Generator fault diagnosis method and system based on vibration trend prediction

Abstract

The invention relates to the technical field of fault diagnosis, and discloses a generator fault diagnosis method and system based on vibration trend prediction. The method comprises the steps that prediction model training is carried out on a regularized vibration sequence through an extreme learning machine algorithm, a vibration trend prediction model is obtained, and the model comprises an input layer, a hidden layer and an output layer; performing adaptive adjustment processing on the vibration trend prediction model through an error sum of squares evaluation function to obtain an optimized prediction model, and when a cycle index error ratio is greater than 2, setting a rule function value to zero; when the cycle index error ratio is less than or equal to 2, the law function value is equal to the product of the last law function value and the natural logarithm of the error ratio; and performing comparative analysis on an output result of the optimized prediction model and real-time monitoring data to obtain a generator fault diagnosis result. According to the invention, the efficiency and accuracy of generator fault diagnosis based on vibration trend prediction are improved.

Description

Technical Field

[0001] The present application relates to the field of fault diagnosis, and in particular to a generator fault diagnosis method and system based on vibration trend prediction. Background Art

[0002] As the core equipment in the power system, the operating status of the generator is directly related to the safety and reliability of the power supply. At present, the fault diagnosis of the generator is mainly achieved through online monitoring of vibration data. Common diagnostic methods include spectrum analysis, wavelet transform, empirical mode decomposition, etc. These methods analyze the vibration signal in the time domain and frequency domain, extract characteristic parameters, and establish a fault diagnosis model. With the development of artificial intelligence technology, fault diagnosis methods based on machine learning have gradually been applied to generator status monitoring. For example, algorithms such as support vector machines and neural networks are used for pattern recognition and fault classification of vibration signals.

[0003] However, existing fault diagnosis methods have some limitations when processing generator vibration signals. First, vibration signals often have nonlinear and non-stationary characteristics, and traditional signal processing methods are difficult to effectively extract fault features. Second, most diagnostic algorithms only focus on the vibration state at the current moment and lack the ability to predict the development trend of faults, resulting in the inability to detect potential fault hazards early. In addition, in practical applications, the reliability and accuracy of diagnostic results are also affected by changes in equipment operating conditions and interference from environmental noise. Summary of the invention

[0004] The present application provides a generator fault diagnosis method and system based on vibration trend prediction, which is used to solve the technical problem that the prior art lacks the ability to predict and analyze vibration trends and is difficult to achieve early warning of faults.

[0005] In a first aspect, the present application provides a generator fault diagnosis method based on vibration trend prediction, the generator fault diagnosis method based on vibration trend prediction comprising: real-time acquisition and processing of generator vibration signals to obtain historical vibration signal sequence data; variational modal decomposition processing of the historical vibration signal sequence data to obtain multiple inherent modal function subsequences; regularization processing of the inherent modal function subsequences, wherein dual-mode processing of the subsequences is performed through a preset regular function to obtain a regularized vibration sequence, the dual-mode processing comprising: when the absolute value of the time error is less than five percent of the regular function value, processing is performed using a regular function value equal to the product of the time error and the regular function period value; when the absolute value of the time error is greater than or equal to five percent of the regular function value, processing is performed using a regular function value equal to the product of the time error and the regular function period value. The regular function value is equal to the sum of the time error and the regular function period value for processing; the regularized vibration sequence is trained with the extreme learning machine algorithm to obtain a vibration trend prediction model, wherein the prediction model includes an input layer, a hidden layer and an output layer, and the weight value of the input layer and the neuron threshold of the hidden layer are randomly generated; the vibration trend prediction model is adaptively adjusted with the error square sum evaluation function to obtain an optimized prediction model, wherein when the cycle number error ratio is greater than two, the regular function value is reset to zero; when the cycle number error ratio is less than or equal to two, the regular function value is equal to the product of the previous regular function value and the natural logarithm of the error ratio; the output result of the optimized prediction model is compared and analyzed with the real-time monitoring data to obtain the generator fault diagnosis result.

[0006] In a second aspect, the present application provides a generator fault diagnosis system based on vibration trend prediction, the generator fault diagnosis system based on vibration trend prediction comprising: The acquisition module is used to collect and process the vibration signal of the generator in real time to obtain historical vibration signal sequence data; A decomposition module, used for performing variational modal decomposition processing on the historical vibration signal sequence data to obtain a plurality of intrinsic modal function subsequences; A processing module is used to perform regular processing on the subsequence of the intrinsic mode function, wherein the subsequence is subjected to dual-mode processing by a preset regular function to obtain a regular vibration sequence, wherein the dual-mode processing includes: when the absolute value of the time error is less than 5% of the regular function value, the regular function value is equal to the product of the time error and the regular function period value for processing; when the absolute value of the time error is greater than or equal to 5% of the regular function value, the regular function value is equal to the sum of the time error and the regular function period value for processing; A training module is used to perform prediction model training on the regularized vibration sequence through an extreme learning machine algorithm to obtain a vibration trend prediction model, wherein the prediction model includes an input layer, a hidden layer and an output layer, and the weight value of the input layer and the neuron threshold of the hidden layer are randomly generated; An adjustment module is used to adaptively adjust the vibration trend prediction model through an error square sum evaluation function to obtain an optimized prediction model, wherein when the cycle number error ratio is greater than two, the regular function value is set to zero; when the cycle number error ratio is less than or equal to two, the regular function value is equal to the product of the previous regular function value and the natural logarithm of the error ratio; The comparison module is used to compare and analyze the output result of the optimized prediction model with the real-time monitoring data to obtain the generator fault diagnosis result.

