Improved minimum noise amplitude deconvolution bearing fault diagnosis method

Abstract

The invention relates to an improved minimum noise amplitude deconvolution bearing fault diagnosis method. The method comprises the following steps: acquiring an original signal of a fault bearing; estimating a fault period by using self-correlation analysis; setting an optimization range, and initializing parameters of an improved sparrow search algorithm, including the maximum number of iterations m and the number of search agents n; enhancing the original signal by using a minimum noise amplitude deconvolution technology; the fitness value of the enhanced signal is calculated, and the optimal parameter is updated during each iteration; checking whether a termination condition is satisfied; if so, ending iteration; otherwise, continuing iteration; a termination condition is realized, and an optimal parameter is obtained; enhancing the original signal by using a minimum noise amplitude deconvolution technology with optimal parameters; fault features are detected through envelope analysis, and final fault diagnosis is carried out. According to the method, the fault period is estimated by using self-correlation analysis instead of depending on priori knowledge; the method has the capability of adaptively selecting optimal parameters according to working condition changes, and blind selection of the parameters is avoided.

Description

Technical Field

[0001] The invention belongs to the technical field of mechanical fault diagnosis, and in particular relates to an improved minimum noise amplitude deconvolution bearing fault diagnosis method. Background Art

[0002] At present, rotating machinery has been widely used in various fields of industrial applications such as manufacturing, energy, transportation, and petrochemicals. Due to long-term operation under heavy loads, high temperatures, or other harsh environments, key components such as gears, bearings, and rotors are easily damaged. These unexpected failures may lead to unplanned downtime and even catastrophic casualties. Therefore, it is of great significance to identify the fault type and assess the fault level as early as possible. Generally speaking, the process of mechanical fault diagnosis is first to measure diagnostic information through various sensors, such as vibration sensors, current sensors, acoustic emission sensors, voltage sensors, and encoders. However, the information fault features in the sensor output are usually masked by severe noise. Therefore, how to use advanced signal processing technology to extract fault features from complex raw signals has become a key step.

[0003] Compared with the classic blind deconvolution algorithm, the minimum noise amplitude deconvolution algorithm can significantly enhance the time domain characteristics, spectrum characteristics and envelope spectrum characteristics of the fault signal. Therefore, the minimum noise amplitude deconvolution technology can effectively enhance the fault characteristics of multi-resonance bearing fault signals, and can effectively enhance the fault characteristics of low signal-to-noise ratio bearing fault signals. The existing minimum noise amplitude deconvolution technology requires the optimal value of the pre-parameter to be set to operate reliably. Therefore, the selection of these parameters largely determines the signal enhancement performance of the deconvolution technology. If the parameter value is not selected appropriately, the final signal enhancement effect will be greatly reduced. In summary, the embodiment of the optimal performance of the minimum noise amplitude deconvolution technology needs to rely on the selection of the pre-parameters, and the selection of the pre-parameters depends to a large extent on the selection of manual experience. Such a selection process is relatively blind and inefficient. Based on this, it is urgent to develop an improved minimum noise amplitude deconvolution bearing fault diagnosis method to effectively solve the above problems. Summary of the invention

[0004] The purpose of the present invention is to provide an improved minimum noise amplitude deconvolution bearing fault diagnosis method to solve the problem that the selection of pre-parameters in the minimum noise amplitude deconvolution technology is relatively blind and inefficient.

[0005] The objective of the present invention is achieved through the following technical solutions:

[0006] An improved minimum noise amplitude deconvolution bearing fault diagnosis method comprises the following steps:

[0007] A. Collect the original signal of the faulty bearing;

[0008] B. Estimate the failure period using autocorrelation analysis;

[0009] C. Set the optimization range and initialize the parameters of the improved sparrow search algorithm, including the maximum number of iterations m and the number of search agents n;

[0010] D. Use minimum noise amplitude deconvolution technique to enhance the original signal;

[0011] E. Calculate the fitness value of the enhanced signal and update the optimal parameters at each iteration;

[0012] F. Check whether the termination condition is met; if so, end the iteration; otherwise, continue the iteration;

[0013] G. Achieve the termination conditions and obtain the optimal parameters;

[0014] H. Use the minimum noise amplitude deconvolution technique with optimal parameters to enhance the original signal;

[0015] I. Use envelope analysis to detect fault characteristics and perform final fault diagnosis.

