Automobile transmission shaft bearing composite fault feature extraction method based on RVME and SWO

Abstract

The invention discloses an automobile transmission shaft bearing composite fault feature extraction method based on RVME and SWO. The method comprises the following steps: step 1, determining optimal values of a penalty factor and a center frequency by using SWO, selecting a modal component by using an effective weight kurtosis index to perform signal reconstruction, and reducing interference of strong background noise; 2, introducing two-dimensional Chebyshev-logic-infinite folding mapping to increase diversity and randomness of a population, preventing SWO from falling into local optimum, shortening operation time, and increasing a fault feature coefficient; and step 3, dividing the population into a plurality of subgroups according to rapid non-dominated sorting, and accelerating the convergence speed of the SWO by adopting a self-adaptive grouping strategy.

Description

Technical Field

[0001] The present invention relates to a composite fault feature extraction method for automobile transmission shaft bearings, and in particular to a composite fault feature extraction method for automobile transmission shaft bearings based on recursive variational mode extraction (RVME) and spiderwasp optimizer (SWO). Background Art

[0002] The automobile transmission shaft is an important part of the automobile, playing the role of motion conversion and load transmission. However, the automobile transmission shaft bearings are prone to various faults due to long-term operation in harsh working environments and complex working conditions such as impact loads. The impact vibration caused by these faults will seriously affect the driving comfort of the car. If these faults are not discovered as early as possible and timely measures are taken, it will affect the safe maintenance of the car and even cause catastrophic accidents with car damage and death. Therefore, the fault diagnosis of automobile transmission shaft bearings is of great significance to the safe operation of the car. The spatial structure of the automobile transmission system is narrow, and the bearing vibration signal is often submerged in the vibration signals of other parts, resulting in the early weak fault information of the bearing being interfered by strong background noise. When multiple faults occur in the bearing, multiple fault features overlap and couple with each other, making it more difficult to identify the fault type. Therefore, the key to fault diagnosis of automobile transmission shaft bearings is to develop signal processing methods to extract early weak fault features and separate composite fault features.

[0003] At present, the commonly used rolling bearing composite fault feature extraction methods include spectral kurtosis, singular spectrum analysis, variational mode decomposition, variational mode extraction (VME) and RVME. In order to solve the problem of adaptive determination of VME parameters, RVME takes the main frequency of the residual signal of the last iteration as the initial center frequency, and introduces the empirical formula to determine the penalty factor value. Compared with other methods, RVME has a stronger ability to separate bearing composite fault features. However, when calculating the initial center frequency, the determination of the rectangular window length lacks a theoretical basis. The penalty factor value at each iteration is calculated by the empirical formula. These two factors reduce the adaptability and decomposition accuracy of RVME in bearing composite fault feature extraction. Swarm intelligence algorithm is a biological heuristic method. Due to its strong adaptability, suitability for parallel computing, strong stability, low computational cost, fast solution speed and no need for gradient information, it has been successfully applied to combinatorial optimization, task scheduling, process control, engineering prediction, image processing and other fields. It provides a novel solution to the problems of bearing fault diagnosis and life prediction. For example, ant colony optimization algorithm, particle swarm optimization (PSO) algorithm, artificial bee colony algorithm, firefly algorithm, bat algorithm, cuckoo search algorithm, grey wolf optimization (GWO) algorithm, whale optimization algorithm (WOA), wild horse optimization algorithm, sparrow search algorithm (SSA) and chimpanzee optimization algorithm. Existing theoretical research and application show that swarm intelligence algorithm is an effective method to solve multi-objective and multi-parameter global optimization problems. Swarm intelligence algorithm is easy to implement, only involves basic mathematical operations, does not rely on the problem itself and the inspiration of prior knowledge, and has absolute advantages in solving complex optimization problems. The difference between automobile transmission shaft bearings and ordinary bearings is that the former is usually used in automobile transmission systems, subjected to high-speed rotation and large loads, and the inner and outer rings will fail at the same time. SWO has fast search speed, high solution accuracy, strong robustness and fast convergence speed. Therefore, SWO has good adaptability and flexibility and can handle complex optimization problems. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that it is difficult to extract composite fault features of automobile transmission shaft bearings under strong background noise interference, and proposes a composite fault feature extraction method for automobile transmission shaft bearings based on RVME and SWO.

