Rolling bearing fault diagnosis method and system
By performing time-frequency conversion and particle movement of vibration signals of rolling bearings, the ridges of the time-frequency matrix are determined, which solves the problem of insufficient fault diagnosis accuracy in the prior art and achieves higher fault diagnosis accuracy.
Patent Information
- Application Number
- CN202411553404.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-02
- Publication Date
- 2025-03-14
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing rolling bearing fault diagnosis methods are prone to problems of missed fault detection and misjudgment when noise interference and nonlinear characteristics are significant.
By collecting the vibration signals of the rolling bearings, performing time-frequency transformation to obtain the time-frequency matrix, initializing multiple particles and measurement matrices, and updating the measurement matrix through horizontal and vertical movement of the particles, and finally determining the ridge of the time-frequency matrix to diagnose the fault.
It improves the accuracy of fault diagnosis, reduces the impact of noise interference, and reduces the situation of missed fault detection and misjudgment.
Smart Images
Figure CN119622399A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of rolling bearings, and in particular to a rolling bearing fault diagnosis method and system. Background Art
[0002] Rolling bearings are lubricants for the operation of different components of equipment. Rolling bearings can support rotating shafts and reduce friction between moving parts, thereby improving the efficiency and stability of mechanical transmission. However, since rolling bearings usually work under high loads, high speeds and complex environments, they are easily affected by external stress, vibration, temperature changes and other factors, resulting in wear, cracks, fatigue and other faults. If these faults cannot be detected and handled in time, they will not only affect the normal operation of the equipment, but may also cause serious equipment damage and even cause safety accidents. Therefore, rolling bearing fault diagnosis technology plays a vital role in mechanical equipment maintenance. At present, rolling bearing fault diagnosis methods mainly rely on signal processing and pattern recognition technologies, such as time domain, frequency domain and time-frequency analysis methods. However, these traditional methods often face difficulties in dealing with complex fault signals, especially in the case of noise interference and significant nonlinear characteristics, which are prone to fault omission and misjudgment. Summary of the invention
[0003] In order to solve the above problems, the present invention provides a rolling bearing fault diagnosis method, which comprises the following steps: Collecting vibration signals of rolling bearings, performing time-frequency transformation on the preprocessed vibration signals to obtain a time-frequency matrix, initializing a plurality of mass points according to the time-frequency matrix, and initializing a metric matrix, an initial temperature, and a number of iterations; In each iteration, for each particle, three points on the left and three points on the right of the particle are obtained in the time-frequency matrix, the horizontal movement direction of the particle is determined based on the values of the points obtained from the time-frequency matrix, after the horizontal movement, the average value of the upper and lower neighborhood points of the particle is obtained, the vertical movement direction of the particle is determined according to the average value and the temperature, and the metric matrix is updated according to the position of the particle in the time-frequency matrix; After all iterations are completed, the ridge of the time-frequency matrix is determined according to the metric matrix, and the rolling bearing fault condition is determined based on the extracted ridge.
[0004] Preferably, the initializing multiple particles according to the time-frequency matrix is specifically: Normalize the value of each position in the time-frequency matrix, and calculate the gradient of each position in the normalized time-frequency matrix to obtain the gradient map of the time-frequency matrix. Calculate the weight of each position based on the normalized gradient map and the normalized time-frequency matrix to obtain the weight matrix of the time-frequency matrix. The weight matrix is expanded into a one-dimensional vector, and the cumulative distribution function of the one-dimensional vector is calculated. A random number is generated in the range of [0,1]. According to the area where the random number falls in the cumulative distribution function, the index corresponding to the area in the cumulative distribution function is determined, and the index is mapped to the position of the time-frequency matrix to obtain the initial position of the particle.
[0005] Preferably, the determining of the horizontal movement direction of the particle based on the value of the point obtained from the time-frequency matrix is specifically: If the particle is located on the left boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the right; If the particle is located on the right boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the left; If the particle is neither on the left boundary nor on the right boundary of the time-frequency matrix, the point to the left of the particle and the upper and lower adjacent points on the left are taken as the three points on the left side of the particle, and the point to the right of the particle and the upper and lower adjacent points on the right are taken as the three points on the right side of the particle. The average value of the three points on the left side of the particle and the average value of the three points on the right side of the particle are calculated, and the probability of the particle moving left and the probability of moving right are calculated based on the ratio of the average values. The horizontal movement direction of the particle is determined according to the probability of moving left and the probability of moving right.
