Bearing vibration signal denoising method based on IBFO optimized VMD

Abstract

The invention relates to the technical field of industrial equipment state monitoring, in particular to a bearing vibration signal denoising method based on IBFO optimized VMD, which comprises the following steps: improving a rhodeus algorithm, initializing parameters of the improved rhodeus algorithm, and calculating a VMD value of the improved rhodeus algorithm; the improvement comprises the steps of optimizing an initial population by using chaotic mapping, introducing an adaptive weight which is exponentially attenuated along with the number of iterations in position updating, introducing a Cauchy mutation operator at an optimal solution, and generating a new candidate solution through disturbance operation; performing variational mode decomposition on the vibration signal, and calculating an initial fitness value in the decomposed signal through the minimum envelope entropy; the method comprises the following steps of: substituting initial parameters of an algorithm into an improved rhodeus algorithm, searching an optimal parameter and a penalty factor in a decomposed signal, performing variational mode decomposition on the decomposed signal based on the optimal parameter and the penalty factor to obtain an IMF component, screening out an effective component based on a Pearson's correlation coefficient, and constructing the effective component into a denoised signal. According to the invention, the fault feature separation capability is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment condition monitoring technology, specifically to a method for denoising bearing vibration signals based on IBFO-optimized VMD. Background Technology

[0002] In the field of industrial equipment condition monitoring technology, rotating machinery (such as motors, fans, pumps, and compressors) is a key piece of equipment in pillar industries such as petrochemicals, power, metallurgy, and manufacturing. Its operating status directly affects production safety, energy efficiency, and cost. Bearings, as one of the most widely used and easily damaged components in rotating machinery, are crucial for equipment fault early warning and diagnosis. Therefore, accurate analysis of bearing vibration signals and early fault diagnosis are key technologies for predictive maintenance and avoiding unplanned downtime, which is of great significance for promoting intelligent industrial transformation and ensuring the safe and reliable operation of equipment.

[0003] However, in existing technologies, the actual vibration signals acquired are often affected by environmental noise, electromagnetic interference, and mechanical resonance, which obscures fault characteristics and severely impacts diagnostic accuracy. Traditional denoising methods (such as wavelet transform and empirical mode decomposition) suffer from problems such as mode aliasing, endpoint effects, and strong parameter dependence, making it difficult to achieve high-precision signal separation. Although variational mode decomposition (VMD) can adaptively decompose signals, its core parameters (number of modes K and penalty factor α) need to be set manually based on experience, lacking optimization basis and easily leading to over-decomposition or under-decomposition. Existing intelligent optimization algorithms (such as BFO) are prone to getting trapped in local optima during parameter optimization, have slow convergence speeds, and uneven initial population distribution weakens global search capabilities. Summary of the Invention

[0004] To address the aforementioned technical problems of traditional VMD parameter optimization relying on experience and the BFO algorithm being prone to local optima, this invention provides a bearing vibration signal denoising method based on IBFO-optimized VMD. This invention primarily utilizes chaotic mapping to optimize the initial population distribution, an exponential decay adaptive weighting strategy, and a Cauchy mutation operator within the Bitterling algorithm to automatically optimize VMD parameters (modality number K, penalty factor α). By screening IMF components using the Pearson correlation coefficient to reconstruct effective signals, it significantly improves fault feature separation capabilities in high-noise environments (-3dB), exhibiting a faster convergence speed than traditional algorithms. This method can be widely applied to on-site inspection and health management of rotating machinery such as motors and fans, providing reliable support for predictive maintenance.

