Signal processing method, device, equipment, medium and product

Abstract

The invention discloses a signal processing method, device and equipment, a medium and a product, and belongs to the technical field of signal processing. An original vibration signal and a multi-component signal model corresponding to the original vibration signal are obtained. According to a discretized scale basis function and a window function corresponding to an original vibration signal, a first time frequency characterization result of each frequency component is determined, on this basis, denoising processing is performed on the first time frequency characterization result based on an instantaneous frequency of the first time frequency characterization result, false components in a noise environment are effectively removed, and on this basis, the noise of the first time frequency characterization result is reduced. And according to the second time frequency characterization result, determining a target instantaneous frequency spectrum of the frequency component at each time center, and obtaining a target time frequency characterization result by fusing the target instantaneous frequency spectrums at each time center, thereby improving the readability of the time frequency characterization result, and improving the accuracy of the time frequency characterization result. Therefore, the fault characteristic frequency of the equipment and the rule of the fault characteristic frequency changing along with time can be identified more easily, and the accuracy of a fault diagnosis result is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of signal processing, and particularly relates to a signal processing method, device, equipment, medium and product. BACKGROUND

[0002] In modern engineering and scientific research, the application of signal processing technology is more and more widely, especially in the fields of vibration analysis, fault diagnosis, medical imaging, speech recognition, etc. With the increase of the complexity of equipment and system, the traditional signal processing method faces many challenges in processing non-stationary signals. The characteristics of non-stationary signals are that the frequency and amplitude change with time, which makes it difficult for traditional time domain or frequency domain analysis methods to effectively capture the instantaneous characteristics of the signal.

[0003] As a technology that represents a signal in both time and frequency domains, time-frequency analysis can more comprehensively describe the characteristics of the signal. Commonly used time-frequency analysis methods include short-time Fourier transform, wavelet transform, etc. Although these methods solve the problem of non-stationary signal analysis to some extent, when facing high noise and complex signals, it is difficult to distinguish the signal characteristics and false components in the noise environment, resulting in poor readability of the final determined time-frequency representation result of the signal, and further affecting the accuracy of the subsequent analysis result. SUMMARY

[0004] The purpose of the embodiments of the present application is to provide a signal processing method, device, equipment, medium and product, which can accurately identify the time-frequency characteristics of the vibration signal, and provide a reference for fault diagnosis and state monitoring of the equipment.

[0005] In a first aspect, the embodiments of the present application provide a signal processing method, comprising:

[0006] obtaining an original vibration signal and a multi-component signal model corresponding to the original vibration signal, the multi-component signal model comprising a plurality of frequency components, each frequency component having a corresponding amplitude and frequency;

[0007] determining a first time-frequency representation result corresponding to the frequency component according to the discretized scale basis function and a window function corresponding to the original vibration signal, the window function being used to traverse the frequency component in the time domain;

[0008] performing denoising processing on the first time-frequency representation result according to the instantaneous frequency of the first time-frequency representation result, to obtain a second time-frequency representation result corresponding to the frequency component;

[0009] determining a target instantaneous frequency spectrum of the frequency component at each time center according to the second time-frequency representation result;

[0010] fusing the target instantaneous frequency spectrum of the frequency component at each time center to obtain a target time-frequency representation result corresponding to the original vibration signal.

[0011] In a second aspect, an embodiment of the present application provides a signal processing apparatus, comprising:

[0012] An acquisition module is configured to acquire an original vibration signal and a multi-component signal model corresponding to the original vibration signal, the multi-component signal model comprising a plurality of frequency components, each frequency component having a corresponding amplitude and frequency;

[0013] A determination module is configured to determine, according to the discretized scale basis function and a window function corresponding to the original vibration signal, a first time-frequency representation result corresponding to the frequency component, the window function being used to traverse the frequency component in a time domain direction;

[0014] A processing module is configured to perform denoising processing on the first time-frequency representation result according to an instantaneous frequency of the first time-frequency representation result, to obtain a second time-frequency representation result corresponding to the frequency component;

[0015] The determination module is further configured to determine, according to the second time-frequency representation result, a target instantaneous frequency spectrum of the frequency component at each time center;

[0016] A fusion module is configured to fuse the target instantaneous frequency spectrum of the frequency component at each time center, to obtain a target time-frequency representation result corresponding to the original vibration signal.

[0017] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor and a memory, the memory storing programs or instructions executable on the processor, the programs or instructions being executed by the processor to implement the steps of the method according to the first aspect.

[0018] In a fourth aspect, an embodiment of the present application provides a readable storage medium, the readable storage medium storing programs or instructions, the programs or instructions being executed by the processor to implement the steps of the method according to the first aspect.

[0019] In a fifth aspect, an embodiment of the present application provides a computer program product, the program product being stored in a storage medium, the program product being executed by at least one processor to implement the steps of the method according to the first aspect.

