Motor pulse energy conversion system bifurcation type fault detection method
By applying DTW and CUSUM technologies in pulse energy conversion systems, combined with adaptive sampling criteria, the problems of low reliability and poor applicability of bifurcation fault detection in the prior art are solved, and more efficient and reliable fault detection is achieved.
Patent Information
- Application Number
- CN202510232218.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-28
- Publication Date
- 2025-05-30
AI Technical Summary
The prior art detects bifurcation faults of pulse energy conversion systems with low detection reliability, especially in the context of strong disturbance noise.
Dynamic time regularization (DTW) method is used to process irregular time series data, identify pattern changes, and accumulate weak fault information in the adaptive data through CUSUM technology, and determine the data acquisition amount in combination with adaptive sampling criteria.
It improves the sensitivity and reliability of bifurcation fault detection, can effectively handle strong background noise signals, and reduces data acquisition, transmission, processing and storage costs.
Smart Images
Figure CN120064970A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of fault diagnosis of electromechanical systems, and particularly relates to a method for detecting bifurcation faults of a motor pulse energy conversion system. Background Art
[0002] The Pulse Energy Conversion System (PECS) is widely used in various motor devices. It adopts Pulse Width Modulation (PWM) technology to control the performance parameters of motor devices, such as rotational speed, torque, response speed, etc., by controlling the duty cycle to meet different requirements. The pulse energy conversion system has strong dynamic characteristics and complex working modes. Especially under high load or impact load, it is prone to bifurcation faults. Under the influence of external or internal factors, the bifurcation faults of the pulse energy conversion system not only lead to a decrease in the energy conversion efficiency and stability of the motor, but also may cause the collapse of the electromechanical system. Therefore, appropriate methods should be developed to detect the bifurcation faults of the pulse energy conversion system in time to avoid the impact of faults on the working performance of the motor system.
[0003] At present, in order to reliably detect bifurcation faults and avoid their hazards, some scholars have proposed methods for detecting bifurcation faults of the pulse energy conversion system. Russian scholar Yury V. Kolokolov detected the bifurcation faults of the pulse energy conversion system based on the Poincaré map theory. Although this method can detect the bifurcation faults of the pulse energy conversion system, the detection reliability is relatively low. If the initial point of the Poincaré map is not selected properly, the period-2 bifurcation faults cannot be detected. In addition, under the background of strong disturbance noise, the applicability of this method is poor. The above disadvantages limit the application of this method in practice.
[0004] In order to overcome the shortcomings of the above detection methods, the present invention discloses a method for detecting bifurcation faults of a motor pulse energy conversion system. The Dynamic Time Warping (DTW) method adopted by the present invention can better process irregular time series data and identify pattern changes. The DTW method not only requires relatively low computational cost, but also has good anti-noise interference ability, especially suitable for complex systems without a clear model. In addition, the DTW method has great advantages in feature extraction and early identification of bifurcation faults.
[0005] In order to reduce data acquisition, transmission and processing costs, etc., this invention patent determines the data acquisition volume through an adaptive sampling criterion, introduces the CUSUM technology to accumulate weak fault information in the adaptive data to identify faults, and can provide more sensitive and accurate detection results, so as to be able to identify the bifurcation-type faults of the pulse energy conversion system in real time and reliably. Summary of the Invention
[0006] Object of the Invention: Aiming at the problems pointed out in the background technology, this invention discloses a method for detecting bifurcation-type faults in a motor pulse energy conversion system. First, the fault characteristics are calculated by using the DTW method; then the CUSUM technology is used to accumulate weak fault information, which can significantly improve the sensitivity and reliability of bifurcation-type fault detection; finally, the sampling data volume of the pulse energy conversion system is adaptively determined, which can more efficiently reduce the costs of data acquisition, transmission, processing and storage.