[0005] The technical means employed in this invention are as follows: A method for denoising bearing vibration signals based on IBFO-optimized VMD includes the following steps: Collect vibration signals from the bearing; The bitterling algorithm is improved by initializing the parameters of the improved bitterling algorithm to obtain the initial parameters of the algorithm. The improvement includes using chaotic mapping to optimize the initial population, introducing adaptive weights that decay exponentially with the number of iterations in the position update, introducing Cauchy mutation operator at the optimal solution and generating new candidate solutions through perturbation operation. The vibration signal is subjected to variational mode decomposition to obtain the decomposed signal, and the initial fitness value in the decomposed signal is calculated by the minimum envelope entropy. Using the improved bitterling algorithm, after substituting the initial parameters of the algorithm, the optimal parameters and penalty factors in the decomposed signal are found. Based on the optimal parameters and penalty factors, variational mode decomposition is performed on the decomposed signal to obtain several IMF components. Among the IMF components, effective components are selected based on the Pearson correlation coefficient, and the effective components are superimposed to construct the denoised signal.

[0006] Furthermore, the formula for calculating the adaptive weights is as follows:

[0007] in, For weight values, This represents the current iteration number. This represents the total number of iterations.

[0008] Furthermore, the calculation formula for the new candidate solution is as follows:

[0009] in, As a new candidate solution, This is the current optimal value. It is the Cauchy factor.

[0010] Furthermore, the formula for calculating the minimum envelope entropy is:

[0011] in, The original signal, For Hilbert transform, The envelope signal after demodulation. For the normalized envelope signal, The envelope entropy value. The length of the normalized envelope signal, For data point sequence numbers, This represents the total number of sampling points for the vibration signal.

[0012] Furthermore, the Pearson correlation coefficient includes the population Pearson correlation coefficient and the sample Pearson correlation coefficient, and the formula for calculating the population Pearson correlation coefficient is as follows:

[0013]

[0014]

[0015] in, The standard deviation of the vibration signal. It is a vibration signal. The standard deviation of the IMF component signal. For IMF component signals, The overall Pearson correlation coefficient, The covariance of the vibration signal and the IMF component signal. The formula for calculating the Pearson correlation coefficient of the sample is as follows:

[0016] in, The sample Pearson correlation coefficient, The arithmetic mean of the vibration signal. It is the arithmetic mean of the IMF component signals.

[0017] Furthermore, the selection of effective components based on the Pearson correlation coefficient includes: Based on the Pearson correlation coefficient of the sample, the components of the IMF that have a correlation coefficient greater than 0.4 with the vibration signal are selected, and the components with a correlation coefficient greater than 0.4 are the effective components.

[0018] Furthermore, the step of substituting the initial parameters of the algorithm and finding the optimal parameters and penalty factor in the decomposed signal includes: S1. Update the population based on the chaotic mapping; S2. Calculate the fitness of the new generation of population and compare it with the initial fitness. If the new fitness is smaller, update the fitness value; otherwise, update the population position based on the adaptive weight. S3. Check if the current iteration count has reached the maximum iteration count. If not, return to S2 and continue the iteration process. If it has, output the optimal solution, which is generated by introducing the Cauchy mutation operator and perturbation operation.

[0019] Furthermore, the initial parameters of the algorithm include: parameter dimension, value range, population size, and maximum number of iterations.

[0020] Compared with the prior art, the present invention has the following advantages: 1. The IBFO-VMD denoising method provided by this invention, through three improvements to the bitterling algorithm (chaotic initialization, adaptive weights, and Cauchy mutation), can more sensitively capture weak fault vibration information, greatly improve the accuracy and completeness of fault feature extraction, and achieve better fault feature extraction capability in strong noise environment.

[0021] 2. The IBFO-VMD denoising method provided by this invention can find the optimal parameter combination by automatically optimizing VMD parameters (K and α) without manual intervention, which greatly improves the automation and efficiency of the denoising process, while avoiding errors caused by human experience, thus overcoming the problem of traditional methods relying on experience settings.

[0022] 3. The IBFO-VMD denoising method provided by this invention achieves faster convergence speed and higher denoising accuracy.

[0023] Based on the above reasons, this invention can be widely applied in fields such as industrial equipment condition monitoring. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating a bearing vibration signal denoising method based on IBFO-optimized VMD according to the present invention.

[0026] Figure 2 This is a hardware design module diagram of an embodiment of the present invention.

[0027] Figure 3 This refers to the current transformer in this embodiment of the invention.