[0005] The composite fault feature extraction method of automobile transmission shaft bearing based on RVME and SWO is characterized in that the method comprises the following steps:

[0006] Step 1: Use SWO to determine the optimal values ​​of the penalty factor and center frequency, and use the effective weight kurtosis index to select the modal components for signal reconstruction to reduce the interference of strong background noise;

[0007] Step 2: Introduce a two-dimensional Chebyshev-Logistic-Infinite collapse map (2D-CLICM) to increase the diversity and randomness of the population, prevent SWO from falling into local optimality, shorten the running time, and increase the fault feature coefficient (FFC);

[0008] Step 3: Divide the population into multiple subgroups according to the fast non-dominated sorting, and use the adaptive grouping strategy to speed up the convergence of SWO.

[0009] Compared with other bearing composite fault feature extraction methods, the automobile transmission shaft bearing composite fault feature extraction method based on RVME and SWO of the present invention has the following beneficial effects:

[0010] 1. Use SWO to determine the optimal values ​​of penalty factor and center frequency, use effective weight kurtosis index to select modal components for signal reconstruction, reduce the interference of strong background noise, and solve the problem that it is difficult to extract the composite fault characteristics of automobile transmission shaft bearings. Therefore, this method can effectively suppress the interference noise in the signal;

[0011] 2.2D-CLICM makes the population distribution more uniform, more random, with a higher Lyapunov index and better unpredictability. The introduction of 2D-CLICM in this method can effectively increase the diversity and randomness of the population, prevent SWO from falling into local optimality, shorten the running time, and increase FFC;

[0012] 3. Adopt adaptive grouping strategy to speed up the convergence of SWO. Divide the population into multiple subgroups according to fast non-dominated sorting, search for local optimal values ​​in each subgroup, and find the global optimal value in each local optimal value. This method takes into account the convergence speed and optimization accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a framework diagram of the composite fault feature extraction method of automobile transmission shaft bearing based on RVME and SWO of the present invention;

[0014] Figure 2 It is the experimental signal of composite fault of automobile transmission shaft bearing;

[0015] Figure 3 is the reconstructed signal of the experimental signal;

[0016] Figure 4 Convergence curves of the six methods. DETAILED DESCRIPTION

[0017] Specific implementation method: Combination Figure 1 The present embodiment is described, a method for extracting composite fault features of automobile transmission shaft bearings based on RVME and SWO, characterized in that the method comprises the following steps:

[0018] Step 1: Load the automobile transmission shaft bearing vibration signal x(t);

[0019] Step 2: Decompose x(t) according to RVME and determine the number of modal components K;

[0020] Step 3: Calculate the sample entropy of each IMF component and use the sample entropy to measure the fitness of the IMF component;

[0021] Step 4: Set the penalty factor α and center frequency f kc Initialize the population size N, spatial dimension dim, crossover rate CR, compromise rate TR and maximum number of iterations t of SWO max ;

[0022] Step 5: Initialize the female spider bee population using 2D-CLICM;

[0023] Step 6: Divide the population into multiple subgroups according to fast non-dominated sorting;

[0024] Step 7: Calculate the fitness of the female spider bees in each subgroup, and select the individual position with the smallest fitness in each subgroup;

[0025] Step 8: Generate a random number r6 and update the position of the female spider bee;

[0026] Step 9: Repeat Step 7 to Step 8 until the maximum number of iterations is reached. Obtain α and f in each subgroup kc The local optimum of

[0027] Step 10: Find α and f in each local optimal value kc The global optimal value of

[0028] Step 11: According to α and f kc The global optimal value of , RVME is used again to decompose x(t);

[0029] Step 12: Calculate the effective weight kurtosis of each IMF component respectively, and reconstruct the signal according to the effective weight kurtosis criterion;

[0030] Step 13: Calculate the envelope spectrum of the reconstructed signal and determine the composite fault characteristic frequency of the automobile transmission shaft bearing.

[0031] The following examples are used to verify the beneficial effects of the present invention.

[0032] Example:

[0033] The steps of the composite fault feature extraction method of automobile transmission shaft bearing based on RVME and SWO in this embodiment are as follows:

[0034] The model of the automobile transmission shaft bearing used in this embodiment is 6206. The outer ring fault (ORF) and inner ring fault (IRF) of the bearing are machined by electrospark. The fault width of the outer ring and inner ring of the bearing is 0.5mm. The speed of the drive motor is 30Hz, the sampling frequency is 10 240Hz, and the radial load is 600N. The fault characteristic frequencies of the outer ring and inner ring are f o =107.05Hz and f i =162.95Hz. The experimental signal of the composite fault of the automobile transmission shaft bearing is as follows Figure 2 The fault characteristic frequencies of the outer and inner races are submerged in the strong noise and cannot be detected by Figure 2 (b) Determine the bearing fault type. The composite fault signal is decomposed using the method proposed in this paper. The reconstructed signal of the experimental signal is as follows: Figure 3 shown. Figure 3 (a) There is an obvious periodic pulse component. Figure 3 (b) It can be seen that f o The first seven harmonics and the inner race fault characteristic frequency f i The first three harmonics of are extracted, and the interference noise is effectively suppressed. This is because the method improves the accuracy of SWO and accelerates its convergence speed by introducing 2D-CLICM and adaptive grouping strategy. SWO is used to determine the penalty factor α and the center frequency f kc The optimal value of is obtained, and the effective weight kurtosis index is used to select the modal components for experimental signal reconstruction, filter the noise components, and extract the composite fault characteristics of the automobile transmission shaft bearing.