[0007] Technical Solution: A method for detecting bifurcation-type faults in a motor pulse energy conversion system according to this invention specifically includes the following steps:
[0008] Step (1): Determine the installation position and detection parameters of the current sensor;
[0009] Step (2): Obtain N groups of periodically synchronized single-point sampling data X;
[0010] Step (3): Calculate the DTW distances between adjacent two groups of the N groups of periodically synchronized single-point sampling data X in the N groups of periodically synchronized single-point sampling data X respectively to obtain DTW distance data D;
[0011] Step (4): Use the CUSUM technology to accumulate the weak fault information of data D to obtain cumulative sum data C;
[0012] Step (5): Judge whether a bifurcation-type fault occurs and realize adaptive data acquisition.
[0013] Further, in step (1), the current sensor is installed at the motor excitation winding to collect the motor current signal; the detection parameters include: the initial number N of synchronously sampled data points in each PWM cycle 0 , the minimum number N min , the maximum number N max , the number of PWM cycles L, the reference value M, and the fault threshold T; the initial number N of synchronously sampled data points in each PWM cycle 0 is determined according to the health state of the detected pulse energy conversion system; when the detected pulse energy conversion system is a brand-new device, the initial number N of synchronously sampled data points in each PWM cycle of the pulse energy conversion system 0 is set to the minimum number N min ; the minimum number N minshall be greater than or equal to 2; when the health status of the detected pulse energy conversion system is unknown, the initial number N of synchronous sampling data points within each PWM period of the pulse energy conversion system 0 is set to
[0014] where round() is a rounding function and the result is an integer; the reference value M is obtained from the statistical data when the pulse energy conversion system is healthy; the fault threshold T is obtained based on expert experience.
[0015] Furthermore, the periodic synchronous single-point sampling in step (2) means synchronously obtaining a single sampling data within each PWM period; at different moments within the PWM period, the periodic synchronous single-point sampling data is synchronously collected respectively, and then multiple groups of periodic synchronous single-point sampling data can be obtained; according to the number N of synchronous sampling data points within each PWM period of the pulse energy conversion system, N groups of periodic synchronous single-point sampling data X = [X 1 , X 2 , X 3 , …, X i , …, X N T ,
[0016]
[0017] where the i-th group of periodic synchronous single-point sampling data is X i = [x i , x i+N , x i+2N , …, x i+(L-1)N .
[0018] Furthermore, in step (3), the DTW algorithm is used to calculate the DTW distance between adjacent two groups of periodic synchronous single-point sampling data respectively, and the DTW data D is obtained:
[0019] D = [d 1 d 2 d 3 … d i … d N-1 T ,
[0020] where d i represents the DTW distance between the i-th group and the (i + 1)-th group of periodic synchronous single-point sampling data X i and X i+1 .
[0021] Furthermore, the implementation process of step (4) is as follows. In the CUSUM method, the reference value M is set as the expected value. For each data point in the data D, the difference between it and the reference value M is calculated to obtain the data Q.
[0022] Q = [q(1), q(2), q(3), …, q(i), …, q(N - 1)] T ,
[0023] where q(i) = d i - M;
[0024] Calculate the cumulative sum of data Q to obtain data C,
[0025] C = [c(1), c(2), c(3), …, c(i), …, c(N - 1)],
[0026] where,
[0027] Furthermore, the implementation process of step (5) is as follows. Take the final cumulative value c(N - 1) of the cumulative sum data C as the fault index and compare it with the fault threshold T; if the fault index c(N - 1) is less than T, it is determined that there is no fault, and the number N of synchronous sampling data points in each PWM period of the next detection time window is adaptively adjusted to where round() is the rounding function, and return to step (2) to continue the fault detection; if the fault index c(N - 1) is greater than or equal to the threshold T, it is considered that a fault has occurred and an alarm is issued.
