A method, device, equipment and medium for signal denoising

Abstract

The embodiment of the present invention provides a method, device, equipment and medium for signal denoising, the method comprising: obtaining a vibration signal of a mechanical device collected by a sensor; decomposing the vibration signal to obtain multiple signal components; calculating the energy entropy of the multiple signal components respectively, and determining the noise component and the non-noise component from the multiple signal components according to the energy entropy; denoising the noise component to obtain a processed noise component; reconstructing the signal of the processed noise component and the non-noise component to obtain a denoised signal. Through the embodiment of the present invention, effective denoising and purification of vibration signals of mechanical equipment is achieved, which significantly improves the accuracy of fault diagnosis and the reliability of the early warning system, and provides strong technical support for the maintenance management and safe operation of mechanical equipment.

Description

Technical Field

[0001] The present invention relates to the field of denoising technology, and in particular to a method, device, equipment and medium for signal denoising processing. Background Art

[0002] In the current field of industrial production and equipment maintenance, the condition monitoring and fault diagnosis of mechanical equipment are extremely important. In order to achieve this goal, in the existing technology, vibration signals of mechanical equipment are collected through various sensors. These vibration signals contain rich equipment operation information, which can reflect the health status, workload and potential failure modes of the equipment, and perform condition assessment, fault diagnosis and early warning of mechanical equipment, thereby effectively avoiding production interruptions and safety accidents caused by sudden equipment failures.

[0003] However, in practical applications, since the equipment operating environment is often complex and changeable, the sensor is very susceptible to various external interferences during the process of collecting vibration signals, such as air vibration (such as wind, sound waves, etc.), electrical performance fluctuations caused by humidity changes, and electromagnetic interference (such as radio waves, power line interference, etc.). These interference factors will inevitably be superimposed on the vibration signal, making the collected vibration signal lose accuracy and reliability, and thus making the results of fault diagnosis and early warning unsatisfactory. Summary of the invention

[0004] In view of the above problems, a method, device, equipment and medium for signal denoising processing is proposed to overcome the above problems or at least partially solve the above problems, including:

[0005] A method for signal denoising, the method comprising:

[0006] Obtain vibration signals of mechanical equipment collected by sensors;

[0007] Decomposing the vibration signal to obtain multiple signal components;

[0008] Calculating energy entropies of the multiple signal components respectively, and determining noise components and non-noise components from the multiple signal components according to the energy entropies;

[0009] Performing denoising processing on the noise component to obtain a processed noise component;

[0010] The processed noise component and non-noise component are reconstructed to obtain a denoised signal.

[0011] Optionally, the performing denoising processing on the noise component to obtain a processed noise component includes:

[0012] For noise components of different layers in the noise components, calculating corresponding hierarchical thresholds;

[0013] The noise components of different layers in the noise components are denoised by using corresponding hierarchical thresholds to obtain processed noise components.

[0014] Optionally, the calculating corresponding hierarchical thresholds for noise components of different layers in the noise components includes:

[0015] For noise components of different layers in the noise components, wavelet coefficients are determined, and corresponding hierarchical thresholds are calculated based on the wavelet coefficients.

[0016] Optionally, determining a noise component and a non-noise component from the multiple signal components according to the energy entropy includes:

[0017] Calculating an energy entropy mean value according to the energy entropies of the multiple signal components;

[0018] A noise component and a non-noise component are determined from the multiple signal components according to the energy entropy mean value and the energy entropies of the multiple signal components.

[0019] Optionally, determining a noise component and a non-noise component from the multiple signal components according to the energy entropy mean and the energy entropy of the multiple signal components includes:

[0020] Determine whether the energy entropy of each signal component is greater than the energy entropy mean;

[0021] When the energy entropy is greater than the energy entropy mean, determining that the signal component is a noise component;

[0022] When the energy entropy is less than or equal to the energy entropy mean, it is determined that the signal component is a non-noise component.

[0023] Optionally, decomposing the vibration signal to obtain multiple signal components includes: performing adaptive noise complete set empirical mode decomposition on the vibration signal to determine multiple signal components.

[0024] Optionally, the signal component is a signal component based on an intrinsic mode function.