[0007] In the technical solution provided by the present application, the complete historical vibration signal sequence data is obtained by real-time acquisition and processing of the generator vibration signal, thereby ensuring the accuracy and timeliness of the original data. The variational mode decomposition processing method is used to decompose the historical vibration signal sequence data to obtain multiple intrinsic mode function subsequences, effectively separating the different frequency components in the signal and enhancing the ability to analyze complex vibration signals. By regularizing the intrinsic mode function subsequences, introducing a dual-mode processing mechanism, and using different calculation methods according to the absolute value of the time error, the influence of non-periodic and nonlinear characteristics on the prediction accuracy is effectively reduced. The extreme learning machine algorithm is used to train the prediction model for the regularized vibration sequence. The algorithm simplifies the training process of the traditional neural network and improves the calculation efficiency by randomly generating the input layer weight value and the hidden layer neuron threshold. The error square sum evaluation function is introduced to adaptively adjust the vibration trend prediction model, and the regular function value is dynamically adjusted according to the size of the cycle number error ratio, thereby enhancing the model's adaptability to changes in vibration trends. Finally, by comparing and analyzing the output results of the optimized prediction model with the real-time monitoring data, it is possible to promptly detect abnormal changes in the generator's operating status, accurately determine the type of fault, achieve early warning and accurate diagnosis of generator faults, and improve the safety and reliability of equipment operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0009] Figure 1 A schematic diagram of an embodiment of a generator fault diagnosis method based on vibration trend prediction in an embodiment of the present application; Figure 2 Schematic diagram of an embodiment of a generator fault diagnosis system based on vibration trend prediction in an embodiment of the present application. DETAILED DESCRIPTION

[0010] The embodiment of the present application provides a generator fault diagnosis method and system based on vibration trend prediction. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0011] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of a generator fault diagnosis method based on vibration trend prediction includes: Step S101, collecting and processing the vibration signal of the generator in real time to obtain historical vibration signal sequence data; Step S102, performing variational mode decomposition processing on the historical vibration signal sequence data to obtain a plurality of intrinsic mode function subsequences; Step S103, regularizing the subsequence of the intrinsic mode function, wherein the subsequence is subjected to dual-mode processing by a preset regular function to obtain a regularized vibration sequence, wherein the dual-mode processing includes: when the absolute value of the time error is less than 5% of the regular function value, processing is performed using the regular function value being equal to the product of the time error and the regular function period value; when the absolute value of the time error is greater than or equal to 5% of the regular function value, processing is performed using the regular function value being equal to the sum of the time error and the regular function period value; Step S104, training the regularized vibration sequence with a prediction model by using an extreme learning machine algorithm to obtain a vibration trend prediction model, wherein the prediction model includes an input layer, a hidden layer and an output layer, and the weight value of the input layer and the neuron threshold of the hidden layer are randomly generated; Step S105, the vibration trend prediction model is adaptively adjusted through the error square sum evaluation function to obtain an optimized prediction model, wherein when the cycle number error ratio is greater than two, the regular function value is set to zero; when the cycle number error ratio is less than or equal to two, the regular function value is equal to the product of the previous regular function value and the natural logarithm of the error ratio; Step S106: Compare and analyze the output result of the optimized prediction model with the real-time monitoring data to obtain the generator fault diagnosis result.

[0012] It is understandable that the execution subject of the present application may be a generator fault diagnosis system based on vibration trend prediction, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0013] Specifically, in the real-time acquisition process of the generator vibration signal, a high-precision vibration sensor is used to collect the vibration signal generated during the operation of the generator, and the sampling frequency is set to 10kHz to ensure the acquisition accuracy of the signal. The collected original vibration signal is subjected to wavelet denoising to remove the influence of environmental noise and interference signals. The characteristic parameters such as the root mean square value, peak value, and kurtosis of the vibration signal are obtained by time domain feature extraction, and the frequency domain conversion is performed to obtain the frequency domain feature sequence. The time domain and frequency domain features are fused to form a complete historical vibration signal sequence data. When performing variational modal decomposition on the acquired historical vibration signal sequence data, the number of decomposition layers is first determined to be 5 layers, and the signal is decomposed at multiple scales. By constructing quadratic penalty terms and Lagrangian multipliers, an augmented Lagrangian expression is established, and the alternating direction multiplier method is used for iterative optimization to obtain vibration modes at different scales. Each mode represents an inherent oscillation component in the original signal, and these components are called inherent modal function subsequences. A Hilbert transform is performed on each subsequence to obtain the instantaneous frequency characteristics, thereby achieving a complete decomposition of the vibration signal.

[0014] In the regularization processing stage, a regular function is set for each intrinsic mode function subsequence, and a dual-mode processing method is used for regularization. When the calculated absolute value of the time error is less than 5% of the regular function value, the regular function value is set to the product of the time error and the period value of the regular function; when the absolute value of the time error is less than or equal to 5%, the regular function value is set to the sum of the time error and the period value of the regular function. This dual-mode processing method effectively reduces the impact of non-periodic and nonlinear characteristics on the prediction accuracy, and finally obtains a regularized vibration sequence. The extreme learning machine algorithm is used to process the regularized vibration sequence, and a prediction model including an input layer, a hidden layer, and an output layer is constructed. The number of input layer nodes is set to 12, corresponding to the characteristic dimension of the regularized vibration sequence, the number of hidden layer nodes is set to 24, and the number of output layer nodes is 1. The input layer weight value and the hidden layer neuron threshold are initialized by random generation, and the range is set between [-1,1]. The hidden layer output is nonlinearly mapped by the activation function, and the output layer weight matrix is ​​calculated, and finally the vibration trend prediction model is constructed. In the adaptive adjustment stage of the prediction model, the error square sum evaluation function is introduced to evaluate the model prediction results. Calculate the error between the model prediction value and the actual value, count the number of cycles and calculate the error ratio. When the error ratio of the number of cycles is greater than 2, the law function value is directly set to 0; when the error ratio of the number of cycles is less than or equal to 2, the law function value is updated to the product of the previous law function value and the natural logarithm of the error ratio. This adaptive adjustment mechanism can adjust the model parameters in time according to the changes in the prediction error to obtain an optimized prediction model with better prediction performance. Compare the output results of the optimized prediction model with the real-time collected monitoring data, calculate the feature deviation, divide the deviation interval, match and verify the fault type, and finally obtain the diagnosis results including the fault type, fault probability and severity.