[0020] In the embodiment of the present application, the original vibration signal and a multi-component signal model corresponding to the original vibration signal are obtained, wherein the multi-component signal model includes a plurality of frequency components, each frequency component has a corresponding amplitude and frequency. According to the discretized scale basis function and the window function corresponding to the original vibration signal, the first time-frequency representation result of each frequency component is determined, and on this basis, the first time-frequency representation result is denoised based on the instantaneous frequency of the first time-frequency representation result, effectively removing the false components in the noise environment. On this basis, the target instantaneous frequency spectrum of the frequency component at each time center is determined according to the second time-frequency representation result, and the target time-frequency representation result is obtained by fusing the target instantaneous frequency spectrum at each time center, which improves the readability of the time-frequency representation result, so that the fault characteristic frequency of the device and its change rule over time can be more easily identified, and the accuracy of the fault diagnosis result is improved. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A flowchart of a signal processing method provided by the embodiment of the present application;

[0022] Figure 2 A flowchart of another signal processing method provided by the embodiment of the present application;

[0023] Figure 3 A flowchart of another signal processing method provided by the embodiment of the present application;

[0024] Figure 4 A schematic diagram of a vibration signal with added noise provided by the embodiment of the present application;

[0025] Figure 5 A schematic diagram of an ideal time-frequency representation result of an original vibration signal provided by the embodiment of the present application;

[0026] Figure 6 A schematic diagram of a time-frequency representation result obtained by analyzing the original vibration signal by the embodiment of the present application;

[0027] Figure 7 A schematic diagram of a time-frequency representation result obtained by analyzing the original vibration signal by the related art;

[0028] Figure 8 A structural schematic diagram of a signal processing device provided by the embodiment of the present application;

[0029] Figure 9 A structural schematic diagram of an electronic device provided by the embodiment of the present application. DETAILED DESCRIPTION

[0030] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0031] The terms "first," "second," etc., used in this application's specification are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, in the specification, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.

[0032] As mentioned above, when faced with high noise and complex signals, the currently used time-frequency analysis method is difficult to distinguish between signal characteristics and spurious components in the noisy environment, resulting in poor readability of the final time-frequency characterization results of the signal, which in turn affects the accuracy of subsequent analysis results.

[0033] Therefore, embodiments of this application provide a signal processing method, apparatus, device, medium, and product that can accurately identify the time-frequency characteristics of vibration signals, providing a reference for equipment fault diagnosis and condition monitoring.

[0034] The signal processing methods, apparatus, devices, media, and products provided in this application will be described below with reference to the accompanying drawings and specific embodiments.

[0035] Figure 1 The flowchart illustrates a signal processing method provided in this application embodiment. This signal processing method can be applied to devices with data processing capabilities, such as laptops, tablets, desktops, industrial computers, servers, or controllers of mechanical equipment.

[0036] like Figure 1 As shown, the signal processing method may include the following steps:

[0037] S110. Obtain the original vibration signal and the multi-component signal model corresponding to the original vibration signal.

[0038] The multi-component signal model includes multiple frequency components, each with a corresponding amplitude and frequency.

[0039] S120. Based on the discretized scale basis function and the window function corresponding to the original vibration signal, determine the first time-frequency characterization result corresponding to the frequency component.

[0040] wherein the window function is used to traverse the frequency components in the time domain direction.

[0041] S130, according to the instantaneous frequency of the first time-frequency representation result, denoising processing is performed on the first time-frequency representation result to obtain a second time-frequency representation result corresponding to the frequency component.

[0042] S140, according to the second time-frequency representation result, determining a target instantaneous frequency spectrum of the frequency component at each time center.

[0043] S150, fusing the target instantaneous frequency spectrum of the frequency component at each time center to obtain a target time-frequency representation result corresponding to the original vibration signal.

[0044] In the embodiments of the present application, the original vibration signal and a multi-component signal model corresponding to the original vibration signal are obtained, wherein the multi-component signal model includes a plurality of frequency components, each frequency component has a corresponding amplitude and frequency. According to the discretized scale basis function and the window function corresponding to the original vibration signal, the first time-frequency representation result of each frequency component is determined first, and on this basis, the first time-frequency representation result is denoised based on the instantaneous frequency of the first time-frequency representation result, effectively removing the false components in the noise environment. On this basis, according to the second time-frequency representation result, the target instantaneous frequency spectrum of the frequency component at each time center is determined, and by fusing the target instantaneous frequency spectrum at each time center, the target time-frequency representation result is obtained, which improves the readability of the time-frequency representation result, so that the fault characteristic frequency of the device and its time-varying law can be more easily identified, and the accuracy of the fault diagnosis result is improved.

[0045] The above steps are described in detail as follows:

[0046] In S110, the original vibration signal can be a vibration signal of a target mechanical device collected according to a certain sampling frequency within a certain sampling time. The frequency and the sampling time can be set according to actual needs. For example, in some embodiments, the sampling frequency can be set to 200Hz, and the preset time period is 2s.

[0047] wherein the length of the original vibration signal can be determined based on the sampling frequency and the preset time period, for example, N=F s *T s , N represents the length of the original vibration signal, F s represents the sampling frequency, and T s represents the sampling time.

[0048] The target mechanical equipment may be, for example, a mechanical equipment applied to a complex industrial field such as national defense, energy, aerospace, etc., or a mechanical equipment applied to a field other than the above-mentioned complex industrial field. For example, in some embodiments, the target mechanical equipment may be a rotating machine such as a water pump, a generator, a gear, etc.