[0035] In order to verify the optimization effect of the proposed method, the method is compared with traditional SWO and 2D-ICM-SWO, and FFC is used to evaluate the fault feature content in the reconstructed signal. The parameter settings are as described above, and the optimization effects of the three methods are shown in Table 1. Compared with traditional SWO, 2D-ICM-SWO and this method have shorter running time, faster convergence speed, larger FFC, and higher optimization accuracy. This is because the introduction of chaotic mapping can effectively increase the diversity and randomness of the population, thereby preventing SWO from falling into local optimality. Compared with 2D-ICM-SWO, this method has better optimization effect. The reason is that 2D-CLICM makes the population distribution more uniform, more random, higher Lyapunov index, and better unpredictability. Therefore, this method can solve the local optimal problem of SWO and improve the convergence speed and optimization accuracy of SWO.

[0036] Table 1 Optimization results of three methods

[0037]

[0038] In order to verify the convergence speed of the proposed method, this method is compared with PSO, GWO, WOA, SSA and Pelican optimization algorithm (POA). The parameter settings are as described above. The convergence curves of the six methods are shown in Figure 4 As shown. Figure 4 It can be seen that the optimal fitness of this method is the smallest. The reason is that this method uses 2D-CLICM to initialize the female spider bee population, making the initial population distribution more uniform and avoiding SWO from falling into the local optimum. In addition, the convergence speed of this method is fast. This is due to the fact that this method uses an adaptive grouping strategy to speed up the convergence speed of SWO. The population is divided into multiple subgroups according to the fast non-dominated sorting, and the local optimal value is searched in each subgroup, and the global optimal value is found in each local optimal value. In general, this method takes into account the convergence speed and optimization accuracy.

Claims

1. The composite fault feature extraction method of automobile transmission shaft bearing based on recursive variational mode extraction (RVME) and spider wasp optimizer (SWO) includes the following steps: Step 1: Load the automobile transmission shaft bearing vibration signal x(t); Step 2: Decompose x(t) according to RVME and determine the number of modal components K; Step 3: Calculate the sample entropy of each IMF component and use the sample entropy to measure the fitness of the IMF component; Step 4: Set the penalty factor α and center frequency f kc Initialize the population size N, spatial dimension dim, crossover rate CR, compromise rate TR and maximum number of iterations t of SWO max ; Step 5: Initialize the female spider bee population using a two-dimensional Chebyshev-Logistic-Infinite collapse map (2D-CLICM); Step 6: Divide the population into multiple subgroups according to fast non-dominated sorting; Step 7: Calculate the fitness of the female spider bees in each subgroup, and select the individual position with the smallest fitness in each subgroup; Step 8: Generate a random number r6 and update the position of the female spider bee; Step 9: Repeat Step 7 to Step 8 until the maximum number of iterations is reached. Obtain α and f in each subgroup kc The local optimum of Step 10: Find α and f in each local optimal value kc The global optimal value of Step 11: According to α and f kc The global optimal value of , RVME is used again to decompose x(t); Step 12: Calculate the effective weight kurtosis of each IMF component respectively, and reconstruct the signal according to the effective weight kurtosis criterion; Step 13: Calculate the envelope spectrum of the reconstructed signal and determine the composite fault characteristic frequency of the automobile transmission shaft bearing.

2. The method for extracting composite fault features of automobile transmission shaft bearings based on RVME and SWO according to claim 1 is characterized in that In the step 5, 2D-CLICM is used to initialize the female spider bee population, increase the diversity and randomness of the population, avoid SWO from falling into the local optimum, shorten the running time, and increase the fault characteristic coefficient.

3. The method for extracting composite fault features of automobile transmission shaft bearings based on RVME and SWO according to claim 1 is characterized in that In the step six, the population is divided into multiple subgroups according to the fast non-dominated sorting, and an adaptive grouping strategy is adopted to accelerate the convergence speed of SWO.

Images