[0015] Taking a generator as an example, during normal operation, the amplitude of the original vibration signal collected by the vibration sensor fluctuates between 0.1mm and 0.5mm. After variational mode decomposition, five intrinsic mode function subsequences are obtained, among which the main vibration components are concentrated in the second and third subsequences. These subsequences are processed in a regularized manner. When the absolute value of the time error at a certain moment is 0.03 (less than 5% of the regular function value of 0.8), the product mode is used for processing; when the absolute value of the time error is 0.04 (equal to 5%), the sum mode is used for processing. After the prediction model is obtained through extreme learning machine algorithm training, the calculated cycle error ratio is 2.5, at which time the regular function value is set to zero; when the error ratio drops to 1.8, if the previous regular function value is 0.6, the new regular function value is updated to 0.35. The deviation between the vibration trend output by the final prediction model and the real-time monitoring data is within 3%, and the generator rotor misalignment fault is accurately identified, with a fault probability of 85%, which is of medium severity.

[0016] In the embodiment of the present application, the complete historical vibration signal sequence data is obtained by real-time acquisition and processing of the generator vibration signal, thereby ensuring the accuracy and timeliness of the original data. The variational mode decomposition processing method is used to decompose the historical vibration signal sequence data to obtain multiple intrinsic mode function subsequences, effectively separating the different frequency components in the signal and enhancing the ability to analyze complex vibration signals. By regularizing the intrinsic mode function subsequences, introducing a dual-mode processing mechanism, and using different calculation methods according to the absolute value of the time error, the influence of non-periodic and nonlinear characteristics on the prediction accuracy is effectively reduced. The extreme learning machine algorithm is used to train the prediction model for the regularized vibration sequence. The algorithm simplifies the training process of the traditional neural network and improves the calculation efficiency by randomly generating the input layer weight value and the hidden layer neuron threshold. The error square sum evaluation function is introduced to adaptively adjust the vibration trend prediction model, and the regular function value is dynamically adjusted according to the size of the cycle number error ratio, thereby enhancing the model's adaptability to changes in vibration trends. Finally, by comparing and analyzing the output results of the optimized prediction model with the real-time monitoring data, it is possible to promptly detect abnormal changes in the generator's operating status, accurately determine the type of fault, achieve early warning and accurate diagnosis of generator faults, and improve the safety and reliability of equipment operation.

[0017] In a specific embodiment, the process of executing step S101 may specifically include the following steps: (1) Setting the sampling frequency of the generator vibration signal to obtain the vibration signal sampling parameters, and performing the Nyquist sampling theorem test on the vibration signal sampling parameters to obtain the effective sampling parameters; (2) Collecting vibration signals based on effective sampling parameters to obtain original vibration signal data, and performing wavelet denoising on the original vibration signal data to obtain denoised vibration signal data; (3) Extracting time domain features from the denoised vibration signal data to obtain a time domain feature sequence, and performing frequency domain conversion processing on the time domain feature sequence to obtain a frequency domain feature sequence; (4) Performing feature fusion processing on the time domain feature sequence and the frequency domain feature sequence to obtain a fused feature sequence, and normalizing the fused feature sequence to obtain a standardized feature sequence; (5) The standardized feature sequence is divided into sliding time windows to obtain time series segments, and the time series segments are checked for data integrity to obtain historical vibration signal sequence data.

[0018] Specifically, the sampling frequency of the vibration sensor is reasonably set. According to the characteristics of the generator vibration signal, the sampling frequency is generally set between 1kHz and 20kHz. The specific setting needs to meet the requirements of the Nyquist sampling theorem, that is, the sampling frequency should not be less than twice the highest frequency of the signal. Through the analysis of the vibration signal spectrum, it is found that the main vibration frequency components of the generator are distributed in the range of 10Hz to 2kHz. Therefore, the selection of a sampling frequency of 10kHz fully meets the requirements of the sampling theorem and ensures the integrity and accuracy of the signal. When collecting vibration signals, a highly sensitive piezoelectric acceleration sensor is used, and the sensor is installed in key parts such as the bearing seat and the base of the generator. The sensor converts the collected mechanical vibration signal into an electrical signal, amplifies and filters it through the signal conditioning circuit, and then performs analog-to-digital conversion through a high-precision data acquisition card to obtain the original digital vibration signal data. In order to eliminate the influence of measurement noise and environmental interference, the original signal is processed by wavelet denoising method, the db4 wavelet basis function is selected, the signal is decomposed into 4 layers, the wavelet coefficients are processed by the soft threshold function, and finally the noise-reduced vibration signal data is reconstructed.

[0019] The time domain features of the vibration signal data after noise reduction are extracted, and statistical feature parameters such as the root mean square value, peak value, kurtosis factor, and margin factor are calculated. The root mean square value reflects the energy of the vibration, the peak value indicates the maximum amplitude of the vibration, the kurtosis factor describes the pulse characteristics of the signal, and the margin factor reflects the steepness of the waveform. These time domain features constitute a time domain feature sequence, which can describe the characteristics of the vibration signal from different angles. At the same time, the vibration signal is subjected to a fast Fourier transform to obtain a frequency domain feature sequence, including feature parameters such as the main frequency component, frequency band energy distribution, and harmonic components in the spectrum. In order to fully reflect the characteristics of the vibration signal, it is necessary to fuse the time domain feature sequence and the frequency domain feature sequence. The feature hierarchical fusion method is adopted. First, weights are assigned to different feature parameters. Important features are given larger weights, and secondary features are given smaller weights. The fused feature sequence is obtained through weighted combination. Then the fused feature sequence is normalized, and the feature parameters of different dimensions are unified into the [0,1] interval to obtain a standardized feature sequence, which is convenient for subsequent data processing and analysis.