[0049] The embodiment does not limit the collection manner of the original vibration signal, which may be collected by a vibration sensor or other equipment, for example.

[0050] The multi-component signal model is a signal model corresponding to the original vibration signal, which is composed of a plurality of frequency components, and each frequency component is modulated by amplitude and frequency, that is, each frequency component has a corresponding amplitude and frequency.

[0051] Exemplarily, the original vibration signal may be subjected to Hilbert transform to convert a real value signal into a complex value signal, thereby obtaining an analytic signal of the original vibration signal.

[0052] Exemplarily, the analytic signal may be described as the following formula (1):

[0053]

[0054] wherein s ′ (t) is the analytic signal, x(t) is the original vibration signal, x(t) ∈ L 2 (R), is the imaginary unit.

[0055] Based on the analytic signal, a corresponding multi-component signal model may be obtained, which may be specifically referred to formula (2).

[0056]

[0057] wherein s(t) represents the multi-component signal model, k represents the number of frequency components of the original vibration signal, A i (t) and f i (t) represent the amplitude and instantaneous frequency of the i-th frequency component, respectively, and exp(·) represents the exponential function.

[0058] Each frequency component may be expressed by the following formula (3):

[0059] s i (t) = A i (t) · exp(j2π∫0 t f i (t)dt) (3)

[0060] In S120, the scale basis function is a basis function related to time and frequency, which helps to obtain high resolution and good readability of time-frequency representation.

[0061] Exemplarily, the scale basis function can be represented by the following formula (4):

[0062]

[0063] wherein Base(t) represents the scale basis function, t c represents the time center, f c represents the frequency center, and θ is the discrete angle of the scale basis function, and m0 is the parameter for weakening the nonlinearity of the tangent function.

[0064] Exemplarily, θ can be discretized in the range of (-π / 2, π / 2), which can be described as the following formula (5):

[0065]

[0066] wherein N n represents the number of equal-angle discretization of θ in (-π / 2, π / 2).

[0067] The discretized scale basis function can be obtained by substituting formula (5) into formula (4).

[0068] The window function here is used to traverse the frequency component in the time domain direction, and exemplarily, the window function can be a Gaussian window function. The width and moving step of the window function can be set according to actual needs. The window function can traverse the complete frequency component by moving in the time domain direction.

[0069] The first time-frequency representation result is the time-frequency representation result corresponding to each frequency component obtained initially, and the first time-frequency representation result is a high-dimensional time-frequency matrix.

[0070] Exemplarily, for each frequency component, the window function can be used to traverse the frequency component, and the traversal result can be matched with the discretized scale basis function, so as to obtain the first time-frequency representation result corresponding to the frequency component. Thus, higher resolution can be achieved in time-frequency analysis, and the instantaneous frequency change of the signal can be captured more clearly. Especially when dealing with complex non-stationary signals, similar frequency components can be effectively distinguished, so that the signal characteristics can be more accurately identified.

[0071] In S130, the instantaneous frequency of the first time-frequency representation result can be obtained by taking the partial derivative of time of the first time-frequency representation result.

[0072] Based on the instantaneous frequency of the first time-frequency representation result, the noise in the first time-frequency representation result can be effectively removed, which helps to enhance the accuracy of signal analysis and the readability of the time-frequency representation result.

[0073] Exemplarily, a noise processing operator can be constructed based on the instantaneous frequency, and the noise in the first time-frequency representation result can be removed based on the noise processing operator to obtain a second time-frequency representation result.

[0074] In S140, the instantaneous spectrum is a frequency distribution result corresponding to each time center of the second time-frequency representation result. The time center can be determined based on the sampling frequency and the sampling time of the original vibration signal. Exemplarily, the original vibration signal can be discretized to obtain a discrete signal, and the length of the discrete signal is equal to the product of the sampling frequency and the sampling time. Within the length of the discrete signal, each discrete point corresponds to a time center point.

[0075] In actual application, each time center can correspond to multiple discrete angles of a scale basis function, and each discrete angle can correspond to an instantaneous spectrum. The target instantaneous spectrum is one or more of the instantaneous spectra. Exemplarily, the target instantaneous spectrum can be selected from the instantaneous spectra based on statistical characteristics of the instantaneous spectra. The statistical characteristics may, for example, include the kurtosis value of the instantaneous spectrum.

[0076] Thus, for each time center of a certain frequency component, a corresponding target instantaneous spectrum can be obtained, which provides a basis for subsequent determination of the time-frequency representation result corresponding to each frequency component.

[0077] Considering that there is a certain multiple relationship between different frequency components of the same vibration signal, in actual application, the target instantaneous spectrum of only one frequency component at each time center can be determined, and the target instantaneous spectrum of other frequency components can be directly obtained based on this frequency component, thereby reducing the amount of calculation and improving the processing efficiency of the signal.

[0078] In S150, the target time-frequency representation result corresponding to the original vibration signal can be generated from the time-frequency representation results corresponding to each frequency component.

[0079] The time-frequency representation result corresponding to each frequency component can be obtained by fusing the target instantaneous spectrum of the frequency component at each time center.

