synchronous motor. The thermal imaging camera and the permanent magnet synchronous motor are located at the same level. The horizontal distance between the two should be selected so that the images collected by the thermal imaging camera include For the entire permanent magnet synchronous motor, the relative position of the two is fixed each time it is collected. The data transmission cable is electromagnetically shielded, the length of the transmission cable is optimized to reduce signal attenuation, and the aperture and focal length of the visual sensor of the infrared thermal imager are adjusted to ensure that the collected thermal distribution images are clear and accurate.

A fault detection method and system for a permanent magnet synchronous motor provided by the present invention has been introduced in detail above. In this paper, specific examples are used to illustrate the principles and implementation modes of the present invention, and the descriptions of the above embodiments are only used to help understand the core idea of the present invention. It should be pointed out that for those skilled in the art, without departing from the principles of the present invention, some improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Claims (10)

1. a kind of permanent magnet synchronous motor fault detection method, which is characterized in that the described method comprises the following steps:

Step S100: a series of N number of timing original graphs of normal and tape jam the thermal transformation of permanent magnet synchronous motor are acquired As data, j=1, j≤N；

Step S200: a series of timing raw image datas of the thermal transformation of jth time are divided using image processing techniques The higher pixel region of permanent magnet synchronous motor surface temperature out；

Step S300: a series of time sequence image datas of the thermal transformation by the jth of image procossing time are subjected to feature and are mentioned It takes, including temperature variation curve feature, interframe variance vibration performance and the higher pixel region histogram of motor surface temperature Correlated characteristic；

Step S400: motor status classification is determined according to the feature that the thermal transformation of jth time extracts；

Step S500: judging whether j > N is true, if not, j=j+1, return step S200 enter step S600 if setting up；

Step S600: the feature and its corresponding motor status classification extracted by the thermal transformation of n times motor establish permanent magnetism Synchronous motor fault diagnosis supporting vector machine model；

Step S700: fault distinguishing is carried out with permanent magnet synchronous motor of the trained model to scene.

2. permanent magnet synchronous motor fault detection method according to claim 1, which is characterized in that using red in step S100 Outer thermal imaging system acquires raw image data, and raw image data is greyscale image data.

3. permanent magnet synchronous motor fault detection method according to claim 2, which is characterized in that heat described in step S100 A series of timing raw image datas of change procedure specifically refer to generally since cold conditions, apply to permanent magnet synchronous motor specified Permanent magnet synchronous motor temperature rises to a series of timing original images when establishing new thermal balance, infrared thermal imagery after step load Instrument is set to a series of timing raw image datas of prefixed time interval acquisition permanent magnet synchronous motor thermal transformation.

4. permanent magnet synchronous motor fault detection method according to claim 3, which is characterized in that in the step 200 Specifically:

Step S210: a series of timing original-gray images that thermal infrared imager collects the thermal transformation of jth time are converted For true temperature pattern；

Step S220: a series of timing temperature patterns of the thermal transformation of jth time are subjected to Gaussian kernel filtering processing；

Step S230: it is randomly selected in a series of timing temperature patterns of the thermal transformation of the jth time by filtering processing One frame image of hot steady state time section, calculates the threshold value of the extraction frame image segmentation, and threshold value will be lower than in the extraction frame image Pixel assignment is 0, and other pixel values remain unchanged；

Step S240: extraction frame image slices vegetarian refreshments mean μ described in step S230 is calculated₁And standard deviation sigma₁, by the extraction frame image In be lower than μ₁-σ₁Pixel be assigned a value of 0, other pixels remain unchanged, so that it is higher to be partitioned into permanent magnet synchronous motor surface temperature Pixel region.

5. permanent magnet synchronous motor fault detection method according to claim 4, which is characterized in that have in the step S300 Body are as follows:

Step S310: determine pixel most according to a series of timing temperature patterns of the thermal transformation of jth described in step S210 time The curve of big temperature value changed over time；

Step S320: determine edges of regions most according to the higher pixel region of permanent magnet synchronous motor surface temperature described in step S240 Small value；

Step S330: using the invariable pixel maximum temperature values of hot stable state and edges of regions minimum value as the upper of feature histogram Section is averagely divided into 10 equal portions by lower section, the higher pixel of permanent magnet synchronous motor surface temperature described in statistic procedure S240 Regional temperature histogram；

Step S340: the mean value of the higher pixel region temperature histogram of permanent magnet synchronous motor surface temperature, standard deviation, partially are calculated State, kurtosis and entropy；

Step S350: the curvilinear characteristic of pixel maximum temperature values described in extraction step S310 changed over time；

Step S360: a series of timing temperature of the thermal transformation of the jth time described in step S220 by filtering processing are calculated The hot stable state interframe variance vibration performance of image.

6. permanent magnet synchronous motor fault detection method according to claim 5, which is characterized in that the step S400 is specific Are as follows:

Step S410: the feature of the extraction of the thermal transformation of jth time is formed into feature vector；

Step S420: motor status classification is determined according to feature vector.

7. permanent magnet synchronous motor fault detection method according to claim 6, which is characterized in that the step S600 is specific Are as follows:

Step S610: the feature vector and its corresponding motor status classification extracted by the thermal transformation of n times motor are established The training set of classifier；

Step S620: classifier uses support vector machines, kernel function uses Radial basis kernel function, and the training set of classification is to extract Feature vector is as input, and using motor status classification as output, Training Support Vector Machines obtain permanent magnet synchronous motor failure and examine Disconnected supporting vector machine model.

8. a kind of permanent magnet synchronous motor fault detection system, which is characterized in that including image capture module, image preprocessing mould Block, characteristic extracting module, model building module and fault detection module, in which:

Image capture module, for acquire permanent magnet synchronous motor normal and tape jam thermal transformation it is N number of a series of when Sequence raw image data, j=1, j≤N are sent to image pre-processing module；

A series of timing of image pre-processing module, the thermal transformation of the jth time for sending image capture module are original Image data is partitioned into the higher pixel region of permanent magnet synchronous motor surface temperature using image processing techniques, is sent to feature and mentions Modulus block；

Characteristic extracting module, the thermal transformation of the jth by image procossing time for sending image pre-processing module A series of time sequence image datas extract feature, including temperature variation curve feature, interframe variance vibration performance and electricity The higher pixel region histogram correlated characteristic of machine surface temperature, is sent to model building module；

Model building module, the extraction feature for being sent according to characteristic extracting module determine motor status classification；Judging j > N is No establishment, if not, j=j+1 returns to image pre-processing module, if so, it is then special by the thermal transformation of n times motor Sign is extracted and motor status classification determines, establishes permanent magnet synchronous motor fault diagnosis supporting vector machine model, is sent to failure inspection Survey module；

Fault detection module, for using the trained model of model building module to carry out failure to the permanent magnet synchronous motor at scene Differentiate.

9. permanent magnet synchronous motor fault detection system according to claim 8, which is characterized in that described image acquisition module It is middle to acquire raw image data using thermal infrared imager.

10. permanent magnet synchronous motor fault detection system according to claim 9, which is characterized in that the thermal infrared imager When acquiring raw image data, thermal infrared imager puts permanent magnet synchronous motor center side to be aligned, thermal infrared imager and permanent magnetism Synchronous motor is located at same level height, and the selection of horizontal distance between the two will make thermal infrared imager acquired image packet Entire permanent magnet synchronous motor is included, both acquisitions relative position immobilizes every time.