[0028] Beneficial effects:
[0029] (1) A method for detecting bifurcation - type faults in a motor pulse energy conversion system disclosed by the present invention overcomes the shortcoming of low reliability of existing methods and can effectively process strong background noise signals;
[0030] (2) A method for detecting bifurcation - type faults in a motor pulse energy conversion system disclosed by the present invention adaptively determines the number of synchronous sampling data points in each PWM period, reduces the data acquisition volume, and thus reduces the costs of data acquisition, processing, transmission, and storage;
[0031] (3) A method for detecting bifurcation - type faults in a motor pulse energy conversion system disclosed by the present invention calculates the bifurcation - type fault characteristics using the DTW algorithm, has a relatively low computational complexity, and can perform real - time fault detection;
[0032] (4) A method for detecting bifurcation - type faults in a motor pulse energy conversion system disclosed by the present invention uses the CUSUM technique to cumulatively and adaptively obtain weak fault information in the data, and can improve the sensitivity and reliability of early fault detection; Description of the Drawings
[0033] Figure 1 It is a flow chart of a method for detecting bifurcation - type faults in a motor pulse energy conversion system disclosed by the present invention;
[0034] Figure 2 The current data and cycle synchronous sampling data when there is no bifurcation fault in the pulse energy conversion system;
[0035] Figure 3 The current data and cycle synchronous sampling data when there is a bifurcation fault in the pulse energy conversion system;
[0036] Figure 4 The detection result of the method of the present invention when there is no bifurcation fault in the pulse energy conversion system;
[0037] Figure 5 The number of synchronous sampling data points within each PWM cycle when there is no bifurcation fault in the pulse energy conversion system;
[0038] Figure 6 The detection result of the method of the present invention when there is a bifurcation fault in the pulse energy conversion system;
[0039] Figure 7 The number of synchronous sampling data points within each PWM cycle when there is a bifurcation fault in the pulse energy conversion system. Specific embodiments
[0040] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings.
[0041] The calculation process of a method for detecting bifurcation faults in a motor pulse energy conversion system disclosed by the present invention is as Figure 1 shown, and specifically includes the following steps:
[0042] Step (1): Determine the installation position and detection parameters of the current sensor;
[0043] The current sensor is installed at the motor excitation winding for collecting the motor current signal; the detection parameters include the initial number N 0 of synchronous sampling data within each PWM cycle, the minimum number N min , the maximum number N max , the number of PWM cycles L, the reference value M, and the fault threshold T; in this example, it is set that the minimum number N min of synchronous sampling data points within each PWM cycle is 45, the maximum number N max is 55, the initial number N 0 is 50, the number of cycles L = 10, the reference value M = 212, and the fault threshold T = 30; assuming that the health state of the measured pulse energy converter is unknown, then the initial number N 0 of synchronous sampling data points within each PWM cycle is set to where round() is a rounding function and the resulting value is an integer.
[0044] Step (2): Obtain N groups of cycle-synchronized single-point sampling data X;
[0045] According to the amount of synchronized sampling data N = 50 within each PWM cycle, the obtained N groups of cycle-synchronized single-point sampling data X = [X 1 , X 2 , X 3 , …, X i , …, X 50 T ,
[0046]
[0047] where the i-th group of cycle-synchronized single-point sampling data is X i = [x i , x i+50 , x i+100 , …, x i+450 .
[0048] Step (3): Calculate the DTW distances between adjacent groups of cycle-synchronized single-point sampling data in the N groups of cycle-synchronized single-point sampling data X respectively, to obtain DTW distance data D;
[0049] Use the DTW algorithm to calculate the DTW distances between adjacent groups of cycle-synchronized single-point sampling data in the N groups of cycle-synchronized single-point sampling data X respectively, to obtain DTW distance data D:
[0050] D = [d 1 d 2 d 3 … d i … d 49 ,
[0051] where d i represents the DTW distance between the i-th and the (i + 1)-th groups of cycle-synchronized single-point sampling data X i and X i+1 .