[0080] Exemplarily, for each frequency component, the instantaneous spectrum information of each time center and the time points near the time center can be calculated, integrated, refined, or transformed, etc. to obtain the time-frequency representation result corresponding to the frequency component. The time-frequency representation result is a two-dimensional time-frequency graph.

[0081] The embodiment adopts scale basis function and window function to map the signal from time domain to time-frequency domain, obtain high-dimensional time-frequency representation result, and improve the resolution of the signal. On this basis, the false components in the signal are effectively removed by using the instantaneous frequency of the high-dimensional time-frequency representation result to denoise the high-dimensional time-frequency representation result, and the authenticity and reliability of the signal feature are improved. On this basis, the target instantaneous spectrum of each frequency component at each time center is determined based on the denoised high-dimensional time-frequency representation result, and the target time-frequency representation result of the original vibration signal is obtained by fusing each target instantaneous spectrum, which improves the readability of the target time-frequency representation result and can clearly show the time-varying characteristics of the signal, which is helpful for quickly and accurately diagnosing the equipment.

[0082] Taking the Gaussian window function as an example of the above window function, the above 120 can include the following steps:

[0083] The frequency component is traversed by using the Gaussian window function to obtain a window function traversal result;

[0084] The discrete scale basis function is matched with the window function traversal result to obtain a matching result;

[0085] The matching result is subjected to Fourier transform to obtain a first time-frequency representation result corresponding to the frequency component.

[0086] The Gaussian window function can be expressed as the following formula (6) for example:

[0087] g(t)=exp(-(t / σ) 2 / 2) (6)

[0088] Wherein, g(t) represents the Gaussian window function, σ represents the window width factor, and for example,

[0089] The traversal process of the Gaussian window function can be represented by the following formula (7):

[0090] Mid(t)=s i (t)g(t-t c ) (7)

[0091] Wherein, Mid(t) represents the window function traversal result obtained by traversing the i-th frequency component of the Gaussian window function.

[0092] The discrete scale basis function is matched with the window function traversal result, and the matching result is subjected to Fourier transform to map the result from time domain to time-frequency domain to obtain the first time-frequency representation result. For details, see the following formula (8):

[0093]

[0094] Wherein, Base′ (t) represents a discretized scale basis function, SBCT x (f c ,t c ,θ) represents a first time-frequency representation result.

[0095] The embodiment utilizes the discretized scale basis function and the sliding window function to analyze each frequency component of the original vibration signal through a time-frequency analysis method to obtain a first time-frequency representation result corresponding to each frequency component, thereby improving the resolution of the signal and helping to clearly capture the instantaneous frequency change of the signal.

[0096] Figure 2 The flowchart of another signal processing method provided by the embodiment of the present application is as follows, Figure 2 which is different from Figure 1 in that, Figure 1 S130 in Figure 2 may be refined as S210-S230 in

[0097] S210, determining an ideal frequency center of the frequency component according to the instantaneous frequency and the first time-frequency representation result.

[0098] The instantaneous frequency of the first time-frequency representation result can be obtained by solving the partial derivative of time of the first time-frequency representation result, and can be specifically represented by the following formula (9):

[0099]

[0100] wherein, represents the partial derivative of time of the first time-frequency representation result, that is, the instantaneous frequency of the first time-frequency representation result.(g) ′ represents the derivative of the window function.

[0101] The ideal frequency center of the frequency component can be constructed based on the instantaneous frequency and the first time-frequency representation result, and can be specifically referred to the following formula (10):

[0102]

[0103] wherein, f r (f c ,t c ,θ) represents the ideal frequency center.

[0104] S220, generating a noise processing operator according to the ideal frequency center and the actual frequency center of the frequency component in combination with the Dirac function.

[0105] The actual frequency center of the frequency component is f cBased on the ideal frequency center and the actual frequency center of the frequency component, a noise processing operator can be constructed to remove false components in the first time-frequency representation result, and improve the authenticity and reliability of the signal feature.

[0106] Exemplarily, based on the ideal frequency center and the actual frequency center of the frequency component, the noise processing operator can be constructed in combination with the Dirac function and the following formula (11):

[0107]

[0108] wherein SSEO x (f c ,t c ,θ) represents the noise processing operator, and δ(·) represents the Dirac function. That is, when f c =f r (f c ,t c ,θ), SSEO x (f c ,t c ,θ) = 1, otherwise, SSEO x (f c ,t c ,θ) = 0, that is, the noise processing operator is a high-dimensional matrix composed of 0 and 1.

[0109] In some embodiments, considering that there is an estimation error between the ideal frequency center and the actual frequency center of the signal, in order to reduce the influence of the estimation error on the noise processing operator, the real part of j· may be operated, and based on the operation result of the real part, the noise processing operator is determined, which can be specifically referred to the following formula (12):

[0110]

[0111] wherein ε is a threshold parameter, that is, when the real part of j· is less than ε, it is considered that f c =f r (f c ,t c ,θ), otherwise, it is considered that f c ≠f r (f c ,t c ,θ). Re(·) represents the real part operation.