[0020] The standardized feature sequence is divided into sliding time windows, with the window length set to 1 second and the sliding step length set to 0.1 second, which not only ensures the continuity of the data, but also can reflect the changes in vibration characteristics in a timely manner. The data in each time window is checked for integrity, and fragments with missing or abnormal data are removed, finally obtaining complete and reliable historical vibration signal sequence data.

[0021] Taking a large steam turbine generator set as an example, when running at a rated speed of 3000rpm, a vibration sensor is installed on the bearing seat for data collection. The sampling frequency is set to 10kHz. Since the highest vibration frequency of the generator is about 2kHz, it satisfies the Nyquist sampling theorem (10kHz>2×2kHz). The amplitude of the original vibration signal collected is between 0.1mm and 0.5mm. After wavelet denoising, the noise amplitude is reduced by about 80%. The time domain feature extraction obtains a root mean square value of 0.25mm, a peak value of 0.48mm, a kurtosis factor of 3.2, and a margin factor of 4.5; the frequency domain analysis shows that the amplitude of the frequency component (50Hz) is 0.15mm, and there are 2-fold and 3-fold harmonic components. When these feature parameters are fused, a weight of 0.4 is assigned to the amplitude-related features and a weight of 0.6 is assigned to the frequency-related features. After normalization, the standardized feature values ​​are all between 0 and 1. The time window is divided into 1 second (containing 10,000 sampling points), and adjacent windows overlap by 9,000 sampling points. The generated historical vibration signal sequence contains complete time domain and frequency domain feature information.

[0022] In a specific embodiment, the process of executing step S102 may specifically include the following steps: (1) Performing data segmentation processing on the historical vibration signal sequence data to obtain multiple vibration signal sub-segments, and performing spectrum analysis on the multiple vibration signal sub-segments to obtain spectrum feature data; (2) Calculate the number of variational mode decomposition layers for the spectral feature data to obtain the decomposition layer number parameters, and estimate the center frequency of the decomposition layer number parameters to obtain the center frequency values ​​of each layer; (3) Setting bandwidth parameters for the center frequency values ​​of each layer to obtain decomposition bandwidth parameters, and constructing secondary penalty terms for the decomposition bandwidth parameters to obtain penalty constraints; (4) Construct the Lagrangian operator for the penalty constraint condition to obtain the augmented Lagrangian expression, and perform alternating direction multiplier iteration on the augmented Lagrangian expression to obtain the decomposition coefficient sequence; (5) Performing Hilbert transform on the decomposition coefficient sequence to obtain an instantaneous frequency sequence, and performing subsequence reconstruction on the instantaneous frequency sequence to obtain multiple intrinsic mode function subsequences.

[0023] Specifically, the processing of vibration signal sequence data first requires data segmentation. The long time series is divided into multiple sub-segments according to a fixed time interval. The length of each sub-segment is set to 1024 sampling points, and the data points between adjacent sub-segments overlap by 50% to ensure the continuity of the signal. Perform fast Fourier transform on each vibration signal sub-segment to obtain spectrum data, from which characteristic parameters such as frequency components, amplitude, and phase are extracted to form a complete spectrum feature data set. Based on the acquired spectrum feature data, the optimal number of decomposition layers of variational mode decomposition is calculated. By analyzing the number of main frequency components and energy distribution in the spectrum, the range of the number of decomposition layers is preliminarily determined, which is generally set between 3 and 8 layers. The center frequency of each layer is estimated. The estimation method is based on the principle of weighted average of spectrum energy, and the frequency region with larger spectrum energy is used as the center frequency of each layer. For example, for the vibration signal of the generator, the center frequency of the first layer usually corresponds to the fundamental frequency (transfer frequency) component, and the subsequent layers correspond to higher harmonics and other characteristic frequency components.

[0024] Set the corresponding bandwidth parameters for the center frequency value of each layer. The bandwidth parameter determines the frequency range of each layer of decomposition, which directly affects the accuracy and effect of decomposition. The bandwidth is usually set to 20%-30% of the center frequency, so that there is both appropriate frequency band overlap between adjacent layers and good resolution. After determining the bandwidth parameters, construct a quadratic penalty term to constrain the bandwidth of each modal component. The construction of the quadratic penalty term takes into account the constraints on the frequency axis and the smoothness requirements on the time axis to form a complete penalty constraint condition. Based on the penalty constraint condition, construct the Lagrangian operator. The Lagrangian operator transforms the original constrained optimization problem into an unconstrained optimization problem, which is easy to solve. By introducing the Lagrangian multiplier, combining the original objective function and the constraint condition, the augmented Lagrangian expression is formed. The augmented Lagrangian expression is iteratively solved using the alternating direction multiplier method. The modal function, center frequency and Lagrangian multiplier are updated in each iteration until the final decomposition coefficient sequence is obtained.