[0006] Preferably, the vertical moving direction of the particle is determined according to the average value and the temperature, specifically: If the particle is located on the upper boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point below the particle, the vertical movement direction of the particle is downward, otherwise, it does not move; If the particle is located at the lower boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point above the particle, the vertical movement direction of the particle is upward, otherwise, it does not move; If the particle is neither on the upper boundary nor on the lower boundary of the time-frequency matrix, then the average value of the eight neighborhoods of the points above the particle is obtained, and the average value of the eight neighborhoods of the points below the particle is obtained, and the probability that the particle does not move vertically and the probability that it moves vertically are calculated according to the temperature; if the average value of the eight neighborhoods of the points above the particle is greater than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves upward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is less than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is equal to the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is half of the probability of the vertical movement, and the probability that the particle moves upward is half of the probability of the vertical movement; the vertical movement direction of the particle is obtained according to the movement probability.
[0007] Preferably, the updating of the metric matrix according to the position of the particle in the time-frequency matrix is specifically: The position of the particle in the time-frequency matrix is obtained, a measurement value is obtained based on the value of the particle in the time-frequency matrix, the position of the particle in the measurement matrix is determined according to the position of the particle in the time-frequency matrix, and the measurement value is increased by the value at the position in the measurement matrix.
[0008] Preferably, the determining of the ridge line of the time-frequency matrix according to the metric matrix is specifically: Adjust the values of points in the measurement matrix that are not greater than the threshold to 0; For each non-zero point in the metric matrix, calculate whether the non-zero point is the maximum point in the neighborhood, and if so, take the non-zero point as the point of the ridge line; The points of adjacent ridge lines are connected to obtain at least one ridge line, and the ridge line is processed and then drawn in the time-frequency matrix.
[0009] On the other hand, the present invention also provides a rolling bearing fault diagnosis system, the system comprising the following modules: A vibration acquisition and initialization module is used to acquire the vibration signal of the rolling bearing, perform time-frequency transformation on the preprocessed vibration signal to obtain a time-frequency matrix, initialize multiple particles according to the time-frequency matrix, and initialize the metric matrix, initial temperature and number of iterations; A ridge point calculation module is used to obtain, for each particle in each iteration, three points on the left side and three points on the right side of the particle in the time-frequency matrix, determine the horizontal movement direction of the particle based on the value of the point obtained from the time-frequency matrix, obtain the average value of the upper and lower neighborhood points of the particle after the horizontal movement, determine the vertical movement direction of the particle according to the average value and the temperature, and update the metric matrix according to the position of the particle in the time-frequency matrix; The fault identification module is used to determine the ridge of the time-frequency matrix according to the metric matrix after all iterations are completed, and determine the rolling bearing fault condition based on the extracted ridge.
[0010] Preferably, the initializing multiple particles according to the time-frequency matrix is specifically: Normalize the value of each position in the time-frequency matrix, and calculate the gradient of each position in the normalized time-frequency matrix to obtain the gradient map of the time-frequency matrix. Calculate the weight of each position based on the normalized gradient map and the normalized time-frequency matrix to obtain the weight matrix of the time-frequency matrix. The weight matrix is expanded into a one-dimensional vector, and the cumulative distribution function of the one-dimensional vector is calculated. A random number is generated in the range of [0,1]. According to the area where the random number falls in the cumulative distribution function, the index corresponding to the area in the cumulative distribution function is determined, and the index is mapped to the position of the time-frequency matrix to obtain the initial position of the particle.
[0011] Preferably, the determining of the horizontal movement direction of the particle based on the value of the point obtained from the time-frequency matrix is specifically: If the particle is located on the left boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the right; If the particle is located on the right boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the left; If the particle is neither on the left boundary nor on the right boundary of the time-frequency matrix, the point to the left of the particle and the upper and lower adjacent points on the left are taken as the three points on the left side of the particle, and the point to the right of the particle and the upper and lower adjacent points on the right are taken as the three points on the right side of the particle. The average value of the three points on the left side of the particle and the average value of the three points on the right side of the particle are calculated, and the probability of the particle moving left and the probability of moving right are calculated based on the ratio of the average values. The horizontal movement direction of the particle is determined according to the probability of moving left and the probability of moving right.
