Auto-correlation bidirectional enhancement integrated envelope method and system

Abstract

The invention discloses a self-correlation bidirectional enhancement integrated envelope method and system. The method comprises the following steps: firstly, acquiring a vibration signal of mechanical equipment and preprocessing the vibration signal to obtain a cyclic coherence spectrum of the vibration signal; then, calculating an autocorrelation function of the cyclic coherence spectrum, and adaptively determining weights of a frequency spectrum frequency direction and a cyclic frequency direction according to peak characteristics of the function after a zero crossing point; on the basis, an enhancement matrix is constructed in combination with the obtained bidirectional weight, weighted enhancement is performed on the original cyclic coherence spectrum, integration is performed along the frequency direction of the spectrum, and finally the self-correlation bidirectional enhancement integrated envelope is generated. According to the method, the periodic enhancement characteristic of the autocorrelation function is fully utilized, the in-band noise in the cyclic coherence spectrum can be effectively suppressed, and the fault characteristic frequency and harmonic waves thereof can be remarkably highlighted, so that the problem that the fault characteristics are difficult to extract under the strong noise background is solved, and the accuracy and reliability of mechanical fault diagnosis are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mechanical fault diagnosis, and particularly relates to a self-correlation bidirectional enhancement integrated envelope method and system. BACKGROUND

[0002] Rolling bearings are important components of key systems such as traction motors, gearboxes and bogies, and bear multiple functions such as supporting mechanical load, maintaining smooth operation and reducing friction loss. The performance of rolling bearings directly determines whether the entire equipment system can operate safely and reliably. Due to long-term operation in harsh environments such as high temperature, heavy load, impact vibration, dust pollution and frequent high-speed start-stop, rolling bearings are prone to problems such as loss of lubrication function, material fatigue damage and accelerated surface wear, thereby significantly increasing the probability of failure. Once a fault occurs, not only will it shorten the service life of the equipment, but it may also cause a chain reaction and pose a threat to safe operation. Therefore, it is of great practical significance to accurately monitor and early diagnose the health status of rolling bearings.

[0003] In the field of fault diagnosis, the commonly used signal processing methods include empirical mode decomposition, variational mode decomposition and characteristic mode decomposition adaptive mode decomposition techniques. Although these methods can partially extract fault information from nonlinear and non-stationary signals, they still have inherent problems such as mode mixing, end effect, basis function dependence and parameter setting sensitivity. In particular, it should be pointed out that such methods can be regarded as a group of adaptive band-pass filters, which are difficult to accurately separate the periodic components generated by faults and other interference signals in the same frequency band, and the extraction ability of fault features is significantly restricted in strong noise background.

[0004] As a second-order statistical method that can simultaneously present the frequency spectrum structure and periodicity of a signal, cyclic coherence spectrum is introduced into mechanical fault diagnosis to simultaneously identify the resonance frequency band and fault feature frequency. However, the traditional cyclic coherence spectrum still performs unsatisfactorily when dealing with complex noise interference in actual engineering, especially in the case of weak fault features and extremely low signal-to-noise ratio, the in-band noise will seriously mask the periodic components caused by faults, resulting in failure of feature extraction.

[0005] In the face of the above challenges, the integrated envelope analysis method can highlight the periodic components related to faults in a high-noise background by effectively integrating and enhancing the frequency domain features, thereby providing clearer and more reliable basis for fault judgment, and has become an important research direction for improving the diagnosis effect. However, the existing integrated envelope method still has deficiencies in noise suppression ability and periodic component enhancement effect. How to further improve the signal-to-noise ratio enhancement efficiency of fault feature frequency and effectively suppress non-related interference is still a technical problem that needs to be overcome in the current field of fault diagnosis. SUMMARY

[0006] To this end, the technical problem to be solved by the present application is to overcome the problems of insufficient signal-to-noise ratio suppression and cycle feature enhancement in the existing integrated envelope analysis technology.

[0007] To solve the above technical problems, the present application provides a self-correlation bidirectional enhancement integrated envelope method, comprising the following steps: S1: preprocessing the obtained vibration signal of the mechanical equipment to obtain a cyclic coherence spectrum of the vibration signal; S2: calculating a self-correlation function of the cyclic coherence spectrum; based on the peak value characteristics of the self-correlation function, determining the weight in the spectral frequency direction and the weight in the cycle frequency direction of the cyclic coherence spectrum; S3: based on the weight in the spectral frequency direction and the weight in the cycle frequency direction, performing enhancement processing on the cyclic coherence spectrum to obtain a self-correlation bidirectional enhancement integrated envelope.