[0025] The obtained decomposition coefficient sequence is processed by Hilbert transform, which can extract the instantaneous frequency characteristics of the signal. Hilbert transform first converts the real signal into an analytical signal, and then calculates the derivative of the phase function to obtain the instantaneous frequency. According to the characteristics of the instantaneous frequency sequence, each subsequence is reconstructed, and finally a complete set of intrinsic mode function subsequences is obtained. Taking the vibration signal processing of a generator set as an example, the 10-second vibration data collected contains a total of 100,000 sampling points, which are divided into 195 subsegments, each with 1024 points, and adjacent subsegments overlap by 512 points. The spectrum analysis of the first subsegment found that the main frequency components are distributed at 50Hz (base frequency), 100Hz (2 times frequency), 150Hz (3 times frequency), 200Hz (4 times frequency) and 300Hz (bearing fault characteristic frequency). Based on the spectrum analysis results, the number of variational mode decomposition layers is set to 5, and the center frequencies of each layer are estimated to be 50Hz, 100Hz, 150Hz, 200Hz and 300Hz respectively. When setting the bandwidth parameters, the bandwidth of the first layer (50Hz) is set to 15Hz, and the bandwidths of the other layers are 25Hz, 35Hz, 45Hz and 60Hz respectively. After Lagrangian operator construction and alternating direction multiplier iteration (the maximum number of iterations is set to 500 times, and the convergence threshold is 1e-6), 5 sets of decomposition coefficient sequences are obtained. The Hilbert transform of these coefficient sequences is performed, and the calculated instantaneous frequency shows that the average frequency of the first natural mode function subsequence is 49.8Hz, which is basically consistent with the theoretical base frequency of 50Hz; the average frequency of the second subsequence is 99.5Hz, corresponding to the 2nd frequency component; the main frequency of the third subsequence is around 151Hz, representing the 3rd frequency feature; the fourth and fifth subsequences reflect the 4th frequency and bearing fault characteristic frequency components respectively. These decomposition results clearly show the various characteristic components in the generator vibration signal, providing an important basis for subsequent fault diagnosis.

[0026] In a specific embodiment, the process of executing step S103 may specifically include the following steps: (1) Perform sliding time window processing on the intrinsic mode function subsequence to obtain a time window sequence, and perform time error calculation on the time window sequence to obtain a time error sequence; (2) Performing absolute value calculation on the time error sequence to obtain an error absolute value sequence, and performing threshold comparison on the error absolute value sequence. Errors greater than the threshold are marked as first-type errors, and those less than or equal to the threshold are marked as second-type errors, thus obtaining the error classification result; (3) Calculate the regular function value of the error classification result to obtain the initial regular function value, and perform period analysis on the initial regular function value to obtain the regular function period value; (4) performing a product operation on the sequences whose absolute value of the error is less than 5% of the regular function value to obtain the regular function value of the first mode, and performing a sum operation on the sequences whose absolute value of the error is greater than or equal to 5% of the regular function value to obtain the regular function value of the second mode; (5) The regular function values ​​of the first mode and the regular function values ​​of the second mode are sequentially merged to obtain a dual-mode processing result, and the dual-mode processing result is normalized to obtain a regularized vibration sequence.

[0027] Specifically, the first step in regularizing the intrinsic mode function subsequence is sliding time window processing, which divides the subsequence into time windows of fixed length (such as 1024 data points), sets the overlap rate between windows to 50%, and slides 512 data points each time, thereby obtaining a series of continuous time window sequences. For the data in each time window, the time difference between it and the adjacent window is calculated to form a time error sequence, which reflects the change characteristics of the data in the time dimension. The absolute value operation is performed on the obtained time error sequence, and the positive and negative errors are uniformly converted into positive values, which is convenient for subsequent threshold comparison. The threshold is set to 5% of the mean value of the current processing window data, and the error absolute value sequence is compared with the threshold. The error greater than the threshold is marked as the first type of error, and the error less than or equal to the threshold is marked as the second type of error, thereby obtaining a complete error classification result.

[0028] Based on the error classification results, the regular function value is calculated for each type of error. The regular function design adopts the form of piecewise continuous function, and different calculation methods are used for different types of errors. By performing Fourier analysis on the initial regular function value, the main periodic components are identified, and the basic periodic value of the regular function is obtained, which reflects the inherent characteristic period of the vibration signal. According to the dual-mode processing strategy, the product operation is performed on the sequence whose absolute error value is less than 5% of the regular function value, that is, the time error is multiplied by the regular function periodic value to obtain the regular function value in the first mode; for the sequence whose absolute error value is less than or equal to 5%, the addition operation is used to add the time error and the regular function periodic value to obtain the regular function value in the second mode.

[0029] The regular function values ​​obtained in the two modes are sequentially merged to form a complete dual-mode processing result. The merged sequence is normalized to unify the value range to the interval [0,1], and finally a regular vibration sequence is obtained.

[0030] Taking the vibration data processing of a generator as an example, the intrinsic mode function subsequence of 100,000 sampling points is selected and processed using a sliding window of 1024 points. When processing the 50th time window, the calculated time error sequence value is [-0.03, 0.02, -0.04, 0.01, 0.05], and the absolute value operation is [0.03, 0.02, 0.04, 0.01, 0.05]. The initial value of the regular function is set to 0.8, and the 5% threshold is 0.04. The error sequence is classified. If 0.03, 0.02, and 0.01 are less than 0.04, the first mode is used for processing; if 0.04 and 0.05 are greater than or equal to 0.04, the second mode is used for processing.

[0031] Periodic analysis shows that the basic period value of the subsequence is 0.02 seconds. In the first mode, for the error value of 0.03, 0.03×0.02=0.0006 is calculated to obtain the first regular function value; in the second mode, for the error value of 0.04, 0.04+0.02=0.06 is calculated to obtain the second regular function value. The processing results of the two modes are combined and normalized. The regularized vibration sequence finally obtained can better reflect the periodic characteristics of the vibration signal. Through this dual-mode regularization processing method, the influence of non-periodic and nonlinear characteristics on the prediction accuracy is effectively reduced, and the obtained regularized vibration sequence is more suitable for subsequent fault diagnosis analysis. The processing results show that for vibration components with strong periodicity, the regularization effect is more significant, which is conducive to extracting stable fault features.