[0012] Preferably, the vertical moving direction of the particle is determined according to the average value and the temperature, specifically: If the particle is located on the upper boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point below the particle, the vertical movement direction of the particle is downward, otherwise, it does not move; If the particle is located at the lower boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point above the particle, the vertical movement direction of the particle is upward, otherwise, it does not move; If the particle is neither on the upper boundary nor on the lower boundary of the time-frequency matrix, then the average value of the eight neighborhoods of the points above the particle is obtained, and the average value of the eight neighborhoods of the points below the particle is obtained, and the probability that the particle does not move vertically and the probability that it moves vertically are calculated according to the temperature; if the average value of the eight neighborhoods of the points above the particle is greater than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves upward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is less than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is equal to the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is half of the probability of the vertical movement, and the probability that the particle moves upward is half of the probability of the vertical movement; the vertical movement direction of the particle is obtained according to the movement probability.
[0013] Preferably, the updating of the metric matrix according to the position of the particle in the time-frequency matrix is specifically: The position of the particle in the time-frequency matrix is obtained, a measurement value is obtained based on the value of the particle in the time-frequency matrix, the position of the particle in the measurement matrix is determined according to the position of the particle in the time-frequency matrix, and the measurement value is increased by the value at the position in the measurement matrix.
[0014] Preferably, the determining of the ridge line of the time-frequency matrix according to the metric matrix is specifically: Adjust the values of points in the measurement matrix that are not greater than the threshold to 0; For each non-zero point in the metric matrix, calculate whether the non-zero point is the maximum point in the neighborhood, and if so, take the non-zero point as the point of the ridge line; The points of adjacent ridge lines are connected to obtain at least one ridge line, and the ridge line is processed and then drawn in the time-frequency matrix.
[0015] Finally, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in the first aspect is implemented.
[0016] Ridge extraction is the key to rolling bearing fault identification. The present invention optimizes the movement mode of particles in ridge extraction. Specifically, the horizontal movement direction is determined according to the situation of the left and right neighborhoods of the particles, and the vertical movement direction is determined according to the average value and temperature of the upper and lower neighborhood points after the particles move horizontally. The present invention also improves the initialization position of the particles. By considering the left and right, upper and lower neighborhood information of the particles, the local maximum in the time-frequency diagram can be better identified, the influence of noise can be reduced, and the ridges can be extracted more accurately, thereby improving the accuracy of identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 This is a flow chart of Embodiment 1; Figure 2 Schematic diagram of three points on the left and three points on the right of the particle; Figure 3 It is a schematic diagram of eight neighborhoods; Figure 4 Schematic diagram of ridge extraction. DETAILED DESCRIPTION
[0018] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.
[0019] In the technical solutions disclosed in this disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the relevant laws and regulations and do not violate public order and good morals. In the technical solutions disclosed in this disclosure, the user's authorization or consent is obtained before obtaining or collecting user personal information.
[0020] The technical solution provided by the present disclosure will be described in detail below with reference to the accompanying drawings and specific embodiments.
[0021] Figure 1 A first embodiment of the present invention is shown. Figure 1 S1 is: collecting the vibration signal of the rolling bearing, performing time-frequency transformation on the preprocessed vibration signal to obtain a time-frequency matrix, initializing multiple particles according to the time-frequency matrix, and initializing the metric matrix, initial temperature and number of iterations; During the operation of the rolling bearing, the vibration signal of the rolling bearing is collected. Preferably, a vibration sensor is used for collection. The vibration sensor is installed on the housing of the rolling bearing or a nearby structure. Due to the influence of noise, the vibration signal is preprocessed after collection. The preprocessing method is filtering and denoising. Then the vibration signal is converted into a time-frequency matrix through short-time Fourier transform or wavelet transform. The time-frequency matrix is an energy distribution matrix of the vibration signal in two dimensions of time and frequency. The time dimension corresponds to the horizontal axis, the frequency dimension corresponds to the vertical axis, and the value is energy or amplitude.
[0022] In the ridge extraction using Crazy Climber, it is necessary to initialize multiple mass points (also called movable points, Climber), and initialize the metric matrix, initial temperature and number of iterations, where the size of the metric matrix is the same as the time-frequency matrix, both of which are M×N, the initial temperature is set to the difference between the maximum and minimum values in the time-frequency matrix, and the number of iterations is set to K. The initialization positions of the mass points are preferably evenly distributed to the time-frequency matrix according to the number of mass points. For example, if there are 9 mass points, the time-frequency matrix is divided into 9 regions of equal size, and the center of each region is the initial position of a mass point.