[0008] In an embodiment of the present application, in step S1, the method for preprocessing the vibration signal to obtain the cyclic coherence spectrum of the vibration signal is: calculating a self-correlation function of the vibration signal, performing Fourier transform on the self-correlation function in two dimensions of time lag and cycle frequency, and constructing a correlation spectrum capable of representing both signal spectral characteristics and cycle characteristics , the expression formula is: , wherein, represents a periodic modulation in the signal; represents a spectral frequency; represents a Fourier transform of the vibration signal self-correlation function, represents the ordinal number of the midpoint; represents the cycle of cyclostationarity.

[0009] In an embodiment of the present application, the method for obtaining the cyclic coherence spectrum of the vibration signal further comprises: performing normalization processing on the correlation spectrum capable of representing both signal spectral characteristics and cycle characteristics to obtain the cyclic correlation spectrum , the expression formula is: , wherein, represents the value at .

[0010] In an embodiment of the present application, in step S2, the method for calculating the self-correlation function of the cyclic coherence spectrum is: calculating the self-correlation function corresponding to all spectral frequencies of the cyclic coherence spectrum, the expression formula is: ,​ wherein, denotes the cycle frequency lag, denotes the autocorrelation calculation, denotes the cycle coherence spectrum, denotes the periodic modulation in the signal; denotes the spectral frequency.

[0011] In one embodiment of the application, in step S2, the method for determining the weight of the cycle coherence spectrum in the spectral frequency direction based on the peak characteristics of the autocorrelation function is: based on the first maximum peak after the zero-crossing point corresponding to the fault characteristic frequency, storing the peak values at the positions of the first maximum peak after the zero-crossing point and its multiples into a matrix, obtaining a peak characteristics matrix : , wherein, denotes the cycle frequency lag corresponding to the first maximum peak after the zero-crossing point, denotes the cycle coherence spectrum spectral frequency sequence number, denotes the cycle frequency sequence number, denotes the first spectral frequency, denotes the first cycle frequency lag, denotes the amplitude of the autocorrelation function at the position of the first maximum peak after the zero-crossing point; obtaining the weight of the spectral frequency based on the peak characteristics matrix : .

[0012] In one embodiment of the application, in step S2, the method for determining the weight of the cycle coherence spectrum in the spectral frequency direction based on the peak characteristics of the autocorrelation function is: based on the periodicity of the vibration signal, constructing a cycle matrix by extending the maximum value to all multiples of the fault characteristic frequency : , wherein, denotes the multiple, denotes the cycle frequency lag corresponding to the first maximum peak after the zero-crossing point, denotes the cycle coherence spectrum spectral frequency sequence number, denotes the cycle frequency sequence number, denotes the first spectral frequency, denotes the first cycle frequency lag, denotes the amplitude of the autocorrelation function at the position of the first maximum peak after the zero-crossing point; The weight of the cyclic frequency is obtained by adding the autocorrelation peak values of all spectral frequencies and performing an exponential operation wherein denotes the exponential operation, denotes the summation over all spectral frequency ranges; Based on the periodicity in the direction of the cyclic frequency, the cyclic frequency lag corresponds to the cyclic frequency , i.e. let , the cyclic frequency weight is obtained: wherein is the spectral frequency, denotes the summation over all spectral frequency ranges.

[0013] In an embodiment of the application, in step S3, the method for obtaining the autocorrelation bidirectional enhancement integrated envelope is: combining the frequency domain and cyclic domain weights to obtain a bidirectional enhancement matrix : denotes the weight of the spectral frequency based on the peak feature matrix, denotes the cyclic frequency weight; The bidirectional enhancement matrix is multiplied by the original cyclic coherence spectrum to obtain an autocorrelation enhancement cyclic coherence spectrum : wherein denotes the autocorrelation function corresponding to all spectral frequency pairs of the cyclic coherence spectrum; The autocorrelation bidirectional enhancement integrated envelope is calculated by summing the spectral frequency directions of the autocorrelation enhancement cyclic coherence spectrum : wherein denotes the sampling frequency.

[0014] In an embodiment of the application, after obtaining the autocorrelation bidirectional enhancement integrated envelope, the mechanical equipment fault diagnosis is performed by identifying the spectral peak corresponding to the fault feature frequency in the autocorrelation bidirectional enhancement integrated envelope.