[0028] Figure 4 This is an infrared thermal image from an embodiment of the present invention.

[0029] Figure 5 This is the temperature and vibration sensor in the embodiment of the present invention.

[0030] Figure 6 This is the speed sensor in the embodiment of the present invention.

[0031] Figure 7 This is a diagram of the data acquisition terminal module in an embodiment of the present invention.

[0032] Figure 8 This is a diagram of the data transmission terminal module in an embodiment of the present invention.

[0033] Figure 9 This is the weighted comparison curve in an embodiment of the present invention.

[0034] Figure 10 This is the test function in the embodiments of the present invention.

[0035] Figure 11 These are the test results from the embodiments of the present invention.

[0036] Figure 12 This is a schematic diagram of the IBFO optimization VMD parameter process in an embodiment of the present invention.

[0037] Figure 13 This is a schematic diagram illustrating the relationship between the magnitude of the correlation coefficient and the degree of correlation in an embodiment of the present invention.

[0038] Figure 14 This is a schematic diagram of the denoising process based on IBFO-VMD in an embodiment of the present invention.

[0039] Figure 15 This is a schematic diagram of the simulation signal parameters in an embodiment of the present invention.

[0040] Figure 16 This is a time-domain diagram of the simulated signal in an embodiment of the present invention.

[0041] Figure 17 This is a schematic diagram of the improved bitterling algorithm parameter settings in an embodiment of the present invention.

[0042] Figure 18 This is the IMF time-frequency diagram of VMD decomposition in an embodiment of the present invention.

[0043] Figure 19 This is the result of the correlation coefficient calculation between the IMF component and the original signal in the embodiments of the present invention. Detailed Implementation

[0044] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0045] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0046] like Figure 1 As shown in the figure, this embodiment of the invention provides a method for denoising bearing vibration signals based on IBFO-optimized VMD, the specific steps of which are as follows: Step 1: Collect the vibration signal of the bearing.

[0047] Step 2: Improve the bitterling algorithm by initializing its parameters. Improvements include using chaotic mapping to optimize the initial population, introducing adaptive weights that decay exponentially with the number of iterations in position updates, and introducing a Cauchy mutation operator at the optimal solution and generating new candidate solutions through perturbation operations. The initial parameters include parameter dimension, value range, population size, and maximum number of iterations.

[0048] The traditional bitterling algorithm (BFO) simulates the behavior of bitterlings, employs probabilistic decision-making and iterative updates, and combines global and local searches in the search space to ultimately find the optimal solution to the problem.

[0049] The specific steps for constructing the traditional bitterling algorithm are as follows: 1. Create the initial BFO ​​population as shown in the formula. Let j represent the dimension of the i-th solution. Create a population in the interval [l, u]. The calculation formula is:

[0050] in, This serves as the lower bound of the search space. Let r be the upper bound of the search space, and r be a random number between 0 and 1.

[0051] Through the objective function To evaluate each fish or solution, as follows:

[0052] in, This represents the fitness value for each fish at its current position.

[0053] 2. Perform position updates to find the optimal solution. The oyster's holding state is shown below:

[0054] in, and Let i and t be the i-th solutions for the fish's current position and new position at times t and t+1, respectively. One of the randomly selected oyster populations, This is the optimal solution. The first random number, The second random number is the first random number, and the range of the first and second random numbers is 0 to 1. The number of steps or the rate at which a fish moves to escape or approach an oyster. A threshold parameter to control the selection probability.

[0055] As J decreases, the fish swarm may be more inclined to local search. To reduce the parameter J and to repeatedly search for the optimal solution in more space around the optimal solution, the parameter P is reduced over time to satisfy the condition r>p:

[0056] in, Let be the step size and jump value for each fish in the initial iteration. Let be the step size for the (t+1)th iteration. For the generated random sequence, This represents the current iteration number. This represents the maximum number of iterations.

[0057] The fish's position update states for the behaviors of escaping, not catching oysters, and producing are shown below, respectively, where parameter a is a set power; R is the distribution radius of the fish around the shell, which decreases with iteration. The value is equal to the average position of the fish school.