[0023] Since the particle is the basis for extracting the ridge, the distribution of the particle is not only related to the number of particle movement times, but also related to the center of the metric matrix. In another embodiment, the initialization of multiple particle points according to the time-frequency matrix is specifically: Normalize the value of each position in the time-frequency matrix, and calculate the gradient of each position in the normalized time-frequency matrix to obtain the gradient map of the time-frequency matrix. Calculate the weight of each position based on the normalized gradient map and the normalized time-frequency matrix to obtain the weight matrix of the time-frequency matrix. The weight matrix is expanded into a one-dimensional vector, and the cumulative distribution function of the one-dimensional vector is calculated. A random number is generated in the range of [0,1]. According to the area where the random number falls in the cumulative distribution function, the index corresponding to the area in the cumulative distribution function is determined, and the index is mapped to the position of the time-frequency matrix to obtain the initial position of the particle.
[0024] First, the time-frequency matrix is normalized, and then the gradient of each position is calculated to obtain a normalized time-frequency matrix and a normalized gradient map. For the same position in the time-frequency matrix, the weight of the position is obtained by the median of the normalized time-frequency matrix and the value of the normalized gradient map. Preferably, the weight calculation method is to perform weighted calculation on the median of the time-frequency matrix and the value in the gradient map at the same position. For example, the value of position [5,19] in the time-frequency matrix is 0.8, the value of position [5,19] in the gradient map is 0.3, the weight of the time-frequency matrix is 0.7, the weight of the gradient map is 0.3, and the weight of the position [5,19] is 0.65.
[0025] The weight matrix is expanded into a one-dimensional vector, and the cumulative distribution function is calculated. The larger the weight, the greater the difference in the cumulative distribution function values of adjacent elements in the one-dimensional vector. A random number is generated, and the probability that the random number falls into this interval is greater, so that the probability of the element with a large weight being selected increases. If an element is selected, the position of the element in the weight matrix can be obtained according to the conversion relationship between the one-dimensional vector and the weight matrix. Since the weight matrix and the time-frequency matrix are of the same size, the position in the time-frequency matrix can be obtained, and the position obtained in the time-frequency matrix is used as the initial position of the particle. If there are 100 particles, 100 random numbers are generated to obtain 100 positions, which are the initial positions of the 100 particles. This method of determining the initial position of the particle is not only random, but also affected by the time-frequency matrix, making the particle more inclined to the initial value or the position of the time-frequency matrix with large changes.
[0026] Figure 1 S2 is: in each iteration, for each particle, three points on the left and three points on the right of the particle are obtained in the time-frequency matrix, the horizontal movement direction of the particle is determined based on the value of the point obtained from the time-frequency matrix, after the horizontal movement, the average value of the upper and lower neighborhood points of the particle is obtained, the vertical movement direction of the particle is determined according to the average value and the temperature, and the metric matrix is updated according to the position of the particle in the time-frequency matrix; In Crazy Climber, the particle moves left and right with equal probability, but this method may cause oscillation and movement to non-ridge lines. In the present invention, three points on the left side and three points on the right side of the particle are obtained, and the horizontal movement direction is determined according to these six points, wherein the three points on the left side of the particle are the point on the left side of the particle and the two points above and below the point on the left side. Assuming that the coordinates of the particle are [i, j], the point on the left side is [i-1, j], and the point on the right side is [i+1, j]. The three points on the left side of the particle are [i-1, j-1], [i-1, j], and [i-1, j+1], and the three points on the right side of the particle are [i+1, j-1], [i+1, j], and [i+1, j+1], respectively. Figure 2If the particle is located at the edge, there is no need to calculate the area beyond the time-frequency matrix. In one implementation, the horizontal movement direction of the particle is determined based on the value of the point obtained from the time-frequency matrix, specifically: If the particle is located on the left boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the right; If the particle is located on the right boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the left; If the particle is neither on the left boundary nor on the right boundary of the time-frequency matrix, the point to the left of the particle and the upper and lower adjacent points on the left are taken as the three points on the left side of the particle, and the point to the right of the particle and the upper and lower adjacent points on the right are taken as the three points on the right side of the particle. The average value of the three points on the left side of the particle and the average value of the three points on the right side of the particle are calculated, and the probability of the particle moving left and the probability of moving right are calculated based on the ratio of the average values. The horizontal movement direction of the particle is determined according to the probability of moving left and the probability of moving right.