[0025] A device for signal denoising, the device comprising:

[0026] A vibration signal acquisition module, used to acquire the vibration signal of the mechanical equipment collected by the sensor;

[0027] A signal decomposition module, used for decomposing the vibration signal to obtain multiple signal components;

[0028] An energy entropy classification module, used to calculate the energy entropies of the multiple signal components respectively, and determine the noise component and the non-noise component from the multiple signal components according to the energy entropy;

[0029] A denoising processing module, used for performing denoising processing on the noise component to obtain a processed noise component;

[0030] The signal reconstruction module is used to reconstruct the processed noise component and non-noise component to obtain a denoised signal.

[0031] An electronic device comprises a processor, a memory and a computer program stored in the memory and capable of running on the processor, wherein the computer program implements the method described above when executed by the processor.

[0032] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described above is implemented.

[0033] The embodiments of the present invention have the following advantages:

[0034] In an embodiment of the present invention, a vibration signal of a mechanical device collected by a sensor is acquired; the vibration signal is decomposed to obtain a plurality of signal components; the energy entropy of the plurality of signal components is respectively calculated, and according to the energy entropy, a noise component and a non-noise component are determined from the plurality of signal components; the noise component is denoised to obtain a processed noise component; and the processed noise component and the non-noise component are reconstructed to obtain a denoised signal, thereby achieving effective denoising and purification of the vibration signal of the mechanical device, significantly improving the accuracy of fault diagnosis and the reliability of the early warning system, and providing strong technical support for the maintenance management and safe operation of the mechanical equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the description of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0036] Figure 1 is a schematic diagram of a signal denoising processing method provided by some embodiments of the present invention;

[0037] Figure 2 is a flowchart of the steps of a signal denoising method provided by some embodiments of the present invention;

[0038] Figure 3is a schematic diagram of an improved threshold function graph provided by some embodiments of the present invention;

[0039] Figure 4 It is a structural block diagram of a signal denoising device provided by some embodiments of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] In related technologies, the vibration signal collected by the sensor is a key data source, and its quality and credibility are crucial for accurately evaluating the equipment status and predicting potential failures. However, due to the complex operating environment of the equipment, the characteristics of the sensor itself, and various interferences during the transmission process, the vibration signal is often mixed with a large amount of noise, which greatly affects the accuracy and reliability of signal analysis. Therefore, how to effectively reduce noise and improve the data quality of vibration signals is an important issue that needs to be solved in the current signal processing field.

[0042] Vibration signals usually have nonlinear and non-stationary characteristics, which makes it difficult for traditional Fourier transform-based denoising algorithms to achieve ideal results. Although Fourier transform can perform detailed analysis of signals in the frequency domain, it has limitations when processing non-stationary signals and cannot accurately capture the local characteristics of the signal.

[0043] The current signal denoising field mainly explores new denoising algorithms based on signal decomposition, signal transformation and deep learning. Denoising algorithms based on signal decomposition, such as singular value decomposition, empirical mode decomposition and variational mode decomposition, decompose the original signal into multiple components, then denoise each component, and finally reconstruct the denoised signal.

[0044] However, these algorithms have certain limitations in practical applications. For example, although the signal processing algorithm based on singular value decomposition can process non-stationary and nonlinear signals such as vibration signals, and has good numerical robustness and unique data processing methods, it has a large amount of calculation and high time complexity, which ultimately leads to a slow algorithm operation speed.

[0045] Empirical mode decomposition can process vibration signals well, does not require prior knowledge, and has good time domain adaptability. After empirical mode decomposition, the signal is decomposed into different IMF components from high frequency to low frequency, which is usually used in signal denoising algorithms. However, empirical mode decomposition is prone to endpoint effects, and the filtered components are not easy to divide, so it is usually necessary to combine it with other noise reduction methods.

[0046] Variational mode decomposition is adaptive and can adapt to the local frequency characteristics of the signal. It can effectively separate different modes of the signal in different frequency ranges and is often used to process non-stationary signals. However, the variational mode decomposition algorithm requires manual parameter setting and has poor adaptability.

[0047] The denoising algorithms based on signal transformation mainly include wavelet transform denoising algorithms. Wavelet transform has the characteristics of multi-resolution analysis and time-frequency localization. It can analyze the signal in both time domain and frequency domain at the same time. It is also commonly used in the field of non-stationary signal denoising. However, wavelet transform requires manual selection of appropriate wavelet basis to achieve better denoising effect, and its adaptability is poor.