[0052] Step (4): Use the CUSUM technique to accumulate the weak fault information of data D, to obtain cumulative sum data C;
[0053] Set the reference value M as the expected value in the CUSUM method. For each data point in data D, calculate the difference between it and the reference value M to obtain data Q,
[0054] Q = [q(1) q(2) q(3) … q(i) … q(49)] T ,
[0055] where q(i) = z(i) - 212; calculate the cumulative sum of data Q to obtain data C,
[0056] C = [c(1) c(2) c(3) … c(i) … c(49)],
[0057] where,
[0058] Step (5): Determine whether a bifurcation fault occurs and implement adaptive data acquisition.
[0059] Take the final cumulative value c(N - 1) of the cumulative sum data C as the fault index. Compare the fault index c(N - 1) with the fault threshold T = 30. If the fault index c(N - 1) is less than T, it is determined that there is no fault, and the number of synchronous sampling data points N within each PWM period of the next detection time window is adaptively adjusted to where round() is a rounding function, and the result is an integer. Then return to step (2) to continue the bifurcation fault detection of the pulse energy conversion system. If the fault index c(N - 1) is greater than the threshold T, it is considered that a bifurcation fault has occurred, and an alarm is issued.
[0060] Figure 2 Shows the current data and synchronous sampling data when there is no bifurcation fault in the pulse energy conversion system. Figure 3 Shows the current data and synchronous sampling data when there is a bifurcation fault in the pulse energy conversion system. It is impossible to tell from these two figures whether a fault has occurred in the pulse energy conversion system.
[0061] Figure 4 Shows the detection results of the method of the present invention when there is no bifurcation fault in the pulse energy conversion system. It can be seen from the figure that the fault indices are all less than the threshold T, so it is determined that the pulse energy conversion system is in a fault - free state. Then the number of synchronous sampling data points N within each PWM period is adaptively adjusted to And return to step (2) to continue the bifurcation fault detection.
[0062] Figure 5 Shows the number of synchronous sampling data points N within each PWM period when there is no bifurcation fault in the pulse energy conversion system. In each fault detection, the number of synchronous sampling data points N within each PWM period changes with the fault index. Since the pulse energy conversion system is in a state without bifurcation faults, in each detection process, the number of synchronous sampling data points N within each PWM period is less than or equal to the maximum value of 55, reducing the data acquisition volume, and thus reducing the data transmission, processing, and storage costs.
[0063] Figure 6The detection results of the method of the present invention when the motor pulse energy conversion system has a bifurcation-type fault are shown. It can be seen from the figure that at the 10th detection, if the fault index c(N - 1) is greater than or equal to T, it is considered that an early fault has occurred in the pulse energy conversion system. At this time, an alarm is issued to notify the relevant maintenance personnel to reasonably arrange the maintenance work of the system.
[0064] Figure 7 N is the number of synchronous sampling data points in each PWM period when the pulse energy conversion system has a bifurcation-type fault. At the 10th fault detection, the fault has been detected, and thus the fault detection work is ended, and the number N of synchronous sampling data points in each PWM period of the next detection time window is no longer adaptively calculated.
[0065] It can be concluded that a method for detecting bifurcation-type faults in a motor pulse energy conversion system disclosed by the present invention first uses the DTW algorithm to find the fault information of the pulse energy conversion system, overcoming the shortcomings of existing methods. Then, it adaptively adjusts the number N of synchronous sampling data points in each PWM period according to the health status, increases or decreases the data acquisition amount in real time, optimizes the sampling data amount, and reduces the costs of data acquisition, processing, transmission, and storage. Finally, the CUSUM technology is used to accumulate the weak fault information in the data, improving the sensitivity and reliability of fault detection. In addition, the method for detecting bifurcation-type faults in a motor pulse energy conversion system disclosed by the present invention has a low calculation cost and can be used for real-time fault detection.
[0066] The above embodiments are only for explaining the technical concept and features of the present invention, and their purpose is to enable those who are familiar with this technology to understand the content of the present invention and implement it accordingly, and it cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit and essence of the present invention should be covered within the protection scope of the present invention.