[0027] In the horizontal direction, there are two ways of moving, namely, moving to the left and moving to the right. If the particle is located on the left boundary, it cannot move to the left and can only move to the right; similarly, if the particle is located on the right boundary, it cannot move to the right and can only move to the left. If the particle is neither located on the left boundary nor on the boundary, it can move to the left or to the right. Calculate the upper and lower adjacent points of the left point of the particle, so that the left point of the particle and the upper and lower adjacent points of the left point are obtained, a total of three points, and calculate the average value of these three points on the time-frequency matrix. Similarly, the average value of the point on the right side of the particle and the upper and lower adjacent points of the right point can be calculated. The probability of the particle moving left and the probability of moving right are calculated based on the ratio of the average values. Preferably, the probability of the particle moving left and the probability of the particle moving right are calculated based on the ratio of the average values as follows: set the probability of moving left and the probability of moving right to be a and b respectively, the ratio of a and b is equal to the ratio of the average values, and a+b=1. Then determine the horizontal movement direction of the particle according to the probability of moving left and the probability of moving right. In this embodiment, the particle tends to move in the direction of a larger average value, which can speed up the movement to the extreme point. It should be noted that if the particle is located at the upper and lower boundaries, and there is only one upper and lower adjacent point between the right point and the left point of the particle, then the average value is calculated from the right point of the particle and one adjacent point of the right point, and the same is true for the left point of the particle. In one embodiment, if the adjacent point of the right point and the left point of the particle exceeds the range of the time-frequency matrix, the value of the adjacent point is set to 0.
[0028] There are three ways for a particle to move in the vertical direction, namely, moving upward, moving downward, and not moving. In a preferred embodiment, the vertical movement direction of the particle is determined according to the average value and the temperature, specifically: If the particle is located on the upper boundary of the time-frequency matrix, if the value of the particle in the time-frequency matrix is less than the value of the point below the particle, the vertical movement direction of the particle is downward, otherwise, it does not move; when the particle is located at the upper boundary point, it can only not move or move downward. If the value of the particle in the time-frequency matrix is less than the value of the point below the particle, it moves downward, otherwise it does not move. It should be noted that in this embodiment, the particle position refers to the new position of the particle after horizontal movement.
[0029] If the particle is located at the lower boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point above the particle, the vertical movement direction of the particle is upward, otherwise, it does not move; when the particle is located at the lower boundary point, it can only not move or move upward. If the value of the particle in the time-frequency matrix is less than the value of the point above the particle, it moves upward, otherwise it does not move.
[0030] If the particle is neither on the upper boundary nor on the lower boundary of the time-frequency matrix, then the average value of the eight neighborhoods of the points above the particle is obtained, and the average value of the eight neighborhoods of the points below the particle is obtained, and the probability that the particle does not move vertically and the probability that it moves vertically are calculated according to the temperature; if the average value of the eight neighborhoods of the points above the particle is greater than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves upward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is less than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is equal to the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is half of the probability of the vertical movement, and the probability that the particle moves upward is half of the probability of the vertical movement; the vertical movement direction of the particle is obtained according to the movement probability.
[0031] If the particle is not located at the upper or lower boundary of the time-frequency matrix, the average value of the eight neighborhoods of the points above the particle and the average value of the eight neighborhoods of the points below the particle are obtained, such as Figure 3As shown. The probability that the particle does not move vertically is calculated using temperature, where the probability of not moving vertically is equal to 1 minus the probability of moving vertically, and the probability of vertical movement is positively correlated with temperature, where the temperature continuously decreases on the basis of the initial temperature. If the average value of the eight neighborhoods of the points above the particle is less than the average value of the eight neighborhoods of the points below the particle, the probability that the particle moves downward is the probability of vertical movement; if the average value of the eight neighborhoods of the points above the particle is equal to the average value of the eight neighborhoods of the points below the particle, the probability that the particle moves downward is half of the probability of vertical movement, and the probability that the particle moves upward is half of the probability of vertical movement; the vertical movement direction of the particle is obtained according to the movement probability. That is, if the particle moves vertically, the particle is more inclined to move in the direction with a larger average value. As the temperature decreases, the probability of vertical movement also decreases and finally stabilizes. Similarly, if there are points in the eight neighborhoods that exceed the range of the time-frequency matrix, the values of the points in the excess are set to 0.