[0032] In a specific embodiment, the process of executing step S104 may specifically include the following steps: (1) Perform data standardization on the regularized vibration sequence to obtain neural network input data, and convert the data format of the neural network input data through the extreme learning machine algorithm to obtain input layer data; (2) Construct a weight matrix for the input layer data to obtain the input layer weight value, and assign random numbers to the input layer weight value to obtain a randomly generated weight matrix; (3) constructing hidden layer nodes for the randomly generated weight matrix to obtain hidden layer structural parameters, and randomly generating neuron thresholds for the hidden layer structural parameters to obtain hidden layer neuron thresholds; (4) Mapping the hidden layer neuron threshold to the activation function to obtain the hidden layer output data, and calculating the output layer connection weights for the hidden layer output data to obtain the output layer parameter matrix; (5) The network topology of the output layer parameter matrix is ​​constructed to obtain a vibration trend prediction model, where the vibration trend prediction model includes an input layer, a hidden layer, and an output layer.

[0033] Specifically, the regularized vibration sequence first needs to be standardized. The Z-score standardization method is used to convert the data to a distribution with a mean of 0 and a standard deviation of 1. The standardized data is used as the input of the neural network and processed by the extreme learning machine algorithm. During the data format conversion process, the continuous time series data is reorganized into a format that meets the input requirements of the neural network. Each sample contains data of n time steps. n is set to 12, which means that the data of the previous 12 time points are used to predict the value of the next time point. The weight matrix of the input layer data is constructed using a fully connected structure, and the matrix dimension is the input dimension × the number of hidden layer nodes. Taking 12 input nodes and 24 hidden layer nodes as an example, a 12×24 weight matrix is ​​constructed. Each element in the weight matrix is ​​initialized with a random number, and the value range is between [-1,1]. It is randomly generated using a uniform distribution to ensure the randomness and diversity of network training.

[0034] The hidden layer node structure is constructed based on the randomly generated weight matrix. The number of hidden layer nodes is set to twice the number of input layer nodes, that is, 24 nodes, and each node is connected to all nodes in the input layer. The threshold parameter is set for each hidden layer node, and it is also randomly generated using a uniform distribution in the interval [-1,1] as the bias value of the neuron. The hidden layer uses the Sigmoid activation function for nonlinear mapping, and the input data is converted into the output of the hidden layer after weight and threshold calculation. Based on the hidden layer output data, the connection weights of the output layer are calculated. The optimal output layer weights are obtained by solving the linear equations. The solution of the equations is solved using the generalized inverse matrix (Moore-Penrose pseudo-inverse) method to obtain the parameter matrix of the output layer.

[0035] The input layer, hidden layer, and output layer are connected through a weight matrix to construct a complete network topology structure to form a vibration trend prediction model. The input layer of the model receives vibration data of 12 time steps, and finally outputs the predicted value of the next moment through nonlinear transformation of 24 hidden layer nodes.

[0036] Taking the vibration trend prediction of a generator as an example, 1000 continuous regularized vibration sequence data points are processed. After data standardization, the mean value changes from the original 0.45 to 0, and the standard deviation changes from 0.15 to 1. The data is reorganized into 988 training samples, each of which contains 12 input values ​​and 1 target output value. A 12×24 input weight matrix is ​​constructed, and the randomly generated weight values ​​are such as [-0.53, 0.86, -0.21]. The thresholds of the 24 hidden layer nodes are randomly generated as [-0.45, 0.32, 0.78]. The hidden layer output is calculated by the Sigmoid function, and then the 24×1 output layer weight matrix is ​​calculated by the generalized inverse matrix. The root mean square error of the final constructed prediction model on the test data is 0.08, showing good prediction performance. This prediction model construction method based on extreme learning machine makes full use of the characteristics of random generation of input weights and hidden layer thresholds, eliminating the cumbersome weight update process of traditional neural networks while maintaining good prediction accuracy. The model structure is simple and clear, with high computational efficiency, and is suitable for real-time prediction of generator vibration trends.

[0037] In a specific embodiment, the process of executing step S105 may specifically include the following steps: (1) Calculate the predicted value of the vibration trend prediction model to obtain the predicted sequence data, and compare the predicted sequence data with the real-time monitoring data to obtain the error sequence; (2) Performing a square operation on the error sequence to obtain an error square sequence, and summing the error square sequence to obtain an error square sum evaluation function value; (3) Perform cycle counts on the error sum evaluation function value to obtain a cycle error sequence, and perform ratio calculation on the cycle error sequence to obtain a cycle error ratio; (4) For data with a cycle number error ratio greater than two, the regular function value is set to zero to obtain the first type of regular function value, and for data with a cycle number error ratio less than or equal to two, the natural logarithm is calculated to obtain a logarithmic sequence; (5) Perform the product operation of the previous regular function value on the logarithmic sequence to obtain the second-type regular function value, and merge the first-type regular function value and the second-type regular function value to obtain the optimized prediction model.

[0038] Specifically, when performing prediction calculations on the vibration trend prediction model, the vibration data at the current moment is first input into the model to obtain the predicted value at the next moment, and continuous prediction forms prediction sequence data. The prediction sequence data is compared point by point with the monitoring data collected in real time, and the difference between the predicted value and the actual value at each moment is recorded to form a complete error sequence. Each value in the error sequence is squared to eliminate the influence of positive and negative errors and obtain an error square sequence. All values ​​of the error square sequence are accumulated and summed to obtain the error square sum evaluation function value, which reflects the overall prediction accuracy of the prediction model.

[0039] For the error square sum evaluation function value, record the number of cycles predicted by the model each time to form a cycle number error sequence. Count the error values ​​in each cycle, calculate the error ratio between two adjacent cycles, and get the cycle number error ratio, which reflects the trend of the prediction error with the number of cycles.