[0015] Based on the same inventive concept, the application further provides an autocorrelation bidirectional enhancement integrated envelope system, comprising the following modules: ​​​​​​The signal acquisition and preprocessing module is configured to preprocess the acquired vibration signal of the mechanical equipment to obtain a cyclic coherence spectrum of the vibration signal. The autocorrelation analysis and weight determination module is configured to calculate an autocorrelation function of the cyclic coherence spectrum, and determine a weight in a spectral frequency direction and a weight in a cycle frequency direction of the cyclic coherence spectrum based on a peak value feature of the autocorrelation function. The spectrum enhancement and envelope generation module is configured to perform enhancement processing on the cyclic coherence spectrum based on the weight in the spectral frequency direction and the weight in the cycle frequency direction to obtain an autocorrelation bidirectional enhancement integrated envelope.

[0016] The application further provides a computer storage medium storing a computer software product, and the computer software product comprises a plurality of instructions for enabling a computer device to execute the autocorrelation bidirectional enhancement integrated envelope method.

[0017] The above technical solution of the application has the following advantages compared with the prior art. The application introduces an autocorrelation function to perform bidirectional enhancement on the cyclic coherence spectrum, and the core mechanism is to adaptively determine bidirectional weights by using a peak value feature, so that the periodic components of the fault are effectively highlighted and the interference is suppressed in strong noise, and the weak fault is clearly extracted. The method further constructs a weight matrix by using the zero-crossing peak value and its multiples, and combines exponential normalization to enhance the capture of fault harmonics, overcoming the shortcomings of traditional methods. This weight generation method based on signal statistical characteristics reduces the dependence on artificial parameters, and enhances the robustness and engineering practicability. Finally, the information fusion in the frequency domain and the cycle domain is realized by constructing a bidirectional enhancement matrix, and a one-dimensional envelope spectrum with concentrated features is obtained through integrated dimension reduction. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to make the content of the application more easily understood, the application will be further described in detail below according to specific embodiments of the application and in conjunction with the drawings.

[0019] Figure 1 is a flowchart of an autocorrelation bidirectional enhancement integrated envelope method provided by an embodiment of the application; Figure 2 is a schematic diagram of the vibration signal in experiment 1; Figure 3 is a schematic diagram of the coherence spectrum of the vibration signal in experiment 1; Figure 4 is a schematic diagram of the enhanced coherence spectrum of the vibration signal in experiment 1; Figure 5 is an enhanced envelope spectrum obtained by the autocorrelation bidirectional enhancement integrated envelope method in experiment 1; Figure 6This is a schematic diagram of the traditional envelope spectrum of the vibration signal in Experiment 1; Figure 7 This is a schematic diagram of the structure of an autocorrelation bidirectional enhanced integrated envelope system provided in an embodiment of the present invention. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0021] Example 1: like Figure 1 As shown, the present invention provides an autocorrelation bidirectional enhanced integrated envelope method, comprising the following steps: S1: Preprocess the acquired vibration signal of the mechanical equipment to obtain the cyclic coherence spectrum of the vibration signal; S2: Calculate the autocorrelation function of the cyclic coherence spectrum; based on the peak characteristics of the autocorrelation function, determine the weights of the cyclic coherence spectrum in the spectral frequency direction and the weights in the cyclic frequency direction; S3: The cyclic coherence spectrum is enhanced based on the weights in the frequency direction and the cyclic frequency direction to obtain an autocorrelation bidirectional enhanced integrated envelope.

[0022] The autocorrelation-based bidirectional enhancement integrated envelope method provided by this invention takes a mechanical vibration signal as input. First, it preprocesses the signal to obtain a cyclic coherence spectrum that simultaneously characterizes the signal's spectral and periodic properties. Based on this, the method further calculates the autocorrelation function of the cyclic coherence spectrum and adaptively assigns weights to the spectral frequency and cyclic frequency directions according to the peak characteristics after the zero-crossing point. Finally, it uses these weights to construct a bidirectional enhancement matrix to enhance the original spectrum and generates an integrated envelope spectrum through integration along the spectral frequency direction, thereby achieving effective highlighting and reliable diagnosis of fault characteristic frequencies in a strong noise background.

[0023] Specifically, in step S1, the vibration signal is preprocessed to obtain a cyclic coherence spectrum. This process aims to transform the original time-domain vibration signal into a two-dimensional representation that can simultaneously characterize the signal's spectral and periodic characteristics, laying the foundation for subsequent feature enhancement and extraction.

[0024] The vibration signals acquired from mechanical equipment typically contain rich dynamic information, but are also mixed with noise interference. The preprocessing stage includes necessary signal conditioning steps to improve the quality of the vibration signals.

[0025] The core processing link aims to construct a correlation spectrum capable of revealing the cyclostationary characteristics of the signal. The specific implementation method is: first, calculate the autocorrelation function of the vibration signal to obtain its statistical correlation at each time lag; then, perform Fourier transform on the autocorrelation function in both time lag and cycle frequency dimensions.