[0058]

[0059]

[0060] 3. The step size and selection probability decrease with each iteration, realizing the transition from global search to local exploration, and eliminating inferior solutions as shown below:

[0061] in, The first in the population i The probability that a solution is eliminated.

[0062] 4. Repeat steps 1 to 3 until the maximum number of iterations is reached, and output the global optimal solution.

[0063] Specifically, the embodiments of the present invention optimize and improve the bitterling algorithm, which includes the following steps: 1. First, optimize the initial population. If the initial population is unevenly distributed, it is easy to generate local optima, which will reduce the ability of global search. Since chaotic mapping has a better effect than pseudo-random numbers, chaotic mapping is used to optimize the initial population.

[0064] 2. Because the weights in the BFO algorithm fluctuate wildly in the early iterations and decrease linearly in the later stages, requiring more iterations to approach the optimal solution, an adaptive weight strategy that decays exponentially with the number of iterations is adopted to enhance the algorithm's convergence and local search capability. The formula for calculating the adaptive weights is:

[0065] in, For weight values, This represents the current iteration number. This represents the total number of iterations. A comparison of the weights before and after the improvement is shown below. Figure 9 .

[0066] 3. Introducing the Cauchy mutation operator at the optimal solution, new candidate solutions are generated through perturbation operations, thereby enhancing the algorithm's ability to escape local optima. The Cauchy mutation strategy is embedded into the Bitterling-Baby algorithm to further improve its global search and local refinement performance. The formula for calculating the new candidate solution is as follows:

[0067] in, As a new candidate solution, This is the current optimal value. It is the Cauchy factor.

[0068] To verify the performance of each algorithm, six algorithms—Grey Wolf Algorithm, Particle Swarm Optimization (PSO), Sparrow Search Algorithm, White Whale Optimization (BFO), and IBFO—were selected for a comparative experiment. The population size was set to 50, and the maximum number of iterations was 500. Tests were conducted on multi-peak (F1, F2) and fixed-dimensional multi-peak (F3, F4) test functions to evaluate the algorithms' global search performance and local search performance. The test functions are as follows: Figure 10 The corresponding experimental convergence curve test results are as follows: Figure 11 As shown.

[0069] Step 3: Perform variational mode decomposition on the vibration signal to obtain the decomposed signal, and calculate the initial fitness value in the decomposed signal using the minimum envelope entropy.

[0070] Specifically, the formula for calculating the minimum envelope entropy is:

[0071] in, The original signal, For Hilbert transform, The envelope signal after demodulation. For the normalized envelope signal, The envelope entropy value. The length of the normalized envelope signal, For data point sequence numbers, This represents the total number of sampling points for the vibration signal.

[0072] Step 4: Using the improved bitterling algorithm, after substituting the initial parameters of the algorithm, find the optimal parameters and penalty factors in the decomposed signal. Based on the optimal parameters and penalty factors, perform variational mode decomposition on the decomposed signal to obtain several IMF components.

[0073] Specifically, after substituting the initial parameters into the algorithm, the optimal parameters and penalty factor in the decomposed signal are found, including: S1. Update the population based on the chaotic mapping.

[0074] S2. Calculate the fitness of the new generation of population and compare it with the initial fitness. If the new fitness is smaller, update the fitness value; otherwise, update the population position based on the adaptive weight.

[0075] S3. Check if the current iteration count has reached the maximum iteration count. If not, return to S2 and continue the iteration process. If it has, output the optimal solution. The optimal solution is generated by introducing the Cauchy mutation operator and performing a perturbation operation.

[0076] Furthermore, the process of IBFO optimizing VMD parameters is as follows: Figure 12 As shown, it specifically includes: 1. Initialize IBFO parameters, setting parameter dimensions, value range, population size, and maximum number of iterations.

[0077] 2. Obtain the decomposition level K and penalty factor α of the initial VMD through chaotic mapping, and calculate the fitness value.