[0112] S230, according to the noise processing operator, the first time-frequency representation result is de-noised to obtain a second time-frequency representation result.

[0113] Exemplarily, the de-noising processing can be implemented by the following formula (13):

[0114] SSECT x (f c ,t c ,θ)=SBCT x (f c ,t c ,θ)·SSEO x (f c ,t c ,θ) (13)

[0115] The embodiment of the present application constructs an ideal frequency center of the frequency component based on the instantaneous frequency and the first time-frequency representation result, constructs a noise processing operator based on the ideal frequency center and the actual frequency center, and removes false components in the first time-frequency representation result and retains effective components in the signal by constructing the noise processing operator, thereby improving the authenticity and reliability of the signal features.

[0116] Figure 3 The flowchart of another signal processing method provided by the embodiment of the present application, Figure 3 and Figure 1 the difference is that, Figure 1 S140 in the embodiment of the present application can be refined as S310-S340 in the embodiment of the present application. Figure 3

[0117] S310, at each time center, determining an instantaneous frequency spectrum corresponding to each discrete angle according to the second time-frequency representation result.

[0118] The discrete angle is a discrete angle corresponding to the scale basis function. Each frequency component can correspond to multiple discrete angles of the scale basis function at each time center. For example, if the scale basis function is discretized into 70 angles, each frequency component can correspond to 70 discrete angles at each time center.

[0119] As known from the above embodiment, the second time-frequency representation result involves three dimensions, namely time, frequency and discrete angle. Each discrete angle corresponding to each time center, that is, when the time and the discrete angle are fixed, the frequency distribution result of the frequency component at the time center at the discrete angle can be obtained based on the second time-frequency representation result, and then the instantaneous frequency spectrum of the frequency component at the time center at the discrete angle can be obtained. Thus, for each discrete angle of each time center, a corresponding instantaneous frequency spectrum can be obtained.

[0120] S320, determining the kurtosis value of each instantaneous frequency spectrum.

[0121] ​It can be understood that, at each time center, when the scale basis function is optimally matched with each frequency component, the instantaneous spectrum of the frequency component has the most prominent and obvious amplitude, and therefore, based on this, the embodiments of the present application can calculate the kurtosis value of the instantaneous spectrum corresponding to each discrete angle at each time center, and determine the optimal matching angle based on the kurtosis value, so that a suitable instantaneous spectrum can be selected based on the optimal matching angle, and the readability of the time-frequency representation result is improved.

[0122] In some embodiments, a first average value of the fourth power of the amplitude of the instantaneous spectrum in a preset frequency interval and a second average value of the square of the amplitude of the instantaneous spectrum in the preset frequency interval can be determined; and the kurtosis value of the instantaneous spectrum is determined according to the first average value and the second average value.

[0123] Exemplarily, the kurtosis value of the instantaneous spectrum can be determined by the following formula (14):

[0124]

[0125] wherein Kur(SSECT x (f c ,t c ,θ)) represents the kurtosis value of the instantaneous spectrum, represents the first average value, represents the second average value, and (0, Ω) represents the preset frequency interval. Generally, Ω=F S / 2.

[0126] Based on the kurtosis value, the discrete angle of the scale basis function is screened, and the corresponding instantaneous spectrum is determined based on the screening result, so that the instantaneous frequency change of the signal can be accurately captured, and the readability of the time-frequency representation result is improved.

[0127] S330, determining a target discrete angle from the discrete angles according to the kurtosis value.

[0128] The kurtosis value corresponding to the target discrete angle is greater than or equal to a preset threshold. The size of the preset threshold can be flexibly determined based on the kurtosis value corresponding to each discrete angle, for example, in some embodiments, the preset threshold can be a threshold that can ensure that the selected kurtosis value is the maximum value among the kurtosis values. That is, the discrete angle with the maximum kurtosis value can be determined as the target discrete angle, so that the instantaneous frequency change of each frequency component at each time center can be accurately captured, and the signal feature can be more accurately identified.

[0129] Exemplarily, the target discrete angle can be represented by the following formula (15):

[0130] θ opt =argmax(Kur(SSECT x (fc t c ,θ))) (15)

[0131] wherein, θ opt represents a target discrete angle.

[0132] S340, the instantaneous spectrum corresponding to the target discrete angle is determined as the target instantaneous spectrum of the frequency component at the time center.

[0133] After the target discrete angle is determined, the instantaneous spectrum corresponding to the target discrete angle can be determined as the target instantaneous spectrum of the frequency component at the corresponding time center.

[0134] The embodiments of the present application evaluate the second time-frequency representation result after denoising by using kurtosis theory, obtain the optimal discrete angle corresponding to the scale basis function at each time center, and further obtain the corresponding target instantaneous spectrum at each time center, which can more accurately capture the instantaneous frequency change of the signal and enhance the readability of the time-frequency representation result.

[0135] In some embodiments, the above S150 can include the following steps:

[0136] For each frequency component, the target instantaneous spectrum of the frequency component at each time center is fused to obtain the time-frequency representation result corresponding to the frequency component.

[0137] According to the time-frequency representation result corresponding to each frequency component, the target time-frequency representation result corresponding to the original vibration signal is generated.