[0026] This step maps the time-domain autocorrelation function to a two-dimensional frequency domain space jointly spanned by the spectral frequency and the cycle frequency , thereby constructing a correlation spectrum that can simultaneously represent the spectral characteristics and periodic characteristics of the signal. The mathematical expression of the correlation spectrum is: , wherein, represents the cycle frequency corresponding to the periodic modulation component in the signal, which is a key variable for identifying fault features and their harmonics; represents the spectral frequency, reflecting the resonance frequency band or inherent spectral structure of the signal; represents the Fourier transform of the autocorrelation function of the vibration signal, represents the ordinal number of the midpoint, represents the period of cyclostationarity.

[0027] This step extracts the periodic component in the signal that is synchronized with a specific cycle , effectively filtering non-periodic random noise and irrelevant periodic interference.

[0028] Further, in order to make the amplitudes of different frequency components comparable and further highlight the periodic modulation strength, after obtaining the correlation spectrum , it also needs to be normalized to obtain the cyclic coherence spectrum , whose calculation formula is: , wherein, represents the value at .

[0029] The normalized cyclic coherence spectrum has its value domain constrained, which can more clearly display the periodic modulation strength relative to the background energy, providing high-quality input for subsequent bidirectional enhancement processing based on the autocorrelation function.

[0030] Step S1 can separate the periodic modulation information related to the fault hidden in the vibration signal from the complex background noise and visualize it by constructing the cyclic coherence spectrum, which is the premise for the entire method to achieve effective enhancement of fault features in high-noise background.

[0031] ​Further, step S2 involves autocorrelation analysis on the aforementioned obtained cyclic coherence spectrum, which is the key step to achieve fault feature enhancement. The core of this step is to further mine and strengthen the periodic fault features contained therein by calculating the autocorrelation function of the cyclic coherence spectrum.

[0032] Specifically, for the cyclic coherence spectrum that has been obtained , the autocorrelation function thereof in the dimension of the cyclic frequency is calculated, which is independently performed for all spectral frequencies , and the mathematical expression thereof is defined as: , wherein, represents the cyclic frequency lag, i.e., the delay variable in the autocorrelation function, for detecting the repetition interval of the periodic component; represents the autocorrelation calculation.

[0033] This operation is the autocorrelation operation on the amplitude square of the cyclic coherence spectrum with respect to the cyclic frequency at each fixed spectral frequency . The purpose of taking the square of is to enhance the peak feature in the cyclic coherence spectrum, so that the periodic fault modulation is more prominent relative to the background noise. The process of multiplying the energy of the original spectrum with the energy of the spectrum after shifting and then taking the expectation can effectively measure the similarity and repetition of the periodic component in the direction of the cyclic frequency.

[0034] For the fault feature with strong periodicity, when the lag amount is exactly equal to the fault period or an integer multiple thereof, the autocorrelation function will have a clear peak.

[0035] This step aims to implement secondary enhancement on the periodic fault feature. Although the cyclic coherence spectrum has achieved the preliminary extraction of the periodic component, the fault feature is often still not obvious enough under strong noise interference. The autocorrelation operation can effectively enhance the amplitude performance of such signals based on the high correlation of the periodic component itself; at the same time, since random noise does not have periodicity, it will be significantly weakened in the autocorrelation processing process.

[0036] This calculation provides a direct basis for the subsequent determination of adaptive weights, and the peak position of the autocorrelation function directly corresponds to the fault feature frequency, and the peak amplitude directly reflects the prominence of the periodic feature at the corresponding frequency point, which lays an important foundation for constructing enhancement weight matrices with clear targeting in the spectral frequency dimension and the cyclic frequency dimension, respectively.

[0037] With this processing mechanism based on the intrinsic statistical characteristics of the signal, the system can reduce the dependence on artificial preset parameters, thereby enhancing its adaptability and robust performance under complex operating conditions.

[0038] Further, the core of assigning weights to the spectral frequency direction in step S2 is to quantify the contribution of different frequency bands to the fault feature, so as to realize targeted signal enhancement; this is achieved by adaptively constructing a weight vector using the periodic characteristics revealed by the autocorrelation function.

[0039] Specifically, the peak characteristics of the autocorrelation function calculated based on the foregoing are analyzed. For each fixed spectral frequency , the autocorrelation function varies in the cycle frequency lag dimension. The periodic impact caused by the fault will exhibit a clear correlation peak synchronized with the fault period in the autocorrelation domain, so the first maximum peak of the autocorrelation function after the zero crossing (i.e. ) is usually corresponds to the potential fault feature frequency.