[0078] 3. Perform VMD decomposition on the signal, calculate the fitness value after decomposition, and determine whether it is the minimum. If it is, update the data; otherwise, leave it unchanged.

[0079] 4. Update the fish's position according to the IBFO algorithm.

[0080] 5. Determine if the maximum number of iterations has been reached. If it has, exit the loop; otherwise, return to VMD decomposition, calculate the fitness value, and repeat the loop.

[0081] 6. Find the optimal parameters K and α in IBFO and pass them into the VMD parameters to decompose the signal again.

[0082] Step 5: Among several IMF components, select the effective components based on the Pearson correlation coefficient, and construct the denoised signal from the effective components.

[0083] The correlation coefficient is a statistic that measures the strength of the linear association between two sets of data. The Pearson correlation coefficient is a commonly used linear correlation coefficient, and its value ranges from -1 to 1.

[0084] Specifically, the Pearson correlation coefficient includes the population Pearson correlation coefficient and the sample Pearson correlation coefficient. The formula for calculating the population Pearson correlation coefficient is as follows:

[0085]

[0086]

[0087] in, The standard deviation of the vibration signal. It is a vibration signal. The standard deviation of the IMF component signal. For IMF component signals, The overall Pearson correlation coefficient, Let be the covariance of the vibration signal and the IMF component signal.

[0088] The Pearson correlation coefficient can be estimated by calculating the sample covariance and sample standard deviation. It is often denoted by the lowercase Latin letter 'r'. The formula for calculating the sample Pearson correlation coefficient is:

[0089] in, The sample Pearson correlation coefficient, The arithmetic mean of the vibration signal. It is the arithmetic mean of the IMF component signals.

[0090] Based on the sample Pearson correlation coefficient, components in the IMF (Inductively Coupled Function) with a correlation coefficient greater than 0.4 with the vibration signal were selected as effective components. The correlation coefficient varies between [0,1]. Literature indicates that when the correlation coefficient between a modal component and the original signal is not lower than 0.4, it means that it contains relatively rich fault characteristics; while components with a correlation coefficient lower than 0.4 can be considered weakly or very weakly correlated and should be discarded. Finally, only components with high correlation coefficients are reconstructed, thus achieving effective signal denoising. The relationship between the magnitude of the correlation coefficient and the degree of correlation is as follows: Figure 13 As shown.

[0091] like Figure 14 As shown, the key parameters of variational mode decomposition (VMD), mode number K and penalty factor α, are first automatically tuned using the improved bitterling optimization (IBFO) algorithm. The bearing vibration signal is then subjected to VMD decomposition using the optimal parameter combination to obtain multiple IMF components. Subsequently, effective components with a correlation coefficient greater than 0.4 with the original signal are selected based on the Pearson correlation coefficient and reconstructed into a new signal to remove noise carried by other low-correlation components, thus completing the signal denoising process.

[0092] Step 6: Test the effectiveness of the IBFO-VMD denoising method.

[0093] Specifically, since the fidelity coefficient and precision coefficient in VMD have a limited impact on the decomposition results, the recommended default settings can be used directly. , A simulation signal is constructed to mimic the vibration characteristics of a rolling bearing failure. Its components include a sinusoidal signal, an amplitude-modulated signal, an impact signal, and noise. By superimposing these multiple signals, the simulation signal more closely approximates the vibration signal of a bearing failure under real-world operating conditions. The specific simulation signal is shown below.

[0094] in, For the synthesized vibration signal, The amplitude of the sinusoidal signal. and The modulation frequency of the amplitude modulation signal. The carrier frequency of the amplitude-modulated signal; The amplitude of periodic impact signals, Indicates the damping coefficient. For the impact cycle, Shaft rotation frequency; It is a sine wave signal. It is an amplitude modulation signal. As an impact signal, it constitutes a signal. It is a Gaussian white noise signal with a signal-to-noise ratio of -3dB.