[0032] In one iteration, all particles determine a movement mode by determining the horizontal movement direction and the vertical movement direction. After the movement mode is determined, the particles move to a new position, and the measurement matrix is updated according to the new position of the particles. In one embodiment, when a particle moves to a new position, the value of the new position is added to the measurement matrix by 1. In another embodiment, the measurement matrix is updated according to the position of the particle in the time-frequency matrix, specifically: The position of the particle in the time-frequency matrix is obtained, a measurement value is obtained based on the value of the particle in the time-frequency matrix, the position of the particle in the measurement matrix is determined according to the position of the particle in the time-frequency matrix, and the measurement value is increased by the value at the position in the measurement matrix.
[0033] The time-frequency matrix and the metric matrix are of the same size. When a particle moves to a new position, the value of the new position in the time-frequency matrix is obtained, the value is used as the metric value, and the metric value is added to the element at the same position in the metric matrix.
[0034] Figure 1 S3 is as follows: After all iterations are completed, the ridge line of the time-frequency matrix is determined according to the metric matrix, and the rolling bearing fault condition is determined based on the extracted ridge line.
[0035] After all iterations are completed, the ridge line of the time-frequency matrix is determined according to the metric matrix. In one embodiment, the ridge line is: Adjust the values of points in the measurement matrix that are not greater than the threshold to 0; For each non-zero point in the metric matrix, calculate whether the non-zero point is the maximum point in the neighborhood, and if so, take the non-zero point as the point of the ridge line; The points of adjacent ridge lines are connected to obtain at least one ridge line, and the ridge line is processed and then drawn in the time-frequency matrix.
[0036] In another embodiment, the ridge line is determined according to the ridge line extraction method in the Crazy Climber algorithm. Figure 4 The extracted ridges are shown.
[0037] After the ridge lines are determined, the fault conditions of the rolling bearing are identified according to the distribution of the ridge lines. Specifically, each ridge line is matched with the fault ridge line. If they match, it means that the corresponding fault exists, otherwise, the corresponding fault does not exist.
[0038] The present invention also provides a rolling bearing fault diagnosis system, the system comprising the following modules: A vibration acquisition and initialization module is used to acquire the vibration signal of the rolling bearing, perform time-frequency transformation on the preprocessed vibration signal to obtain a time-frequency matrix, initialize multiple particles according to the time-frequency matrix, and initialize the metric matrix, initial temperature and number of iterations; A ridge point calculation module is used to obtain, for each particle in each iteration, three points on the left side and three points on the right side of the particle in the time-frequency matrix, determine the horizontal movement direction of the particle based on the value of the point obtained from the time-frequency matrix, obtain the average value of the upper and lower neighborhood points of the particle after the horizontal movement, determine the vertical movement direction of the particle according to the average value and the temperature, and update the metric matrix according to the position of the particle in the time-frequency matrix; The fault identification module is used to determine the ridge of the time-frequency matrix according to the metric matrix after all iterations are completed, and determine the rolling bearing fault condition based on the extracted ridge.
[0039] Preferably, the initializing multiple particles according to the time-frequency matrix is specifically: Normalize the value of each position in the time-frequency matrix, and calculate the gradient of each position in the normalized time-frequency matrix to obtain the gradient map of the time-frequency matrix. Calculate the weight of each position based on the normalized gradient map and the normalized time-frequency matrix to obtain the weight matrix of the time-frequency matrix. The weight matrix is expanded into a one-dimensional vector, and the cumulative distribution function of the one-dimensional vector is calculated. A random number is generated in the range of [0,1]. According to the area where the random number falls in the cumulative distribution function, the index corresponding to the area in the cumulative distribution function is determined, and the index is mapped to the position of the time-frequency matrix to obtain the initial position of the particle.
[0040] Preferably, the determining of the horizontal movement direction of the particle based on the value of the point obtained from the time-frequency matrix is specifically: If the particle is located on the left boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the right; If the particle is located on the right boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the left; If the particle is neither on the left boundary nor on the right boundary of the time-frequency matrix, the point to the left of the particle and the upper and lower adjacent points on the left are taken as the three points on the left side of the particle, and the point to the right of the particle and the upper and lower adjacent points on the right are taken as the three points on the right side of the particle. The average value of the three points on the left side of the particle and the average value of the three points on the right side of the particle are calculated, and the probability of the particle moving left and the probability of moving right are calculated based on the ratio of the average values. The horizontal movement direction of the particle is determined according to the probability of moving left and the probability of moving right.