[0016] Furthermore, in step B, the Hilbert transform envelope analysis method is used to determine the best candidate point of the signal fault period by locating the peak position in the envelope signal autocorrelation function to estimate the fault period.

[0017] Furthermore, in step B, the steps of estimating the failure period are as follows:

[0018] B1. Obtain the analytical signal of the input signal through formula (1);

[0019]

[0020] B2. Calculate the autocorrelation of the analytical signal using formula (2), and select the maximum value as the number of iterations;

[0021]

[0022]

[0023] Where x(t) is the analytical signal obtained by Hilbert transform and τ is the lag.

[0024] Furthermore, the candidate range excludes the portion from 0 to the first point where the autocorrelation equals zero.

[0025] Furthermore, in step C, the improved sparrow search algorithm is optimized by combining the three parameter values ​​of sine, cosine and Cauchy mutation.

[0026] Furthermore, the search factor in the improved sparrow search algorithm is defined as:

[0027]

[0028] Among them, a is a constant, t is the number of iterations, Iter max is the maximum number of iterations;

[0029] The new mathematical equation describing the discoverer is as follows:

[0030]

[0031] Introducing the Cauchy variational strategy into the original equation, we get another follower equation:

[0032]

[0033] Furthermore, the envelope spectrum peak factor is used as the objective function, and this index is used as the objective function of the optimization parameter. The index can be defined as:

[0034]

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The improved minimum noise amplitude deconvolution bearing fault diagnosis method of the present invention proposes an adaptive minimum noise amplitude deconvolution technology, which combines an improved sparrow search algorithm and autocorrelation analysis to solve the specific problems faced; this adaptive minimum noise amplitude deconvolution technology can use autocorrelation analysis to estimate the fault cycle instead of relying on prior knowledge, thereby expanding its application range; in addition, the technology also has the ability to adaptively select the optimal parameters according to changes in operating conditions, avoiding blind selection of parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 The original signal and the envelope spectrum of the original signal;

[0039] Figure 2 The optimization process of the improved sparrow search algorithm;

[0040] Figure 3 Improved minimum noise amplitude deconvolution of the enhanced signal;

[0041] Figure 4 Flowchart of the steps of the improved minimum noise amplitude deconvolution bearing fault diagnosis method. DETAILED DESCRIPTION

[0042] The present invention will be further described below in conjunction with embodiments:

[0043] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0044] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0045] The detection of cyclic pulse components in vibration signals is an important step in bearing fault extraction. Deconvolution method has been proven to be an effective technical tool to highlight the cyclic pulse components caused by bearing faults. Among them, the minimum noise amplitude deconvolution technique can effectively enhance the fault periodic pulse signal from the severely disturbed bearing fault signal. However, there are some shortcomings in the original minimum noise amplitude deconvolution technique, which mainly comes from the strict requirements on multiple input parameters. These pre-parameters largely determine the performance of the deconvolution technique. If the parameter values ​​are not selected appropriately, the final signal enhancement effect will be greatly reduced.

[0046] In order to solve the above problems, the present invention proposes an adaptive minimum noise amplitude deconvolution technology, which uses the autocorrelation analysis of the envelope signal to estimate a priori parameter of the method, namely the key parameter of the fault cycle, instead of relying on a given fault cycle. The method uses an improved sparrow search algorithm to optimize other priori parameters to determine the values ​​of the three key parameters of the optimal filter length, number of iterations and noise ratio. Compared with the original minimum noise amplitude deconvolution technology, the adaptive minimum noise amplitude deconvolution technology can select the optimal parameters according to the environmental adaptive parameters under different operating conditions, so that the method can achieve the best performance.