[0138] The above embodiments obtain the target instantaneous spectrum at each time center. To obtain a complete time-frequency representation result, for example, for each frequency component, the target instantaneous spectrum of the frequency component at each time center can be fused to obtain the time-frequency representation result corresponding to the frequency component.

[0139] Other frequency components can perform similar operations, so that for each frequency component, the corresponding time-frequency representation result can be obtained.

[0140] Combining the time-frequency representation results corresponding to each frequency component included in the original vibration signal, the time-frequency representation result corresponding to the original vibration signal can be obtained.

[0141] The embodiments of the present application first fuse the instantaneous spectrum of each frequency component at each time center to obtain the time-frequency representation result corresponding to each frequency component, and then obtain the time-frequency representation result of the original vibration signal based on the time-frequency representation result of each frequency component, which improves the readability of the time-frequency representation result and can intuitively display the change of the frequency component of the signal with time, helping the analyst better understand the characteristics of the signal.

[0142] In some embodiments, the signal processing method can further include the following steps:

[0143] According to the target time-frequency characterization result, the device corresponding to the original vibration signal is diagnosed for fault to obtain a fault diagnosis result.

[0144] Exemplarily, the device can be diagnosed for fault based on whether there is a fault characteristic frequency in the target time-frequency characterization result. For example, if there is a fault characteristic frequency in the target time-frequency characterization result, it is considered that the device is faulty, otherwise, it is considered that the device is normal.

[0145] Since the accuracy and readability of the target time-frequency characterization result are both improved, the device can be accurately diagnosed for fault based on the target time-frequency characterization result, and the accuracy of the diagnosis result is improved.

[0146] The scheme of the present application will be described below through a simulation example.

[0147] Take a multi-component signal model of a certain original vibration signal described as the following formula (16) as an example:

[0148]

[0149] That is, the original vibration signal contains four frequency components. Among them, f t1 = -t 2 + 10t + 15, f t2 = 1.4f t1 , f t3 = 1.8f t1 , f t4 = 2.2f t1 , and ε represents the added Gaussian white noise. The schematic diagram of the signal added with noise is shown in Figure 4 . The ideal time-frequency characterization result of each frequency component of the original vibration signal can be seen in Figure 5 .

[0150] Suppose the sampling frequency of the original vibration signal is Fs = 200 Hz, the sampling time is 4s, the total sampling point number of the signal N = 800 is calculated, the length of the Gaussian window function is set to Nw = 200, the coverage length of the window function is Nv = 199, that is, the moving step of the window function is 1, the number of discrete angles corresponding to the scale basis function Nn = 70, and then the actual time-frequency characterization result of the original vibration signal determined according to the scheme of the above embodiment can be seen in Figure 6 .

[0151] Since noise is added, the waveform of the signal is relatively complex and it is difficult to directly analyze the frequency components. After processing by the time-frequency analysis scheme provided in the embodiments of the present application, as shown in Figure 6As shown, the changes of different frequency components can be clearly distinguished, the frequency characteristics of the signal can effectively remove false components and retain real frequency components, clear frequency components can be seen without noise interference, and the actual performance of these instantaneous frequencies is successfully extracted. Compared with the results obtained by the traditional scheme Figure 7 The embodiments of the present application can more accurately capture the instantaneous frequency changes of the signal, especially in the presence of noise in the signal, the time-frequency spectrum obtained provides higher readability, can intuitively show the changes of the frequency components of the signal over time, and can help the analyst better understand the characteristics of the signal.

[0152] Based on the same inventive concept, the embodiments of the present application also provide a signal processing device. Figure 8 A structural schematic diagram of a signal processing device provided by the embodiments of the present application.

[0153] As Figure 8 shown, the signal processing device 800 can include:

[0154] The acquisition module 801 is configured to acquire an original vibration signal and a multi-component signal model corresponding to the original vibration signal, the multi-component signal model including a plurality of frequency components, each frequency component having a corresponding amplitude and frequency;

[0155] The determination module 802 is configured to determine a first time-frequency representation result corresponding to the frequency component according to the discretized scale basis function and the window function corresponding to the original vibration signal, the window function being used to traverse the frequency component in the time domain direction;

[0156] The processing module 803 is configured to perform denoising processing on the first time-frequency representation result according to the instantaneous frequency of the first time-frequency representation result, to obtain a second time-frequency representation result corresponding to the frequency component;

[0157] The determination module 802 is further configured to determine a target instantaneous frequency spectrum of the frequency component at each time center according to the second time-frequency representation result;

[0158] The fusion module 804 is configured to fuse the target instantaneous frequency spectrum of the frequency component at each time center to obtain a target time-frequency representation result corresponding to the original vibration signal.

[0159] In the embodiment of the present application, the original vibration signal and a multi-component signal model corresponding to the original vibration signal are obtained, wherein the multi-component signal model includes a plurality of frequency components, each frequency component has a corresponding amplitude and frequency. According to the discretized scale basis function and the window function corresponding to the original vibration signal, the first time-frequency representation result of each frequency component is determined, and on this basis, the first time-frequency representation result is denoised based on the instantaneous frequency of the first time-frequency representation result, so that the false components in the noise environment are effectively removed. On this basis, according to the second time-frequency representation result, the target instantaneous frequency spectrum of the frequency component at each time center is determined, and the target time-frequency representation result is obtained by fusing the target instantaneous frequency spectrum at each time center, thereby improving the readability of the time-frequency representation result, so that the fault characteristic frequency of the device and its time-varying law can be more easily identified, and the accuracy of the fault diagnosis result is improved.