[0041] Preferably, the vertical moving direction of the particle is determined according to the average value and the temperature, specifically: If the particle is located on the upper boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point below the particle, the vertical movement direction of the particle is downward, otherwise, it does not move; If the particle is located at the lower boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point above the particle, the vertical movement direction of the particle is upward, otherwise, it does not move; If the particle is neither on the upper boundary nor on the lower boundary of the time-frequency matrix, then the average value of the eight neighborhoods of the points above the particle is obtained, and the average value of the eight neighborhoods of the points below the particle is obtained, and the probability that the particle does not move vertically and the probability that it moves vertically are calculated according to the temperature; if the average value of the eight neighborhoods of the points above the particle is greater than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves upward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is less than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is equal to the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is half of the probability of the vertical movement, and the probability that the particle moves upward is half of the probability of the vertical movement; the vertical movement direction of the particle is obtained according to the movement probability.
[0042] Preferably, the updating of the metric matrix according to the position of the particle in the time-frequency matrix is specifically: The position of the particle in the time-frequency matrix is obtained, a measurement value is obtained based on the value of the particle in the time-frequency matrix, the position of the particle in the measurement matrix is determined according to the position of the particle in the time-frequency matrix, and the measurement value is increased by the value at the position in the measurement matrix.
[0043] Preferably, the determining of the ridge line of the time-frequency matrix according to the metric matrix is specifically: Adjust the values of points in the measurement matrix that are not greater than the threshold to 0; For each non-zero point in the metric matrix, calculate whether the non-zero point is the maximum point in the neighborhood, and if so, take the non-zero point as the point of the ridge line; The points of adjacent ridge lines are connected to obtain at least one ridge line, and the ridge line is processed and then drawn in the time-frequency matrix.
[0044] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, including at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned virtual switch flow table update method in the enterprise website.
[0045] According to an embodiment of the present disclosure, the present disclosure further provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the above-mentioned method for updating a virtual switch flow table in an enterprise website.
[0046] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.
[0047] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0048] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0049] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.
[0050] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.
Claims
1. A rolling bearing fault diagnosis method, characterized in that: The method comprises the following steps: Collecting vibration signals of rolling bearings, performing time-frequency transformation on the preprocessed vibration signals to obtain a time-frequency matrix, initializing a plurality of mass points according to the time-frequency matrix, and initializing a metric matrix, an initial temperature, and a number of iterations; In each iteration, for each particle, three points on the left and three points on the right of the particle are obtained in the time-frequency matrix, the horizontal movement direction of the particle is determined based on the values of the points obtained from the time-frequency matrix, after the horizontal movement, the average value of the upper and lower neighborhood points of the particle is obtained, the vertical movement direction of the particle is determined according to the average value and the temperature, and the metric matrix is updated according to the position of the particle in the time-frequency matrix; After all iterations are completed, the ridge of the time-frequency matrix is determined according to the metric matrix, and the rolling bearing fault condition is determined based on the extracted ridge.
2. The method according to claim 1, characterized in that The initialization of multiple particles according to the time-frequency matrix is specifically as follows: Normalize the value of each position in the time-frequency matrix, and calculate the gradient of each position in the normalized time-frequency matrix to obtain the gradient map of the time-frequency matrix. Calculate the weight of each position based on the normalized gradient map and the normalized time-frequency matrix to obtain the weight matrix of the time-frequency matrix. The weight matrix is expanded into a one-dimensional vector, and the cumulative distribution function of the one-dimensional vector is calculated. A random number is generated in the range of [0,1]. According to the area where the random number falls in the cumulative distribution function, the index corresponding to the area in the cumulative distribution function is determined, and the index is mapped to the position of the time-frequency matrix to obtain the initial position of the particle.
3. The method according to claim 1, characterized in that The method of determining the horizontal moving direction of the particle based on the value of the point obtained from the time-frequency matrix is specifically as follows: If the particle is located on the left boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the right; If the particle is located on the right boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the left; If the particle is neither on the left boundary nor on the right boundary of the time-frequency matrix, the point to the left of the particle and the upper and lower adjacent points on the left are taken as the three points on the left side of the particle, and the point to the right of the particle and the upper and lower adjacent points on the right are taken as the three points on the right side of the particle. The average value of the three points on the left side of the particle and the average value of the three points on the right side of the particle are calculated, and the probability of the particle moving left and the probability of moving right are calculated based on the ratio of the average values. The horizontal movement direction of the particle is determined according to the probability of moving left and the probability of moving right.