[0048] In recent years, deep learning has been widely used in the field of noise reduction due to its flexible network structure and powerful learning ability. The noise reduction algorithm based on deep learning has achieved good noise reduction effect by designing different network structures and training a large number of samples. According to the different network structures, it can be mainly divided into three categories: noise reduction model based on noise reduction autoencoder, deep noise reduction network based on residual learning, and noise reduction model based on generative adversarial network. Although deep learning has good learning ability and can well restore the detailed features of the signal, its training requires a large amount of sample data and label data. At the same time, the increase in model parameters is also prone to overfitting, and it is difficult to converge during training.

[0049] In addition, existing denoising algorithms often combine threshold denoising algorithms with other decomposition algorithms in order to achieve better denoising effects. However, threshold algorithms in related technologies often use fixed thresholds, which have limited denoising effects for signals containing different noise components. At the same time, traditional threshold functions also have limitations, resulting in unsatisfactory denoising effects.

[0050] Since the hard threshold function in the related art keeps the components greater than or equal to the threshold unchanged and sets the components less than the threshold to 0, this method can completely retain the information greater than the threshold, but the signal is prone to oscillation points and jump points, and the smoothness is poor. Secondly, its soft threshold function reduces the components greater than the threshold and sets the components less than the threshold to 0, thus avoiding the oscillation phenomenon near the threshold and having better smoothness, but a constant deviation will be generated after the signal is compressed.

[0051] Therefore, in an embodiment of the present invention, an adaptive threshold denoising algorithm based on adaptive complete ensemble empirical mode decomposition (CEEMDAN) is proposed. CEEMDAN is an improvement on the empirical mode decomposition algorithm, which introduces adaptive noise, making it more robust in the case of mixed signals and noise, and can avoid the problem of endpoint effect, reduce the generation of modal aliasing, avoid the pollution of residual noise, and obtain IMF (Intrinsic Mode Function) components with different frequency distributions after decomposition. Since noise is usually distributed in different IMF components, and the noise-dominated components generally have higher energy entropy, by calculating the energy entropy of each IMF component, high-frequency components with more noise content can be adaptively screened out. The noise content in different components is different. For different noise components, the wavelet threshold is calculated separately as the hierarchical threshold, which can effectively improve the denoising effect. Secondly, the threshold function is improved to avoid the generation of constant deviation under the premise of ensuring signal smoothness.

[0052] In order to solve the shortcomings of the denoising algorithm in the related art on non-stationary signals such as vibration signals and the defects of soft and hard threshold functions, the denoising algorithm has a better denoising effect, effectively improves the signal-to-noise ratio of the vibration signal, and has strong adaptability. In the embodiment of the present invention, the main points are as follows:

[0053] 1. First, CEEMDAN is used to decompose the signal, which can avoid the endpoint effect problem caused by the EMD (Empirical Mode Decomposition) algorithm and decompose the signal into IMF components of different frequencies.

[0054] 2. The noise content of the decomposed IMF components is different. We calculate the energy entropy of each component and filter the components according to the size of the energy entropy to filter out the noise components and low-frequency components.

[0055] 3. Keep the low-frequency components and perform adaptive threshold denoising on the filtered high-frequency components. Most traditional threshold denoising algorithms use the same threshold, which has limited denoising effects on components with different noise contents. Therefore, this paper calculates a threshold for each layer of IMF components. Traditional soft and hard threshold denoising functions have certain problems. Based on the traditional soft and hard threshold functions, this paper introduces an improved threshold function, which improves the smoothness of the signal while avoiding the generation of constant errors. It can effectively enhance the adaptability of the denoising algorithm and improve the denoising effect.

[0056] 4. Reconstruct the high-frequency component after threshold processing and the retained low-frequency component to obtain the denoised signal.

[0057] In the specific implementation, Figure 1 , the embodiment of the present invention may include the following steps:

[0058] [1] Input noise component.

[0059] [2] CEEMDAN decomposition.

[0060] [3] Energy entropy screening.

[0061] [4] Determine whether it is greater than the energy entropy threshold. If so, proceed to [5]; otherwise, proceed to the low-frequency component and proceed to [9].

[0062] [5] Noise component.

[0063] [6] Wavelet thresholding calculates hierarchical thresholds.

[0064] [7] Improve the threshold function to handle noise components.

[0065] [8] Noise component after processing.

[0066] [9]Signal reconstruction.

[0067]

[10] Get the denoised signal.