[0040] Different processing strategies are adopted according to the size of the error ratio of the number of cycles. When the error ratio is greater than 2, the regular function value is directly set to 0 to obtain the first type of regular function value; when the error ratio is less than or equal to 2, the natural logarithm of the ratio is calculated to form a logarithmic sequence. This classification processing method can effectively deal with different degrees of prediction errors. The obtained logarithmic sequence is multiplied by the regular function value obtained by the previous calculation to obtain the second type of regular function value. The first type of regular function value and the second type of regular function value are combined to form a complete optimization prediction model.

[0041] Taking the vibration data of a generator as an example, 1000 data points are taken for prediction. The output sequence of the prediction model is [0.42, 0.45, 0.38, 0.41], and the corresponding actual monitoring value is [0.40, 0.43, 0.39, 0.40]. The calculated error sequence is [0.02, 0.02, -0.01, 0.01]. The square operation of the error sequence is [0.0004, 0.0004, 0.0001, 0.0001], and the sum of the squares of the error is 0.001. In four consecutive cycles, the square sums of the errors are [0.001, 0.0015, 0.0008, 0.0012], and the calculated cycle error ratio is [1.5, 0.53, 1.5]. For the second ratio 0.53 (less than 2), its natural logarithm is calculated to be -0.63. If the previous regular function value is 0.8, the second type regular function value is -0.504; for the first ratio 1.5 (less than 2), the second type regular function value is calculated to be -0.405; for the last ratio 1.5 (less than 2), the second type regular function value is calculated to be -0.405.

[0042] Through this adaptive adjustment mechanism, the prediction model can dynamically adjust parameters according to the real-time prediction error to maintain good prediction accuracy. For example, the average error of the adjusted model in the prediction of the next 100 data points was reduced to 0.008, showing good prediction performance.

[0043] In a specific embodiment, the process of executing step S106 may specifically include the following steps: (1) Predict the sequence data of the optimized prediction model to obtain the prediction output result, and collect the characteristic parameters of the real-time monitoring data to obtain the real-time characteristic sequence; (2) Calculate the numerical deviation of the predicted output results and the real-time feature sequence to obtain feature deviation data, and divide the feature deviation data into threshold ranges to obtain a deviation interval sequence; (3) Match the fault types of the deviation interval sequence to obtain the initial fault characteristics, and verify the initial fault characteristics with historical data to obtain candidate fault types; (4) Calculate the failure probability of the candidate failure types to obtain the failure probability distribution, and perform reliability analysis on the failure probability distribution to obtain the reliability index; (5) Perform fault severity evaluation on the reliability index to obtain the generator fault diagnosis result, where the generator fault diagnosis result includes fault type, fault probability and fault severity.

[0044] Specifically, the optimized prediction model predicts the vibration trend at fixed time intervals (such as 0.1 seconds) and continuously outputs the prediction result sequence. At the same time, characteristic parameters, including key indicators such as amplitude, frequency, and phase, are collected from the real-time operating status of the generator to form a real-time characteristic sequence. The acquisition frequency of the characteristic parameters is synchronized with the prediction frequency to ensure the time correspondence of the data. The difference between the predicted output result at each moment and the corresponding real-time characteristic sequence data is calculated to obtain the characteristic deviation data. According to historical operating experience and equipment specifications, the deviation value is divided into multiple level intervals, such as normal interval (deviation less than 5%), slightly abnormal interval (deviation 5%-15%), moderate abnormal interval (deviation 15%-30%) and severe abnormal interval (deviation greater than 30%), forming a complete deviation interval sequence.

[0045] The deviation interval sequence is matched with the pre-established fault feature library to identify the initial fault features. The fault feature library contains common generator fault types and their typical vibration characteristic patterns. By comparing and verifying with historical fault case data, the most likely candidate fault types are screened out to improve the accuracy of diagnosis. For each candidate fault type, the probability of its occurrence is calculated based on Bayesian probability theory. The probability calculation takes into account multiple factors, including feature matching, historical occurrence frequency, operating conditions, etc. At the same time, the reliability analysis of the fault probability distribution is performed to evaluate the credibility of the diagnosis results and obtain a reliability index reflecting the accuracy of the diagnosis.

[0046] According to the reliability index and the impact of the fault on the equipment operation, the severity of the fault is graded and evaluated, and finally a diagnostic result report is generated including the fault type, fault probability and severity.

[0047] Taking the fault diagnosis of a generator bearing as an example, the vibration trend prediction value output by the prediction model is 0.45mm, the vibration value monitored in real time is 0.52mm, the calculated deviation is 0.07mm, and the deviation percentage is 13.5%, which belongs to the slight abnormal range. This deviation feature is consistent with the typical characteristics of the inner ring failure of the bearing. According to historical data verification, 85 of the past 100 similar failures were confirmed as inner ring failures of the bearing. Based on this calculation, the failure probability is 85% and the reliability index is 0.92. Considering the rapid development of the bearing failure and the potential harm to the equipment, the fault is diagnosed as medium severity and needs to be repaired during the next planned shutdown. Through this systematic fault diagnosis process, equipment failures can be accurately identified.