[0047] The improved minimum noise amplitude deconvolution bearing fault diagnosis method of the present invention comprises the following steps:

[0048] 1. Adaptive minimum noise amplitude deconvolution

[0049] 1.1. Estimation of failure cycle

[0050] The Hilbert transform envelope analysis method transfers the frequency analysis from the resonant frequency range to a lower range containing the fault frequency, which can effectively simplify the signal. According to the definition of autocorrelation, the best candidate point of the signal fault period can be determined by locating the peak position within the envelope signal autocorrelation function. The steps for estimating the fault period are as follows:

[0051]

[0052]

[0053]

[0054] Where x(t) is the analytical signal obtained by Hilbert transform, and τ is the lag. First, the analytical signal of the input signal is obtained by formula (1); then, the autocorrelation of the analytical signal is calculated using formula (2), and the maximum value is selected as the number of iterations. It should be noted that the candidate range excludes the part from 0 to the first point where the autocorrelation is equal to zero.

[0055] 2. Parameter selection

[0056] Accurately and reasonably selecting multiple input parameters such as filter size, number of iterations, noise ratio, etc. can ensure the performance of fault pulse enhancement. However, it is a huge challenge to select three optimal parameter values ​​simultaneously based on empirical knowledge alone. Therefore, the present invention studies a commonly used parameter selection optimization framework. The optimization framework adopts an improved sparrow search algorithm, combined with sine cosine and Cauchy mutation three parameter values ​​for optimization. In the present invention, sine cosine and Cauchy mutation are also a search optimization strategy, which is combined with the sparrow search algorithm. Corresponding to the following formulas (5) and (6) respectively.

[0057] 2.1. Improved sparrow search algorithm

[0058] During the sparrow hunting process, the food source and location may change. When the finder encounters the most ideal local food source, it will attract a large number of followers to gather there. This leads to a reduction in the positional diversity of the finder and the entire group, and ultimately, leads to being trapped in a local extreme situation. Therefore, the present invention incorporates the sine-cosine algorithm into the original sparrow search algorithm and uses the fluctuation pattern of the sine and cosine model to identify the location of the explorer. This method effectively retains the diversity of individual explorers, thereby enhancing the global search capability and avoiding falling into the local optimum.

[0059] The search factor is defined as:

[0060]

[0061] Among them, a is a constant, t is the number of iterations, Itermax is the maximum number of iterations.

[0062] The new mathematical equation describing the discoverer is as follows:

[0063]

[0064] In order to avoid the problem of local optimality, the search algorithm further introduces the Cauchy variational strategy into the original equation and obtains another follower equation:

[0065]

[0066] 2.2 Objective Function

[0067] The establishment of the objective function is also an important step in parameter selection. Considering the impact characteristics of the bearing fault signal, the envelope spectrum peak factor is used as the objective function. Due to its frequency domain analysis characteristics, the indicator is also immune to random pulse noise. Using this indicator as the objective function of the optimization parameter can enhance the periodic pulse component in the signal and improve the signal-to-noise ratio of the signal. The higher the value of this indicator, the stronger the signal-to-noise ratio of the signal and the more prominent the periodic pulse component. This indicator can be defined as:

[0068]

[0069] The following is the process of the adaptive minimum noise amplitude deconvolution technology.

[0070] Step 1, collecting the original signal of the faulty bearing;

[0071] Step 2: Estimate the failure period using autocorrelation analysis;

[0072] Step 3: Set the optimization range and initialize the parameters of the improved sparrow search algorithm, including the maximum number of iterations m and the number of search agents n; these are the pre-parameters of the sparrow search algorithm, and set m=10 and n=30.

[0073] Step 4: Use minimum noise amplitude deconvolution technique to enhance the original signal;

[0074] Step 5: Calculate the fitness value of the enhanced signal and update the optimal parameters in each iteration;

[0075] Step 6: Check whether the termination condition is met. If yes, the iteration ends. Otherwise, the iteration continues;

[0076] Step 7: Implement the termination condition and obtain the optimal parameters;

[0077] Step 8: Use the minimum noise amplitude deconvolution technique with optimal parameters to enhance the original signal;

[0078] Step 9: Use envelope analysis to detect fault characteristics and perform final fault diagnosis.