[0160] In some possible implementations of the embodiment of the present application, the determination module 802 is further configured to determine the ideal frequency center of the frequency component according to the instantaneous frequency and the first time-frequency representation result.

[0161] The processing module 803 is specifically configured to:

[0162] According to the ideal frequency center and the actual frequency center of the frequency component, a noise processing operator is generated in combination with the Dirac function;

[0163] The first time-frequency representation result is denoised according to the noise processing operator to obtain the second time-frequency representation result.

[0164] In some possible implementations of the embodiment of the present application, the determination module 802 is specifically configured to:

[0165] At each time center, the instantaneous frequency spectrum corresponding to each discrete angle is determined according to the second time-frequency representation result, and the discrete angle is the discrete angle corresponding to the scale basis function;

[0166] The kurtosis value of each instantaneous frequency spectrum is determined;

[0167] According to the kurtosis value, a target discrete angle is determined from the discrete angles, and the kurtosis value corresponding to the target discrete angle is greater than or equal to a preset threshold;

[0168] The instantaneous frequency spectrum corresponding to the target discrete angle is determined as the target instantaneous frequency spectrum of the frequency component at the time center.

[0169] In some possible implementations of the embodiment of the present application, the determination module 802 is specifically configured to:

[0170] The first average value of the fourth power of the amplitude of the instantaneous frequency spectrum in a preset frequency interval and the second average value of the square of the amplitude of the instantaneous frequency spectrum in the preset frequency interval are determined;

[0171] According to the first average value and the second average value, a kurtosis value of the instantaneous spectrum is determined.

[0172] In some possible implementation of the embodiments of the present application, the fusion module 804 is specifically configured to:

[0173] For each frequency component, the fusion module 804 fuses the target instantaneous spectrum of the frequency component at the time center to obtain a time-frequency representation result corresponding to the frequency component.

[0174] According to the time-frequency representation result corresponding to each frequency component, the fusion module 804 generates a target time-frequency representation result corresponding to the original vibration signal.

[0175] In some possible implementation of the embodiments of the present application, the window function includes a Gaussian window function.

[0176] The determination module 802 is specifically configured to:

[0177] The determination module 802 traverses the frequency components by using the Gaussian window function to obtain a window function traversal result.

[0178] The determination module 802 matches the discretized scale basis function with the window function traversal result to obtain a matching result.

[0179] The determination module 802 performs Fourier transform on the matching result to obtain a first time-frequency representation result corresponding to the frequency component.

[0180] In some possible implementation of the embodiments of the present application, the signal processing apparatus 800 can further include:

[0181] The diagnosis module is configured to perform fault diagnosis on the device corresponding to the original vibration signal according to the target time-frequency representation result to obtain a fault diagnosis result.

[0182] Based on the same inventive concept, the embodiments of the present application further provide an electronic device, which is described below in combination with Figure 9 The electronic device provided by the embodiments of the present application is described.

[0183] As shown in Figure 9 The electronic device 900 can include a processor 901 and a memory 902 configured to store computer program instructions.

[0184] The processor 901 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or can be configured as one or more integrated circuits implementing the embodiments of the present application.

[0185] The memory 902 can include mass storage for data or instructions. As an example and not by way of limitation, the memory 902 can include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disc (e.g., a compact disc (CD) or a digital versatile disc (DVD)), a solid-state drive (SSD), a USB drive, or a combination of two or more of these. In one example, the memory 902 can include removable or non-removable (or fixed) media, where the memory 902 is a nonvolatile solid-state memory in one example. In one example, the memory 902 can be a read-only memory (ROM). In one example, the ROM can be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically alterable ROM (EAROM), or a flash memory, or a combination of two or more of these.

[0186] The processor 901 implements the methods of the embodiments by reading and executing computer program instructions stored in the memory 902. Figures 1-3 The methods of the embodiments shown, and the corresponding technical effects achieved by the embodiments performing the methods, are not repeated here for brevity. Figures 1-3 The methods of the embodiments shown, and the corresponding technical effects achieved by the embodiments performing the methods, are not repeated here for brevity.

[0187] In one example, the electronic device 900 can also include a communication interface 903 and a bus 904. As shown, the processor 901, the memory 902, and the communication interface 903 are connected by the bus 904 and complete communication among each other. Figure 9

[0188] The communication interface 903 is mainly used to realize the communication between the modules, devices, and / or equipment in the embodiments of the present application.

[0189] ​Bus 904 includes a hardware, software, or both that couples components of electronic device 900 to each other. As an example and not by way of limitation, bus 904 can include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand™ interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association local (VLB) bus, or another suitable bus or a combination of two or more of these. Where suitable, bus 904 can include one or more buses. Although this application describes and illustrates a particular bus, this application contemplates any suitable bus or interconnect.