[0040] In order to utilize this feature, a peak feature matrix is constructed, which is defined as follows: , wherein denotes the cycle frequency lag corresponding to the first maximum peak after the zero crossing, denotes the cycle coherence spectrum spectral frequency ordinal number, denotes the cycle frequency ordinal number, denotes the th spectral frequency, denotes the th cycle frequency lag, denotes the amplitude of the autocorrelation function at the position of the first maximum peak after the zero crossing.

[0041] The construction process of this matrix is a feature selection process. At each spectral frequency , only the amplitude at the position of the first key peak is retained, and the values at all other lag positions are set to 0. The matrix clearly marks the position and strength of the strongest periodic component related to the fault fundamental frequency in each frequency band.

[0042] Based on this peak feature matrix, the weight in the spectral frequency direction is calculated by summing along the cycle frequency lag direction: .

[0043] This step effectively suppresses the interference of non-contributing frequency bands. For those bands that do not contain fault periodic components or have weak components, the weight value is small, and in the subsequent enhancement step, the information of these bands will be weakened, thereby reducing the influence of irrelevant noise and interference on the final diagnosis result.

[0044] Further, on the basis of determining the frequency direction weight of the spectrum, the weight of the cycle frequency direction is determined.

[0045] Specifically, a periodic matrix is constructed, which is based on the harmonic structure characteristics of the fault feature on the cycle frequency axis, and its mathematical definition is as follows: , wherein, is a positive integer multiple ( ), used to represent the fundamental frequency ( ) of the fault feature frequency and its harmonics; is the cycle frequency lag corresponding to the first maximum peak after the autocorrelation function zero-crossing, that is, the estimated position of the fault feature fundamental frequency. This matrix marks the peak position corresponding to the fault feature fundamental frequency and its harmonics in the cycle frequency-spectrum frequency two-dimensional space.

[0046] After obtaining the periodic matrix, the weight in the cycle frequency direction is obtained by aggregating the autocorrelation peaks on all spectrum frequencies, and the calculation process is represented as: , wherein, represents an exponential operation, which performs a nonlinear transformation on the summation result to enhance the prominence of fault-related components; represents the summation of all spectrum frequency ranges, which accumulates the harmonic peak energy distributed on each cycle frequency lag to form a feature vector reflecting the strength of periodicity.

[0047] Based on the periodicity in the cycle frequency direction, the cycle frequency lag corresponding to the cycle frequency , that is, letting , the cycle frequency weight is obtained: , wherein, represents the summation of all spectrum frequency ranges.

[0048] By constructing the periodic matrix and the exponential weighted normalization processing, the fault feature frequency and its harmonic can be effectively highlighted in the cyclic frequency position, and the recognizability of the periodic fault component in the strong noise background can be significantly improved.

[0049] Further, in step S3, based on the bidirectional weight obtained in the foregoing steps, the original cyclic coherence spectrum is directionally enhanced, and an integrated envelope spectrum for fault diagnosis is finally generated.

[0050] The weight vector is combined with to construct a two-dimensional bidirectional enhancement matrix : ; The enhancement matrix constitutes a two-dimensional weighted template, and the amplitude distribution of the template in the cyclic frequency-spectral frequency plane reflects the relative importance of different positions to the fault feature. At the cyclic frequency corresponding to the fault feature frequency and its harmonic, and at the spectral frequency corresponding to the fault resonance frequency band, the matrix element has a larger value, so that the key area is directionally enhanced.

[0051] Further, the bidirectional enhancement matrix is multiplied by the original cyclic coherence spectrum to obtain an autocorrelation enhanced cyclic coherence spectrum : ; This operation is equivalent to spatially variable gain filtering of the original spectrum, and under the modulation of the weighted template, the periodic modulation component related to the fault is significantly enhanced, while the irrelevant noise and interference components are suppressed, thereby improving the signal-to-noise ratio of the signal.

[0052] Further, the autocorrelation bidirectional enhanced integrated envelope is calculated by summing the spectral frequency direction of the autocorrelation enhanced cyclic coherence spectrum : , wherein f s represents the sampling frequency, the integral operation compresses the two-dimensional spectrum information to the cyclic frequency dimension to generate a one-dimensional integrated envelope spectrum, and in the integrated envelope spectrum, the fault feature frequency and its harmonics present prominent spectral peaks, providing a clear and reliable basis for the identification and diagnosis of mechanical faults.