Claims
1. A method for detecting bifurcation faults in a motor pulse energy conversion system, characterized in that: include: Step (1): Determine the installation position and detection parameters of the current sensor; Step (2): Obtain N groups of periodic synchronous single-point sampling data X; Step (3): respectively calculating the DTW distances of two adjacent groups of periodic synchronous single-point sampling data in the N groups of periodic synchronous single-point sampling data X to obtain DTW distance data D; Step (4): Use CUSUM technology to accumulate the weak fault information of data D to obtain cumulative sum data C; Step (5): Determine whether a bifurcation fault occurs and implement adaptive data acquisition.
2. A method for detecting bifurcation type faults in a motor pulse energy conversion system according to claim 1, characterized in that: The current sensor in step (1) is installed at the motor excitation winding to collect the motor current signal.
3. A method for detecting bifurcation type faults in a motor pulse energy conversion system according to claim 1, characterized in that: The detection parameters set in step (1) include: the initial number N0 of synchronous sampling data in each PWM cycle, the minimum number N min , the maximum number N max , PWM cycle number L, reference value M and fault threshold T; the initial number N0 of synchronously sampled data points in each PWM cycle is determined according to the health status of the pulse energy conversion system being tested; when the pulse energy conversion system being tested is a new device, the initial number N0 of synchronously sampled data points in each PWM cycle of the pulse energy conversion system is set to the minimum number N min ; The minimum number N min It must be greater than or equal to 2; when the health status of the detected pulse energy converter is unknown, the initial number N0 of synchronous sampling data points in each PWM cycle is set to Among them, round() is a rounding function, and the result is an integer; the reference value M is obtained by the statistical data when the pulse energy conversion system is healthy; and the fault threshold T is obtained based on expert experience.
4. A method for detecting bifurcation type faults in a motor pulse energy conversion system according to claim 1, characterized in that: The periodic synchronous single-point sampling data in step (2) refers to the synchronous acquisition of single sampling data in each PWM cycle; the N groups of periodic synchronous single-point sampling data are represented as X = [X1, X2, X3, ..., X i ,…,X N ] T , Among them, the i-th group of periodic synchronous single-point sampling data is X i =[x i ,x i+N ,x i+2N ,…,x i+(L-1)N ].
5. A method for detecting bifurcation faults in a motor pulse energy conversion system according to claim 1, characterized in that: The step (3) uses the DTW algorithm to respectively calculate the DTW distances of two adjacent groups of periodic synchronous single-point sampling data to obtain DTW distance data D=[d1 d2 d3 ... d i …d N-1 ] T , where d i Indicates the periodic synchronous single-point sampling data X of the i-th group and the i+1-th group i and X i+1 DTW distance.
6. A method for detecting bifurcation faults in a motor pulse energy conversion system according to claim 1, characterized in that: The implementation process of step (4) is as follows: in the CUSUM method, the reference value M is set as the expected value, and for each data point in the data D, the difference between it and the reference value M is calculated to obtain the data Q=[q(1),q(2),q(3),…,q(i),…q(N-1)] T , Where q(i) = d i -M; calculate the cumulative sum of data Q to obtain data C, C=[c(1)c(2)c(3)…c(i)…c(N-1)] in, 7. A method for detecting bifurcation faults in a motor pulse energy conversion system according to claim 1, characterized in that: The implementation process of step (5) is as follows: the final cumulative value c(N-1) of the cumulative sum data C is used as a fault indicator to compare with the fault threshold T; if the fault indicator c(N-1) is less than T, it is determined that there is no fault, and the number of synchronous sampling data points N in each PWM cycle of the next detection time window is adaptively adjusted to And return to step (2) to continue fault detection; if the fault indicator c(N-1) is greater than or equal to the threshold T, it is considered that a fault has occurred and an alarm is issued.