4. The method according to claim 1, characterized in that The method of determining the vertical moving direction of the particle according to the average value and the temperature is specifically as follows: If the particle is located on the upper boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point below the particle, the vertical movement direction of the particle is downward, otherwise, it does not move; If the particle is located at the lower boundary of the time-frequency matrix, then if the value of the particle in the time-frequency matrix is less than the value of the point above the particle, the vertical movement direction of the particle is upward, otherwise, it does not move; If the particle is neither on the upper boundary nor on the lower boundary of the time-frequency matrix, then the average value of the eight neighborhoods of the points above the particle is obtained, and the average value of the eight neighborhoods of the points below the particle is obtained, and the probability that the particle does not move vertically and the probability that it moves vertically are calculated according to the temperature; if the average value of the eight neighborhoods of the points above the particle is greater than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves upward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is less than the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is the probability of the vertical movement; if the average value of the eight neighborhoods of the points above the particle is equal to the average value of the eight neighborhoods of the points below the particle, then the probability that the particle moves downward is half of the probability of the vertical movement, and the probability that the particle moves upward is half of the probability of the vertical movement; the vertical movement direction of the particle is obtained according to the movement probability.
5. The method according to claim 1, characterized in that The updating of the metric matrix according to the position of the particle in the time-frequency matrix is specifically as follows: The position of the particle in the time-frequency matrix is obtained, a measurement value is obtained based on the value of the particle in the time-frequency matrix, the position of the particle in the measurement matrix is determined according to the position of the particle in the time-frequency matrix, and the measurement value is increased by the value at the position in the measurement matrix.
6. The method according to claim 1, characterized in that The ridge line of the time-frequency matrix is determined according to the metric matrix, specifically: Adjust the values of points in the measurement matrix that are not greater than the threshold to 0; For each non-zero point in the metric matrix, calculate whether the non-zero point is the maximum point in the neighborhood, and if so, take the non-zero point as the point of the ridge line; The points of adjacent ridge lines are connected to obtain at least one ridge line, and the ridge line is processed and then drawn in the time-frequency matrix.
7. A rolling bearing fault diagnosis system, characterized in that: The system includes the following modules: A vibration acquisition and initialization module is used to acquire the vibration signal of the rolling bearing, perform time-frequency transformation on the preprocessed vibration signal to obtain a time-frequency matrix, initialize multiple particles according to the time-frequency matrix, and initialize the metric matrix, initial temperature and number of iterations; A ridge point calculation module is used to obtain, for each particle in each iteration, three points on the left side and three points on the right side of the particle in the time-frequency matrix, determine the horizontal movement direction of the particle based on the value of the point obtained from the time-frequency matrix, obtain the average value of the upper and lower neighborhood points of the particle after the horizontal movement, determine the vertical movement direction of the particle according to the average value and the temperature, and update the metric matrix according to the position of the particle in the time-frequency matrix; The fault identification module is used to determine the ridge of the time-frequency matrix according to the metric matrix after all iterations are completed, and determine the rolling bearing fault condition based on the extracted ridge.
8. The system according to claim 7, characterized in that The initialization of multiple particles according to the time-frequency matrix is specifically as follows: Normalize the value of each position in the time-frequency matrix, and calculate the gradient of each position in the normalized time-frequency matrix to obtain the gradient map of the time-frequency matrix. Calculate the weight of each position based on the normalized gradient map and the normalized time-frequency matrix to obtain the weight matrix of the time-frequency matrix. The weight matrix is expanded into a one-dimensional vector, and the cumulative distribution function of the one-dimensional vector is calculated. A random number is generated in the range of [0,1]. According to the area where the random number falls in the cumulative distribution function, the index corresponding to the area in the cumulative distribution function is determined, and the index is mapped to the position of the time-frequency matrix to obtain the initial position of the particle.
9. The system according to claim 7, characterized in that The method of determining the horizontal moving direction of the particle based on the value of the point obtained from the time-frequency matrix is specifically as follows: If the particle is located on the left boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the right; If the particle is located on the right boundary of the time-frequency matrix, the horizontal movement direction of the particle is to the left; If the particle is neither on the left boundary nor on the right boundary of the time-frequency matrix, the point to the left of the particle and the upper and lower adjacent points on the left are taken as the three points on the left side of the particle, and the point to the right of the particle and the upper and lower adjacent points on the right are taken as the three points on the right side of the particle. The average value of the three points on the left side of the particle and the average value of the three points on the right side of the particle are calculated, and the probability of the particle moving left and the probability of moving right are calculated based on the ratio of the average values. The horizontal movement direction of the particle is determined according to the probability of moving left and the probability of moving right.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the method according to any one of claims 1 to 6.