[0053] After obtaining the autocorrelation bidirectional enhanced integrated envelope , the fault diagnosis of the mechanical equipment can be performed based on the spectrum. The specific implementation process is as follows: the autocorrelation bidirectional enhanced integrated envelope is obtained in the cyclic frequency A one-dimensional function in the dimension, the spectral peak position corresponds to the frequency of the periodic impulse component present in the signal. When mechanical specific components (such as rolling bearings, gears, etc.) occur local damage, its fault characteristic frequency (such as the bearing through frequency) and each order harmonic can be obtained according to its geometric parameters and operating conditions as a theoretical value.

[0054] Fault diagnosis is completed by identifying whether there are significant spectral peaks in the envelope spectrum that match these theoretical fault characteristic frequencies. Specifically, on the calculated envelope spectrum, the local maximum points with amplitude exceeding the preset threshold are located, and the cycle frequency corresponding to these points is the detected significant periodic component. Then, the detected spectral peak frequency is compared with the theoretical characteristic frequency and its harmonics of each type of fault (such as bearing inner ring fault, outer ring fault, rolling element fault, etc.) that may occur in the component to be diagnosed.

[0055] If a prominent spectral peak is observed near a certain theoretical fault characteristic frequency (or its multiple frequency), and its amplitude is significantly higher than the background noise level, it can be determined that the equipment has a fault type corresponding to the frequency. The amplitude of the spectral peak can reflect the severity of the fault to some extent. This method effectively suppresses noise and enhances fault characteristics, so that even in the early stage of weak fault, the characteristic frequency spectral peak can be clearly visible in the integrated envelope , thereby achieving effective monitoring of the running state of mechanical equipment and early and accurate diagnosis of faults.

[0056] Experiment 1: To verify the effectiveness of the autocorrelation bidirectional enhancement integrated envelope method proposed in the present application, simulation experiments were carried out as shown in Figure 2 , using a set of simulated vibration signals as input: , The vibration signal model includes four parts: the first term represents the periodic pulse caused by outer ring failure; the second term represents a random pulse signal; the third term represents the periodic harmonic interference caused by shaft rotation or gear meshing; and the last component represents Gaussian white noise. In the model, represents the repetition period of the fault pulse, represents the random slip effect of the rolling element, which is a random value in the range of [0.01T, 0.02T], is the fault pulse sequence number, is the maximum number of fault pulses, is used to adjust the amplitude of the first two fault pulse components; is the fault pulse sequence number, is the maximum number of fault pulses, represents the repetition period of the fault pulse, Indicates the amplitude of a random pulse; This represents the random slip effect of rolling elements. This represents the amplitude of a periodic harmonic. Indicates the first The frequency of each component; This represents the impact response function of rotating machinery.

[0057] Each vibration signal is 20,000 units long and sampled at a frequency of 20,000 Hz. The fault characteristic frequencies of the outer ring are as follows: = 80Hz, and its excitation resonance frequency is 3800Hz. The excitation resonance frequency of the random shock wave is 8000Hz. Gaussian white noise is introduced, and its signal-to-noise ratio is set to -18dB.

[0058] First, the above vibration signal was preprocessed and its cyclic coherence spectrum was calculated. The results are shown in Figure 3. Figure 3 The distribution of the original signal on the two-dimensional plane of cyclic frequency-spectral frequency is shown. It shows that there are weak periodic modulation components at the fault characteristic frequency of 80Hz and its harmonics, but the overall spectrum is severely interfered with by background noise, and the fault characteristics are not obvious.

[0059] Subsequently, the autocorrelation function of the cyclic coherence spectrum is calculated, and the weights of the spectral frequency direction and the cyclic frequency direction are adaptively determined based on the peak characteristics of the autocorrelation function after the zero-crossing point. By constructing a fault feature enhancement weight matrix, the original cyclic coherence spectrum is weighted and enhanced, resulting in the enhanced cyclic coherence spectrum as shown below. Figure 4 As shown. With Figure 3 In comparison, the amplitude of the enhanced spectrum at the fault characteristic frequency and its harmonics is significantly increased, while background noise and irrelevant interference are effectively suppressed, indicating that the bidirectional enhancement mechanism successfully highlights the periodic fault component.

[0060] Finally, the enhanced cyclic coherence spectrum is integrated along the frequency direction to obtain the autocorrelation bidirectional enhancement integrated envelope. ,like Figure 5 As shown, the envelope spectrum exhibits sharp and prominent peaks at the fault characteristic frequencies of 80 Hz and 160 Hz, which are completely consistent with the preset real fault components in the simulation signal, clearly revealing the periodic characteristics of the outer ring fault.