[0068] Compared with the related art, the embodiments of the present invention have the following beneficial effects:

[0069] 1. The vibration signal denoising algorithms in the related art are all based on EMD decomposition, but the EMD algorithm will produce modal aliasing and endpoint effects during the decomposition process, which will affect the final decomposition effect. Therefore, in this embodiment, the CEEMDAN algorithm is used for decomposition, which can solve the problem of residual noise transmission during the decomposition process, avoid contaminating the reconstructed signal, and can effectively solve the problems of modal aliasing and endpoint effects.

[0070] 2. In the denoising algorithm in the related art, the IMF components after decomposition mostly need to be divided manually. However, there are large errors in manual division and poor adaptability. Therefore, in this embodiment, an energy entropy screening algorithm is adopted to realize adaptive screening of components by calculating the energy entropy of different components, which effectively avoids the occurrence of human errors.

[0071] 3. In the threshold denoising algorithms in the related art, a fixed threshold is mostly used to denoise all components, which has poor adaptability for components with different noise contents. At the same time, the traditional soft and hard threshold denoising algorithms have certain defects in theory. In this embodiment, the thresholds are calculated for components with different noise contents, and an improved threshold algorithm is also introduced, which can avoid the problem of discontinuous signals after denoising by the hard threshold algorithm in the related art, and solve the problem of constant errors generated by the soft threshold function, which can effectively improve the signal-to-noise ratio of the denoised signal.

[0072] The vibration signal adaptive denoising algorithm based on CEEMDAN proposed in the embodiment of the present invention can effectively improve the denoising effect and the signal-to-noise ratio while improving the adaptability.

[0073] The present invention is described below with reference to the accompanying drawings:

[0074] Reference Figure 2 , shows a flowchart of a signal denoising method provided by some embodiments of the present invention, which may specifically include the following steps:

[0075] Step 201: Acquire a vibration signal of a mechanical device collected by a sensor.

[0076] Among them, a sensor is a device that can detect physical quantities (such as temperature, pressure, displacement, vibration, etc.) and convert them into measurable signals. In mechanical equipment vibration monitoring, for example, an accelerometer, velocity sensor or displacement sensor can be used to detect the vibration of the equipment.

[0077] As some examples, sensors can be installed in mechanical equipment to capture its vibrations. When the mechanical equipment is running, the moving parts inside it will generate vibrations, which are sensed by the sensor and converted into electrical signals. The electrical signal (such as an analog signal) output by the sensor is converted into a digital signal after pre-processing such as amplification and filtering for subsequent analysis and processing.

[0078] In some examples, the collected vibration signals can be processed by methods such as spectrum analysis, time domain analysis, wavelet analysis, etc. to extract useful information. This information can be used to identify the operating status of mechanical equipment, such as whether it is operating normally, whether there is a fault or wear, etc.

[0079] Step 202: decompose the vibration signal to obtain multiple signal components.

[0080] In practical applications, the vibration signals generated by mechanical equipment during operation often contain multiple frequency components, which may come from different parts of the equipment or different operating states. The complexity of the vibration signal makes direct analysis difficult, so it is necessary to decompose it into simpler signal components for more in-depth research and analysis.

[0081] Signal decomposition refers to decomposing the original vibration signal into multiple independent signal components with clear physical meanings. These components can represent the vibration characteristics of different parts in the equipment, or reflect the vibration behavior of the equipment under different operating conditions.

[0082] For example, the vibration signal can be decomposed by CEEMDAN to obtain the IMF components from high frequency to low frequency.

[0083] In the embodiment of the present invention, by changing the way of adding auxiliary noise and the decomposition process through CEEMDAN, the transmission of residual noise can be avoided, the residual noise can be isolated, the reconstructed signal can be prevented from being contaminated by noise, and the endpoint effect and modal aliasing problems can be effectively solved.

[0084] In some embodiments of the present invention, the signal component is a signal component based on an intrinsic mode function.

[0085] In practical applications, the intrinsic mode function is the core concept in the empirical mode decomposition method. An IMF is a signal component that meets specific conditions and represents an inherent vibration mode or frequency component in the signal. Among them, the IMF must meet two conditions: one is that the number of local extreme points and zero crossing points is equal or differs by at most one; the other is that at any point, the average value of the upper envelope defined by the local maximum point and the lower envelope defined by the local minimum point is zero.

[0086] As some examples, the vibration signal can be decomposed into a series of signal components, each of which represents a natural vibration mode in the signal.