[0095] By replacing the x3(t) signal with two sine functions and one cosine function, we can verify whether the optimized VMD can effectively remove noise while maintaining the independence of each frequency component. The simulation signal is shown below:

[0096] in, For synthesized vibration signals It is a sinusoidal signal. It is an amplitude modulation signal. For cosine signals, The signal is Gaussian white noise with a signal-to-noise ratio of -3dB. , Sine wave amplitude, The amplitude of the cosine signal. The frequency of the sinusoidal signal. The frequency of the cosine signal. The phase of the sine sign. The phase of the cosine signal.

[0097] Simulation signal parameters are as follows Figure 15 As shown.

[0098] Furthermore, the testing steps in this process are as follows: 1. Set the sampling time to 1 second and the sampling frequency to 1.5 kHz. The time-domain waveforms of the signals x1(t), x2(t), x3(t), x4(t), noise, and the synthesized vibration signal x(t) are shown below. Figure 16 .

[0099] 2. The simulated vibration signal is input into the IBFO-VMD framework for modal decomposition. The parameters K and α in VMD are optimized using the IBFO algorithm to select the optimal parameter combination. According to the IBFO VMD parameter optimization process, the parameters are first initialized. The improved bitterling algorithm parameter settings are as follows: Figure 17 As shown.

[0100] Using the minimum envelope entropy as the objective function, and substituting the initial parameters, we obtain the iterative fitness values ​​and the number of mode decompositions. The iterative change process of the penalty factor α.

[0101] 3. Substitute the optimal parameters obtained from step 2 into VMD for decomposition to obtain the time-domain plot and spectrum of the IMF, as shown below. Figure 18 As shown.

[0102] 4. Calculate the correlation coefficient between each IMF component in step 3 and the original signal, and obtain the results as follows: Figure 19The Pearson correlation coefficient between each obtained IMF component and the original signal is calculated. Components that meet the correlation threshold are selected for reconstruction to remove noise.

[0103] The hardware components involved in this invention include a current acquisition module composed of a current transformer, a temperature acquisition module composed of infrared thermal imaging, a vibration acquisition module composed of temperature and vibration sensors, a speed module, a data acquisition terminal module, and a data transmission terminal module. The hardware design module diagram is shown below. Figure 2 As shown.

[0104] Furthermore, the specific steps for the above hardware design are as follows: The current acquisition module is a clamp-on, open-type isolated transmitter for measuring single-phase AC current. The clamp-on orifice diameter is 50mm. Utilizing electromagnetic isolation, it converts the input AC current signal into a standard 4-20mA analog signal output, enabling rapid measurement of AC current signals in the field. Current transformers, such as... Figure 3 As shown.

[0105] The temperature acquisition module has a camera on the back, and above the camera is a FLIR Lepton 3.5 professional thermal imaging camera with a resolution of 160×120 and a measurement range of -20 to +400 degrees Celsius. The CAT S62 Pro's MyFLIR-Pro thermal imaging software can capture photos, videos, and time-lapse photography. It combines visible light images and thermal images via MSX (Multispectral Dynamic Imaging), offering various imaging styles. At the top are tools for adding anchor points, reference lines, etc. Infrared thermal images are shown below. Figure 4 .

[0106] The vibration acquisition module is a new type of integrated sensor that measures temperature and vibration in three directions in real time. It can simultaneously provide time-domain characteristic values ​​(RMS, peak value, kurtosis coefficient) of velocity and acceleration, spectral analysis data, and fault diagnosis results. It is widely applicable to condition monitoring and health management of rotating machinery such as motors, pumps, fans, bearings, air compressors, gas engines, generators, reducers, and gearboxes. Temperature and vibration sensors, such as... Figure 5 As shown.

[0107] The speed module uses the CZ480 speed transmitter, which can be widely used in the power, metallurgy, petrochemical, and paper industries to measure the speed and zero speed of large rotating machinery (such as steam turbines, compressors, motors, fans, pumps, etc.). The CZ480 speed transmitter senses the raised teeth or recessed grooves on the magnetic conductor using a Hall element, converts it into an analog signal of 4-20mA or an RS485 communication output, and then processes the signal through internal precision circuitry to achieve speed measurement. It can be used to detect the speed of various rotating machinery. Speed ​​sensors, such as... Figure 6 As shown.