[0061] For comparison, such as Figure 6As shown, it shows the result of the same signal processed by the traditional envelope analysis method. In the envelope spectrum generated by this method, the peak value at the fault characteristic frequency is low in recognition, and is basically submerged in the extensive noise background, and it is difficult to effectively identify the fault component from it. This result strongly confirms that the method has a significant advantage in enhancing the fault characteristics and improving the reliability of diagnosis in a strong noise environment.

[0062] Embodiment two: As Figure 7 shown, the application also provides a self-correlation bidirectional enhancement integrated envelope system for implementing the self-correlation bidirectional enhancement integrated envelope method described in embodiment one, comprising the following modules: A signal acquisition and preprocessing module is used for preprocessing the acquired vibration signal of the mechanical equipment to obtain the cyclic coherence spectrum of the vibration signal. A self-correlation analysis and weight determination module is used for calculating the autocorrelation function of the cyclic coherence spectrum, and determining the weight in the spectral frequency direction and the weight in the cyclic frequency direction of the cyclic coherence spectrum based on the peak value characteristics of the autocorrelation function. A spectrum enhancement and envelope generation module is used for enhancing the cyclic coherence spectrum based on the weight in the spectral frequency direction and the weight in the cyclic frequency direction to obtain a self-correlation bidirectional enhancement integrated envelope.

[0063] The system in this embodiment completely corresponds to the core process of the method in embodiment one, forming a closed-loop processing link from signal input to diagnosis result output. The design of the system ensures the implementability and repeatability of the method in engineering application.

[0064] Through the modular system construction, the complex signal processing process can be decomposed into logical and clear functional units, effectively reducing the complexity of algorithm deployment and maintenance. The data flow between the modules of the system is clear, and the interface definition is clear, which is convenient for integration into the existing diagnosis platform or development as an independent diagnosis tool.

[0065] The system has good adaptability in actual engineering, and can automatically complete the conversion process from the original vibration signal to the enhanced envelope spectrum, significantly improving the efficiency and reliability of fault diagnosis. The modular structure also provides convenience for subsequent function expansion and performance optimization.

[0066] Embodiment three: The application also provides a computer storage medium, which stores a computer software product, and the computer software product comprises a plurality of instructions for causing a computer device to execute the self-correlation bidirectional enhancement integrated envelope method described in embodiment one.

[0067] Those skilled in the art will appreciate that embodiments of the application can be devised for a method, a system, or a computer program product. Accordingly, the present application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0068] The present application is described in reference to the flowchart illustrations and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0069] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0070] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0071] Obviously, the above-described embodiments are only examples and are not intended to limit the present application. Based on the above description, those skilled in the art can make other variations and modifications of the present application without departing from the present application. Neither requiring nor intending to limit the present application to the exact forms of implementations shown and described, the application is amenable to a number of changes and modifications.

Claims

1. A self-correlation bidirectional enhancement integrated envelope method, characterized by, The method comprises the following steps: S1: preprocessing the acquired vibration signal of the mechanical equipment to obtain a cyclic coherence spectrum of the vibration signal; S2: calculating an autocorrelation function of the cyclic coherence spectrum; determining a weight in a spectral frequency direction and a weight in a cyclic frequency direction of the cyclic coherence spectrum based on a peak value feature of the autocorrelation function; S3: performing enhancement processing on the cyclic coherence spectrum based on the weight in the spectral frequency direction and the weight in the cyclic frequency direction to obtain an autocorrelation bidirectional enhancement integrated envelope.

2. The self-correlation bidirectional reinforcement integrated envelope method according to claim 1, characterized in that: In step S1, the vibration signal is preprocessed to obtain a cyclic coherence spectrum of the vibration signal. The method is as follows: calculating an autocorrelation function of the vibration signal, performing Fourier transform on the autocorrelation function in two dimensions of time lag and cycle frequency, and constructing a correlation spectrum capable of representing both spectral characteristics and cycle characteristics of the signal The expression formula is: ， wherein, denotes a periodic modulation in the signal; denotes a spectral frequency; denotes a Fourier transform of the autocorrelation function of the vibration signal, denotes ordinal number of the mid-point; denotes a period of cyclostationarity.

3. The self-correlation bidirectional reinforced integrated envelope method according to claim 2, characterized in that: The method for obtaining the cyclic coherence spectrum of the vibration signal further comprises: performing correlation spectrum on the characteristic spectrum and period of the characteristic signal performing normalization processing to obtain the cyclic coherence spectrum of the vibration signal , and an expression formula is: ， wherein represents a value of when .