[0087] In some embodiments of the present invention, decomposing the vibration signal to obtain multiple signal components includes: performing adaptive noise complete set empirical mode decomposition on the vibration signal to determine multiple signal components.

[0088] As some examples, CEEMDAN (Complete Ensemble Empirical Mode with Adaptive Noise) is an improved empirical mode decomposition (EMD) method that aims to solve the modal aliasing problem in the EMD method and improve the stability and accuracy of the decomposition. Compared with the EMD method, CEEMDAN enhances the decomposition effect of the signal by introducing adaptive noise, making it more robust in the case of mixed signals and noise, and can avoid the endpoint effect problem, reduce the generation of modal aliasing, avoid the pollution of residual noise, and obtain IMF components with different frequency distributions after decomposition, thereby obtaining more accurate signal components.

[0089] In some examples, adaptive noise can play two main roles in CEEMDAN: first, it increases the diversity of the signal, which helps to separate clearer IMFs (intrinsic mode functions); second, it reduces the impact of noise on the decomposition results through multiple iterations and averaging. The introduced adaptive noise can be a white noise of limited amplitude, and its amplitude and amount are adjusted according to the characteristics of the signal during the decomposition process.

[0090] For example, a white noise of limited amplitude can be added to the vibration signal, and then the EMD method is applied to obtain a set of IMFs. Next, the local mean of these IMFs is calculated and used as the new signal for the next EMD decomposition, while adding adaptive noise again. This process is repeated until the predetermined stop condition is met (such as reaching a predetermined number of IMFs or the decomposed residual meets certain conditions). Finally, through multiple iterations and averaging, a stable set of IMFs is obtained, which are multiple signal components of the vibration signal.

[0091] Step 203: Calculate energy entropies of the multiple signal components respectively, and determine noise components and non-noise components from the multiple signal components according to the energy entropies.

[0092] In practical applications, the size of entropy reflects the uncertainty and complexity of the signal. Energy entropy indicates the energy distribution of the signal. A large energy entropy indicates that the proportion of useful signals in the total energy is small, while a small energy entropy indicates that the proportion of useful signals in the total energy is large.

[0093] Among them, the energy entropy of the signal component is an indicator to measure the complexity of the energy distribution of the signal component. The larger the energy entropy, the more complex the energy distribution of the signal component, that is, the more or more dispersed the frequency components contained in the signal. On the contrary, the smaller the energy entropy, the simpler the energy distribution of the signal component, that is, the fewer or more concentrated the frequency components contained in the signal.

[0094] Signal components can be classified according to the size of energy entropy. Among them, noise components have higher energy entropy because they contain multiple frequency components, and these components are random in time and amplitude. Non-noise components (such as the natural vibration mode of the device) have lower energy entropy because they usually contain fewer frequency components, and these components have certain regularity in time and amplitude.

[0095] As some examples, an energy entropy threshold may be set, and signal components with energy entropy higher than the threshold may be determined as noise components, and signal components with energy entropy lower than the threshold may be determined as non-noise components.

[0096] As some examples, energy entropy can be used to distinguish noise signals from useful signals (non-noise components), and the calculated value of the energy entropy of each signal component can be:

[0097] H(u k )=-P k log2P k

[0098] Among them, H(u k ) represents the component u k The energy entropy value, P k represents the kth component u k The proportion of energy in the total energy, E k is the energy of component k, and E is the sum of the total energies of the components.

[0099] In some embodiments of the present invention, determining the noise component and the non-noise component from the multiple signal components based on the energy entropy includes: calculating an energy entropy mean based on the energy entropy of the multiple signal components; and determining the noise component and the non-noise component from the multiple signal components based on the energy entropy mean and the energy entropy of the multiple signal components.

[0100] After calculating the energy entropies of multiple signal components, the energy entropies of all signal components can be averaged to obtain the energy entropy mean, which represents the average level of the energy distribution complexity of all signal components.

[0101] Among them, after obtaining the energy entropy mean, it can be used as a reference standard to distinguish between noise components and non-noise components. The energy entropy of each signal component is compared with the energy entropy mean. If the energy entropy of the signal component is higher than the energy entropy mean, it can be determined as a noise component; if the energy entropy of the signal component is lower than the energy entropy mean, it can be determined as a non-noise component.