[0190] In addition, in combination with the signal processing method in the above embodiments, an application embodiment can provide a readable storage medium to implement. The readable storage medium stores program instructions; the program instructions are executed by a processor to implement any one of the signal processing methods in the above embodiments.

[0191] In addition, in combination with the signal processing method in the above embodiments, an application embodiment can provide a computer program product to implement. The program product is stored in a storage medium, and the program product is executed by at least one processor to implement any one of the signal processing methods in the above embodiments.

[0192] Although the present application has been described with reference to the preferred embodiments, various modifications can be made to the application without departing from the scope of the application. In particular, the technical features mentioned in each embodiment can be combined in any manner as long as there is no structural conflict. The present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A signal processing method, characterized by, The method comprises: acquiring an original vibration signal and a multi-component signal model corresponding to the original vibration signal, the multi-component signal model comprising a plurality of frequency components, each frequency component having a corresponding amplitude and frequency; determining a first time-frequency representation result corresponding to the frequency component according to a discretized scale basis function and a window function corresponding to the original vibration signal, the window function being used to traverse the frequency component in a time domain direction; performing denoising processing on the first time-frequency representation result according to an instantaneous frequency of the first time-frequency representation result, to obtain a second time-frequency representation result corresponding to the frequency component; determining a target instantaneous spectrum of the frequency component at each time center according to the second time-frequency representation result; fusing the target instantaneous spectra of the frequency component at the time centers to obtain a target time-frequency representation result corresponding to the original vibration signal.

2. The method of claim 1, wherein, The method further comprises: determining an ideal frequency center of the frequency component according to the instantaneous frequency and the first time-frequency representation result; generating a noise processing operator according to the ideal frequency center and an actual frequency center of the frequency component in combination with a Dirac function; performing denoising processing on the first time-frequency representation result according to the noise processing operator to obtain the second time-frequency representation result.

3. The method according to claim 1 or 2, characterized in that, The method further comprises: at each time center, determining an instantaneous spectrum corresponding to each discrete angle according to the second time-frequency representation result, the discrete angle being a discrete angle corresponding to the scale basis function; determining a kurtosis value of each instantaneous spectrum; determining a target discrete angle from the discrete angles according to the kurtosis values, the kurtosis value corresponding to the target discrete angle being greater than or equal to a preset threshold value; determining the instantaneous spectrum corresponding to the target discrete angle as the target instantaneous spectrum of the frequency component at the time center.

4. The method of claim 3, wherein, The method further comprises: determining a first average value of the fourth power of the amplitude of the instantaneous spectrum in a preset frequency interval and a second average value of the square of the amplitude of the instantaneous spectrum in the preset frequency interval; determining the kurtosis value of the instantaneous spectrum according to the first average value and the second average value.

5. The method according to claim 1 or 2, characterized in that, The method further comprises: for each frequency component, fusing the target instantaneous spectra of the frequency component at the time centers to obtain a time-frequency representation result corresponding to the frequency component; and generating the target time-frequency representation result corresponding to the original vibration signal according to the time-frequency representation results corresponding to the frequency components.

6. The method of claim 1 or 2, wherein, The window function comprises a Gaussian window function. The method further comprises: The Gaussian window function is used to traverse the frequency component, and a window function traversal result is obtained; The discrete scale basis function is matched with the window function traversal result, and a matching result is obtained; The matching result is subjected to Fourier transform, and a first time-frequency representation result corresponding to the frequency component is obtained.

7. The method according to claim 1 or 2, characterized in that, The method further includes: According to the target time-frequency representation result, a fault diagnosis is performed on the equipment corresponding to the original vibration signal, and a fault diagnosis result is obtained.

8. A signal processing device, characterized by It includes: An acquisition module is configured to acquire an original vibration signal and a multi-component signal model corresponding to the original vibration signal, the multi-component signal model including a plurality of frequency components, each frequency component having a corresponding amplitude and frequency; A determination module is configured to determine, according to a discrete scale basis function and a window function corresponding to the original vibration signal, a first time-frequency representation result corresponding to the frequency component, the window function being used to traverse the frequency component in a time domain direction; A processing module is configured to perform denoising processing on the first time-frequency representation result according to an instantaneous frequency of the first time-frequency representation result, and obtain a second time-frequency representation result corresponding to the frequency component; The determination module is further configured to determine, according to the second time-frequency representation result, a target instantaneous frequency spectrum of the frequency component at each time center; A fusion module is configured to fuse the target instantaneous frequency spectrum of the frequency component at each time center, and obtain a target time-frequency representation result corresponding to the original vibration signal.

9. An electronic device, comprising: The electronic device includes a processor and a memory, the memory storing programs or instructions executable on the processor, the programs or instructions being executed by the processor to implement the steps of the method according to any one of claims 1-7.

10. A readable storage medium, characterized by, The readable storage medium stores programs or instructions, which are executed by the processor to implement the steps of the method according to any one of claims 1-7.

11. A readable storage medium, characterized by, The readable storage medium stores programs or instructions, which are executed by the processor to implement the steps of the method according to any one of claims 1-7.

Images