4. The self-recursive bidirectional enhancement integrated envelope method of claim 1, wherein: In step S2, the method for calculating the autocorrelation function of the cyclic coherence spectrum is: calculating the autocorrelation function corresponding to all spectral frequencies of the cyclic coherence spectrum The expression formula is: ， wherein, denotes a cycle frequency lag, denotes an autocorrelation calculation, denotes the cyclic coherence spectrum, denotes a periodic modulation in the signal; denotes a spectral frequency.

5. The self-correlation bidirectional enhancement integrated envelope method according to claim 1, wherein: In step S2, based on the peak characteristics of the autocorrelation function, the method for determining the weight of the cyclic coherence spectrum in the frequency spectrum frequency direction is: based on the first maximum peak after the zero-crossing point of the autocorrelation function corresponding to the fault characteristic frequency, the peak values at the positions of the first maximum peak after the zero-crossing point and its multiples are stored into a matrix to obtain a peak characteristic matrix : ， wherein, denotes the cycle frequency lag corresponding to the first maximum peak after the zero crossing, denotes the cycle coherence spectrum frequency index, denotes the cycle frequency index, denotes the first spectrum frequency, denotes the first cycle frequency lag, denotes the amplitude of the autocorrelation function at the position of the first maximum peak after the zero crossing; Obtaining weights for spectral frequencies based on a peak feature matrix : 。 6. The self-recursive bidirectional enhancement integrated envelope method according to claim 1 or 5, characterized in that: In step S2, based on the peak characteristics of the autocorrelation function, the method for determining the weight of the cyclic coherence spectrum in the cyclic frequency direction based on the weight of the cyclic coherence spectrum in the spectral frequency direction is: based on the periodicity of the vibration signal, a period matrix is constructed by expanding the maximum value to all multiples of the fault characteristic frequency : ， wherein, denotes the multiple, denotes the cycle frequency lag corresponding to the first maximum peak after the zero crossing, denotes the cycle coherence spectrum frequency index, denotes the cycle frequency index, denotes the first spectrum frequency, denotes the first cycle frequency lag, denotes the amplitude of the autocorrelation function at the position of the first maximum peak after the zero crossing; The weight of the cyclic frequency is obtained by adding the autocorrelation peaks of all spectral frequencies and taking the exponential : , wherein denotes an exponential operation, denotes a summation over all spectral frequency ranges; Based on the periodicity in the direction of the cycle frequency, the cycle frequency is lagged by the cycle frequency corresponding to the equation , resulting in cycle frequency weights : ， wherein is the spectral frequency, denotes the sum over all spectral frequency ranges.

7. The self-correlation bidirectional enhancement integrated envelope method according to claim 1, wherein: In step S3, the method for obtaining the self-correlation bidirectional enhancement integrated envelope is: combining the frequency domain and cycle domain weights to obtain a bidirectional enhancement matrix : , wherein, denotes a weight of the spectral frequency based on the peak feature matrix, denotes a cyclic frequency weight; Multiplying the two-way enhancement matrix with the original cyclically coherent spectrum results in the autocorrelation-enhanced cyclically coherent spectrum : , wherein denotes the autocorrelation function corresponding to all spectral frequencies of the cyclically coherent spectrum; The autocorrelation bidirectional enhanced integrated envelope is calculated by summing the spectral frequency direction of the autocorrelation enhanced cyclic spectral : ， wherein denotes the sampling frequency.

8. The self-correlation bidirectional enhancement integrated envelope method according to claim 1, wherein: After the autocorrelation bidirectional enhancement integrated envelope is obtained, fault diagnosis of the mechanical equipment is performed by identifying a spectral peak corresponding to a fault characteristic frequency in the autocorrelation bidirectional enhancement integrated envelope.

9. A self-correlation bidirectional enhancement integrated envelope system, characterized by, The method comprises the following modules: a signal acquisition and preprocessing module, configured to preprocess an acquired vibration signal of the mechanical equipment to obtain a cyclic coherence spectrum of the vibration signal; an autocorrelation analysis and weight determination module, configured to calculate an autocorrelation function of the cyclic coherence spectrum and determine a weight in a spectral frequency direction and a weight in a cyclic frequency direction of the cyclic coherence spectrum based on a peak value feature of the autocorrelation function; a spectral enhancement and envelope generation module, configured to perform enhancement processing on the cyclic coherence spectrum based on the weight in the spectral frequency direction and the weight in the cyclic frequency direction to obtain an autocorrelation bidirectional enhancement integrated envelope.

10. A computer storage medium, characterized in that, The computer storage medium stores a computer software product, and the computer software product comprises a plurality of instructions for causing a computer device to execute the autocorrelation bidirectional enhancement integrated envelope method in any one of claims 1 to 8.

Images