[0102] In some embodiments of the present invention, determining noise components and non-noise components from the multiple signal components based on the energy entropy mean and the energy entropy of the multiple signal components includes: judging whether the energy entropy of each signal component is greater than the energy entropy mean; determining that the signal component is a noise component when the energy entropy is greater than the energy entropy mean; and determining that the signal component is a non-noise component when the energy entropy is less than or equal to the energy entropy mean.

[0103] In practical applications, for each signal component, its energy entropy can be compared with the mean energy entropy. If the energy entropy of a signal component is greater than the mean energy entropy, it means that the energy distribution of the signal component is more complex than the overall average level, and may contain more random or irregular components, and can be determined as a noise component. If the energy entropy of a signal component is less than or equal to the mean energy entropy, it means that the energy distribution of the signal component is relatively simple, and may contain fewer random components, and can therefore be determined as a non-noise component.

[0104] Step 204: denoising the noise component to obtain a processed noise component.

[0105] After determining the noise components and non-noise components, the identified noise components can be denoised to reduce or eliminate the impact of these noise components on the vibration signal. The denoising process can be performed using methods such as filtering, transform domain processing, statistical modeling, etc. After the noise components are denoised, the original noise components can be weakened or eliminated to a certain extent.

[0106] In some embodiments of the present invention, the denoising process of the noise component to obtain the processed noise component includes: calculating corresponding hierarchical thresholds for noise components of different layers in the noise component; and denoising the noise components of different layers in the noise component using the corresponding hierarchical thresholds to obtain the processed noise component.

[0107] As some examples, the noise components may be decomposed into different levels or layers by wavelet transform, multi-scale analysis, etc., to determine the noise components of different layers in the noise components, and then calculate the corresponding hierarchical thresholds.

[0108] For each level of noise components, a corresponding hierarchical threshold can be calculated according to its characteristics (such as energy, amplitude, frequency, etc.). This threshold is used to distinguish the noise components and possible useful signal components in the level to ensure that the useful signal is retained as much as possible while removing the noise.

[0109] The calculated hierarchical threshold is used to perform denoising on the noise components at each level of the noise component, such as comparing the noise components at each level with the corresponding hierarchical threshold, and deciding whether to retain or remove certain components at that level based on the comparison result.

[0110] The denoising process may include hard threshold processing (directly removing components exceeding the threshold), soft threshold processing (attenuating components exceeding the threshold), or other more complex processing strategies.

[0111] As some examples, soft and hard threshold functions can be used as a threshold denoising method, where the mathematical expression of the hard threshold function is:

[0112]

[0113] The mathematical expression of the soft threshold function is:

[0114]

[0115] From the above expression, we can see that the hard threshold function keeps the coefficients greater than the threshold unchanged and sets the coefficients less than the threshold to 0. The hard threshold function completely retains the components greater than the threshold, but loses the smoothness of the signal, and there will be additional oscillation points and jump points. The soft threshold function reduces the components greater than the threshold and sets the components less than the threshold to 0, which can ensure the continuity of the signal. However, due to the compression of the signal, a constant deviation will eventually be generated, which will also affect the final denoising effect.

[0116] In practical applications, the threshold of the threshold denoising algorithm is mostly a fixed threshold for threshold denoising of all components. However, the noise contained in each layer of the signal after CEEADAN decomposition is inconsistent, so the denoising effect of all components using the same threshold is limited. Therefore, in this embodiment, different thresholds are calculated for IMF components with different decomposition degrees to perform threshold denoising. The expression of the threshold is:

[0117]

[0118] Among them, σ represents the noise variance, N represents the signal length. In the field of image denoising, wavelet transform can be used to estimate noise, because after wavelet transform, the energy is mainly concentrated in the sub-bands with large scales, and the high-frequency sub-bands with small scales have small amplitudes and low energy.

[0119] Therefore, when the noise variance is relatively large, the highest frequency subband coefficients can all be regarded as noise, and the noise variance can be estimated accordingly. The estimation formula for the noise standard variance can be:

[0120]

[0121] Wherein, MAD is the median of the HH subband wavelet coefficients, and σ represents the noise variance.

[0122] Since both the soft and hard threshold functions have certain disadvantages, this embodiment adopts an improved soft threshold function, which is expressed as follows:

[0123]

[0124] When the threshold is set to 2, the image is expressed as follows Figure 3 According to the expression and image, the improved threshold function is between the soft and hard threshold functions. Compared with the soft threshold function, it has continuity and solves the problem of constant error of the soft threshold function. With w i，j The error between |w i，j | becomes larger and gradually becomes smaller, which has good adaptability and can effectively improve the denoising effect.