[0108] The design of the data acquisition terminal module and the data transmission terminal module is as follows: Figure 7 , Figure 8 As shown.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for denoising bearing vibration signals based on IBFO-optimized VMD, characterized in that, Includes the following steps: Collect vibration signals from the bearing; The bitterling algorithm is improved by initializing the parameters of the improved bitterling algorithm to obtain the initial parameters of the algorithm. The improvement includes using chaotic mapping to optimize the initial population, introducing adaptive weights that decay exponentially with the number of iterations in the position update, introducing Cauchy mutation operator at the optimal solution and generating new candidate solutions through perturbation operation. The vibration signal is subjected to variational mode decomposition to obtain the decomposed signal, and the initial fitness value in the decomposed signal is calculated by the minimum envelope entropy. Using the improved bitterling algorithm, after substituting the initial parameters of the algorithm, the optimal parameters and penalty factors in the decomposed signal are found. Based on the optimal parameters and penalty factors, variational mode decomposition is performed on the decomposed signal to obtain several IMF components. Among the IMF components, effective components are selected based on the Pearson correlation coefficient, and the effective components are superimposed to construct the denoised signal.

2. The bearing vibration signal denoising method based on IBFO-optimized VMD according to claim 1, characterized in that, The formula for calculating the adaptive weight is: in, For weight values, This represents the current iteration number. This represents the total number of iterations.

3. The bearing vibration signal denoising method based on IBFO-optimized VMD according to claim 1, characterized in that, The formula for calculating the new candidate solution is: in, As a new candidate solution, This is the current optimal value. It is the Cauchy factor.

4. The bearing vibration signal denoising method based on IBFO-optimized VMD according to claim 1, characterized in that, The formula for calculating the minimum envelope entropy is: in, The original signal, For Hilbert transform, The envelope signal after demodulation. For the normalized envelope signal, The envelope entropy value. The length of the normalized envelope signal, For data point sequence numbers, This represents the total number of sampling points for the vibration signal.

5. The bearing vibration signal denoising method based on IBFO-optimized VMD according to claim 1, characterized in that, The Pearson correlation coefficient includes the overall Pearson correlation coefficient and the sample Pearson correlation coefficient. The formula for calculating the overall Pearson correlation coefficient is as follows: in, The standard deviation of the vibration signal. It is a vibration signal. The standard deviation of the IMF component signal. For IMF component signals, The overall Pearson correlation coefficient, The covariance of the vibration signal and the IMF component signal. The formula for calculating the Pearson correlation coefficient of the sample is as follows: in, The sample Pearson correlation coefficient, The arithmetic mean of the vibration signal. It is the arithmetic mean of the IMF component signals.

6. The bearing vibration signal denoising method based on IBFO-optimized VMD according to claim 1 or 5, characterized in that, The selection of effective components based on Pearson correlation coefficient includes: Based on the Pearson correlation coefficient of the sample, the components of the IMF that have a correlation coefficient greater than 0.4 with the vibration signal are selected, and the components with a correlation coefficient greater than 0.4 are the effective components.

7. The bearing vibration signal denoising method based on IBFO-optimized VMD according to claim 1, characterized in that, The step of substituting the initial parameters of the algorithm and finding the optimal parameters and penalty factor in the decomposed signal includes: S1. Update the population based on the chaotic mapping; S2. Calculate the fitness of the new generation of population and compare it with the initial fitness. If the new fitness is smaller, update the fitness value; otherwise, update the population position based on the adaptive weight. S3. Check if the current iteration count has reached the maximum iteration count. If not, return to S2 and continue the iteration process. If it has, output the optimal solution, which is generated by introducing the Cauchy mutation operator and perturbation operation.

8. The bearing vibration signal denoising method based on IBFO-optimized VMD according to claim 1, characterized in that, The initial parameters of the algorithm include: parameter dimension, value range, population size, and maximum number of iterations.

Images