[0079] like Figure 1-Figure 3 As shown, the present invention is directed to early bearing fault feature extraction under the proposed bearing fault diagnosis scheme.

[0080] It can be seen that in the case of strong noise, the weak early fault pulse has been masked by the noise, and its envelope spectrum cannot directly extract the fault characteristic frequency.

[0081] Through optimization, the optimized parameters are obtained: filter length L=20, iteration number N=185 and noise ratio ρ=0.2.

[0082] Figure 3 The enhanced signal is obtained by deconvolving the minimum noise amplitude with the optimal value. It can be seen that compared with the original signal, the fault pulse sequence of the enhanced signal is highlighted, and the envelope spectrum of the enhanced signal can clearly identify the fault characteristic spectrum and its multiples, which proves the superiority of this method.

[0083] In view of the deficiency of needing to calculate the priori fault period in advance, the present invention proposes an envelope signal autocorrelation analysis method to automatically estimate the fault period, and in view of the problem of parameter selection, proposes an adaptive minimum noise amplitude deconvolution technology based on an improved sparrow search algorithm; the present invention can calculate the fault period through the envelope signal autocorrelation analysis method, thereby avoiding the deficiency of manually calculating the fault period and expanding the application scope of the method; the improved sparrow search algorithm can adaptively select the optimal parameters, thereby avoiding the blind and inefficient manual selection method and improving the efficiency and application scope of the algorithm.

[0084] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. An improved minimum noise amplitude deconvolution bearing fault diagnosis method, characterized in that: The following steps are involved: A. Collect the original signal of the faulty bearing; B. Estimate the failure period using autocorrelation analysis; C. Set the optimization range and initialize the parameters of the improved sparrow search algorithm, including the maximum number of iterations m and the number of search agents n; D. Use minimum noise amplitude deconvolution technique to enhance the original signal; E. Calculate the fitness value of the enhanced signal and update the optimal parameters at each iteration; F. Check whether the termination conditions are met; If yes, then end the iteration; otherwise, continue the iteration; G. Achieve the termination conditions and obtain the optimal parameters; H. Use the minimum noise amplitude deconvolution technique with optimal parameters to enhance the original signal; I. Use envelope analysis to detect fault characteristics and perform final fault diagnosis.

2. The improved minimum noise amplitude deconvolution bearing fault diagnosis method according to claim 1, characterized in that: Step B, using the Hilbert transform envelope analysis method, by locating the peak position in the envelope signal autocorrelation function, determine the best candidate point of the signal fault period and estimate the fault period.

3. The improved minimum noise amplitude deconvolution bearing fault diagnosis method according to claim 2 is characterized in that: Step B, the steps for estimating the failure period are as follows: B1. Obtain the analytical signal of the input signal through formula (1); B2. Calculate the autocorrelation of the analytical signal using formula (2), and select the maximum value as the number of iterations; Where x(t) is the analytical signal obtained by Hilbert transform and τ is the lag.

4. The improved minimum noise amplitude deconvolution bearing fault diagnosis method according to claim 3 is characterized in that: The candidate range excludes the portion from 0 to the first point where the autocorrelation equals zero.

5. The improved minimum noise amplitude deconvolution bearing fault diagnosis method according to claim 1, characterized in that: Step C, the improved sparrow search algorithm is combined with the three parameter values ​​of sine, cosine and Cauchy mutation for optimization.

6. The improved minimum noise amplitude deconvolution bearing fault diagnosis method according to claim 5, characterized in that: The search factor in the improved sparrow search algorithm is defined as: Among them, a is a constant, t is the number of iterations, Iter max is the maximum number of iterations; The new mathematical equation describing the discoverer is as follows: Introducing the Cauchy variational strategy into the original equation, we get another follower equation:

7. The improved minimum noise amplitude deconvolution bearing fault diagnosis method according to claim 6, characterized in that: The envelope spectrum peak factor is used as the objective function, and the index is used as the objective function of the optimization parameter. The index is defined as:

Images