[0125] In some embodiments of the present invention, calculating corresponding layered thresholds for noise components of different layers in the noise components includes: determining wavelet coefficients for noise components of different layers in the noise components, and calculating corresponding layered thresholds based on the wavelet coefficients.

[0126] As some examples, wavelet analysis (a mathematical tool) may be used to decompose noise components of different layers in the noise component into components at different frequencies and locations, which are called wavelet coefficients.

[0127] Among them, wavelet coefficients are the core output of wavelet analysis, reflecting the characteristics of noise components at different scales and locations. For each layer of noise components, there will be a set of wavelet coefficients, which describe the characteristics of the noise components of this layer. When processing noise components, these coefficients are key because they can be used to distinguish the basis of signal and noise.

[0128] After the wavelet coefficients are determined, the stratification threshold can be calculated based on these coefficients. The stratification threshold is a boundary used to distinguish between non-noise and noise. The wavelet coefficient value corresponding to noise will be relatively small, while the wavelet coefficient value corresponding to non-noise will be relatively large. By setting a threshold, the wavelet coefficients below the threshold (considered to be noise) can be distinguished from the wavelet coefficients above the threshold (considered to be non-noise). This threshold can be calculated based on the statistical characteristics of the wavelet coefficients (such as standard deviation, mean, etc.) to ensure that it can effectively separate the signal and noise.

[0129] Step 205: reconstruct the processed noise component and non-noise component to obtain a denoised signal.

[0130] Among them, signal reconstruction refers to the process of denoising the noise component to obtain the processed noise component, and then recombining the processed noise component with the non-noise component into a complete vibration signal to obtain a denoised signal, which reduces the interference of noise while retaining the main features of the original vibration signal.

[0131] In an embodiment of the present invention, a vibration signal of a mechanical device collected by a sensor is acquired; the vibration signal is decomposed to obtain a plurality of signal components; the energy entropy of the plurality of signal components is respectively calculated, and according to the energy entropy, a noise component and a non-noise component are determined from the plurality of signal components; the noise component is denoised to obtain a processed noise component; and the processed noise component and the non-noise component are reconstructed to obtain a denoised signal, thereby achieving effective denoising and purification of the vibration signal of the mechanical device, significantly improving the accuracy of fault diagnosis and the reliability of the early warning system, and providing strong technical support for the maintenance management and safe operation of the mechanical equipment.

[0132] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0133] Reference Figure 4 , shows a schematic diagram of the structure of a signal denoising device provided by some embodiments of the present invention, which may specifically include the following modules:

[0134] A vibration signal acquisition module 401 is used to acquire a vibration signal of a mechanical device collected by a sensor;

[0135] A signal decomposition module 402, used to decompose the vibration signal to obtain multiple signal components;

[0136] An energy entropy classification module 403 is used to calculate the energy entropies of the multiple signal components respectively, and determine the noise component and the non-noise component from the multiple signal components according to the energy entropies;

[0137] A denoising processing module 404 is used to perform denoising processing on the noise component to obtain a processed noise component;

[0138] The signal reconstruction module 405 is used to reconstruct the processed noise component and non-noise component to obtain a denoised signal.

[0139] In some embodiments of the present invention, the denoising processing module 404 includes:

[0140] A layered threshold calculation submodule, used for calculating corresponding layered thresholds for noise components of different layers in the noise components;

[0141] The denoising processing submodule is used to perform denoising processing on noise components of different layers in the noise component by using corresponding layered thresholds to obtain processed noise components.

[0142] In some embodiments of the present invention, the hierarchical threshold calculation submodule includes:

[0143] The wavelet coefficient determination unit is used to determine the wavelet coefficients of the noise components of different layers in the noise components, and calculate the corresponding layered thresholds according to the wavelet coefficients.

[0144] In some embodiments of the present invention, the energy entropy classification module 403 includes:

[0145] An entropy mean calculation submodule, used to calculate an energy entropy mean according to the energy entropies of the multiple signal components;

[0146] The component determination submodule is used to determine the noise component and the non-noise component from the multiple signal components according to the energy entropy mean value and the energy entropy of the multiple signal components.

[0147] In some embodiments of the present invention, the component determination submodule includes:

[0148] An energy entropy judgment unit, used to judge whether the energy entropy of each signal component is greater than the energy entropy mean;

[0149] A first determining unit, configured to determine that the signal component is a noise component when the energy entropy is greater than the energy entropy mean;

[0150] The second determining unit is used to determine that the signal component is a non-noise component when the energy entropy is less than or equal to the energy entropy mean.

[0151] In some embodiments of the present invention, the signal decomposition module 402 is used to perform adaptive noise complete set empirical mode decomposition on the vibration signal to determine multiple signal components.

[0152] In some embodiments of the present invention, the signal component is a signal component based on an intrinsic mode function.

[0153] In an embodiment of the present invention, a vibration signal of a mechanical device collected by a sensor is acquired; the vibration signal is decomposed to obtain a plurality of signal components; the energy entropy of the plurality of signal components is respectively calculated, and according to the energy entropy, a noise component and a non-noise component are determined from the plurality of signal components; the noise component is denoised to obtain a processed noise component; and the processed noise component and the non-noise component are reconstructed to obtain a denoised signal, thereby achieving effective denoising and purification of the vibration signal of the mechanical device, significantly improving the accuracy of fault diagnosis and the reliability of the early warning system, and providing strong technical support for the maintenance management and safe operation of the mechanical equipment.

[0154] Some embodiments of the present invention further provide an electronic device, comprising a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the above method is implemented when the computer program is executed by the processor.

[0155] Some embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented.

[0156] Some embodiments of the present invention further provide a computer program product, including a computer program, which implements the above method when executed by a processor.

[0157] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0158] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0159] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0160] Those skilled in the art will appreciate that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0161] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0162] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0164] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0165] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the above elements.

[0166] The above is a detailed introduction to the provided signal denoising processing method, device, equipment and medium. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for signal denoising, characterized in that: The method comprises: Obtain vibration signals of mechanical equipment collected by sensors; Decomposing the vibration signal to obtain multiple signal components; Calculating energy entropies of the multiple signal components respectively, and determining noise components and non-noise components from the multiple signal components according to the energy entropies; Performing denoising processing on the noise component to obtain a processed noise component; The processed noise component and non-noise component are reconstructed to obtain a denoised signal.

2. The method according to claim 1, characterized in that The denoising process is performed on the noise component to obtain a processed noise component, including: For noise components of different layers in the noise components, calculating corresponding hierarchical thresholds; The noise components of different layers in the noise components are denoised by using corresponding hierarchical thresholds to obtain processed noise components.

3. The method according to claim 2, characterized in that The step of calculating corresponding hierarchical thresholds for noise components of different layers in the noise components comprises: For noise components of different layers in the noise components, wavelet coefficients are determined, and corresponding hierarchical thresholds are calculated based on the wavelet coefficients.

4. The method according to any one of claims 1 to 3, characterized in that: Determining the noise component and the non-noise component from the multiple signal components according to the energy entropy includes: Calculating an energy entropy mean value according to the energy entropies of the multiple signal components; A noise component and a non-noise component are determined from the multiple signal components according to the energy entropy mean value and the energy entropies of the multiple signal components.

5. The method according to claim 4, characterized in that The determining of the noise component and the non-noise component from the multiple signal components according to the energy entropy mean and the energy entropy of the multiple signal components comprises: Determine whether the energy entropy of each signal component is greater than the energy entropy mean; When the energy entropy is greater than the energy entropy mean, determining that the signal component is a noise component; When the energy entropy is less than or equal to the energy entropy mean, it is determined that the signal component is a non-noise component.

6. The method according to any one of claims 1 to 3, characterized in that: Decomposing the vibration signal to obtain multiple signal components includes: performing adaptive noise complete set empirical mode decomposition on the vibration signal to determine multiple signal components.

7. The method according to claim 1, characterized in that The signal component is a signal component based on an intrinsic mode function.

8. A signal denoising device, characterized in that: The device comprises: A vibration signal acquisition module, used to acquire the vibration signal of the mechanical equipment collected by the sensor; A signal decomposition module, used for decomposing the vibration signal to obtain multiple signal components; An energy entropy classification module, used to calculate the energy entropies of the multiple signal components respectively, and determine the noise component and the non-noise component from the multiple signal components according to the energy entropy; A denoising processing module, used for performing denoising processing on the noise component to obtain a processed noise component; The signal reconstruction module is used to reconstruct the processed noise component and non-noise component to obtain a denoised signal.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program implements the method according to any one of claims 1 to 7 when executed by the processor.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images