[0048] The generator fault diagnosis method based on vibration trend prediction in the embodiment of the present application is described above. The generator fault diagnosis system based on vibration trend prediction in the embodiment of the present application is described below. Figure 2 In the embodiment of the present application, an embodiment of a generator fault diagnosis system based on vibration trend prediction includes: The acquisition module 201 is used to collect and process the vibration signal of the generator in real time to obtain historical vibration signal sequence data; A decomposition module 202 is used to perform variational modal decomposition processing on the historical vibration signal sequence data to obtain multiple intrinsic modal function subsequences; The processing module 203 is used to perform regularization processing on the subsequence of the intrinsic mode function, wherein the subsequence is subjected to dual-mode processing by a preset regular function to obtain a regularized vibration sequence, wherein the dual-mode processing includes: when the absolute value of the time error is less than 5% of the regular function value, the regular function value is equal to the product of the time error and the regular function period value for processing; when the absolute value of the time error is greater than or equal to 5% of the regular function value, the regular function value is equal to the sum of the time error and the regular function period value for processing; A training module 204 is used to perform prediction model training on the regularized vibration sequence by using an extreme learning machine algorithm to obtain a vibration trend prediction model, wherein the prediction model includes an input layer, a hidden layer and an output layer, and the weight value of the input layer and the neuron threshold of the hidden layer are randomly generated; The adjustment module 205 is used to perform adaptive adjustment processing on the vibration trend prediction model through the error square sum evaluation function to obtain an optimized prediction model, wherein when the cycle number error ratio is greater than two, the regular function value is set to zero; when the cycle number error ratio is less than or equal to two, the regular function value is equal to the product of the previous regular function value and the natural logarithm of the error ratio; The comparison module 206 is used to compare and analyze the output result of the optimized prediction model with the real-time monitoring data to obtain the generator fault diagnosis result.

[0049] Through the cooperation of the above components, the complete historical vibration signal sequence data is obtained by real-time acquisition and processing of the generator vibration signal, ensuring the accuracy and timeliness of the original data. The variational mode decomposition processing method is used to decompose the historical vibration signal sequence data to obtain multiple intrinsic mode function subsequences, effectively separating the different frequency components in the signal and enhancing the analysis ability of complex vibration signals. By regularizing the intrinsic mode function subsequences, introducing a dual-mode processing mechanism, and using different calculation methods according to the absolute value of the time error, the influence of non-periodic and nonlinear characteristics on the prediction accuracy is effectively reduced. The extreme learning machine algorithm is used to train the prediction model of the regularized vibration sequence. The algorithm simplifies the training process of the traditional neural network and improves the calculation efficiency by randomly generating the input layer weight value and the hidden layer neuron threshold. The error square sum evaluation function is introduced to adaptively adjust the vibration trend prediction model, and the regular function value is dynamically adjusted according to the size of the cycle number error ratio, which enhances the model's adaptability to changes in vibration trends. Finally, by comparing and analyzing the output results of the optimized prediction model with the real-time monitoring data, it is possible to promptly detect abnormal changes in the generator's operating status, accurately determine the type of fault, achieve early warning and accurate diagnosis of generator faults, and improve the safety and reliability of equipment operation.

[0050] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A generator fault diagnosis method based on vibration trend prediction, characterized in that: The generator fault diagnosis method based on vibration trend prediction includes: Collect and process the generator vibration signal in real time to obtain historical vibration signal sequence data; Performing variational mode decomposition processing on the historical vibration signal sequence data to obtain a plurality of intrinsic mode function subsequences; The subsequence of the intrinsic modal function is subjected to regularization processing, wherein the subsequence is subjected to dual-mode processing by a preset regular function to obtain a regularized vibration sequence, wherein the dual-mode processing includes: when the absolute value of the time error is less than 5% of the regular function value, the regular function value is equal to the product of the time error and the regular function period value for processing; when the absolute value of the time error is greater than or equal to 5% of the regular function value, the regular function value is equal to the sum of the time error and the regular function period value for processing; The regularized vibration sequence is trained with an extreme learning machine algorithm to form a prediction model, where the prediction model includes an input layer, a hidden layer, and an output layer, and the weight value of the input layer and the neuron threshold of the hidden layer are randomly generated; The vibration trend prediction model is adaptively adjusted through an error square sum evaluation function to obtain an optimized prediction model, wherein when the cycle number error ratio is greater than two, the regular function value is set to zero; when the cycle number error ratio is less than or equal to two, the regular function value is equal to the product of the previous regular function value and the natural logarithm of the error ratio; The output result of the optimized prediction model is compared and analyzed with the real-time monitoring data to obtain the generator fault diagnosis result.

2. The generator fault diagnosis method based on vibration trend prediction according to claim 1 is characterized in that: The real-time collection and processing of the generator vibration signal to obtain historical vibration signal sequence data includes: The sampling frequency of the generator vibration signal is set to obtain the vibration signal sampling parameters, and the vibration signal sampling parameters are tested by Nyquist sampling theorem to obtain effective sampling parameters; Acquiring vibration signals on the effective sampling parameters to obtain original vibration signal data, and performing wavelet denoising on the original vibration signal data to obtain denoised vibration signal data; Performing time domain feature extraction on the noise-reduced vibration signal data to obtain a time domain feature sequence, and performing frequency domain conversion processing on the time domain feature sequence to obtain a frequency domain feature sequence; Performing feature fusion processing on the time domain feature sequence and the frequency domain feature sequence to obtain a fused feature sequence, and performing normalization processing on the fused feature sequence to obtain a standardized feature sequence; The standardized feature sequence is divided into sliding time windows to obtain time series segments, and the time series segments are checked for data integrity to obtain historical vibration signal sequence data.

3. The generator fault diagnosis method based on vibration trend prediction according to claim 1 is characterized in that: The variational modal decomposition process is performed on the historical vibration signal sequence data to obtain a plurality of intrinsic modal function subsequences, including: Performing data segmentation processing on the historical vibration signal sequence data to obtain a plurality of vibration signal sub-segments, and performing spectrum analysis on the plurality of vibration signal sub-segments to obtain spectrum feature data; Calculating the number of variational mode decomposition layers of the frequency spectrum feature data to obtain a decomposition layer number parameter, and estimating the center frequency of the decomposition layer number parameter to obtain a center frequency value of each layer; Setting bandwidth parameters for the center frequency values ​​of each layer to obtain decomposition bandwidth parameters, and constructing secondary penalty terms for the decomposition bandwidth parameters to obtain penalty constraint conditions; Constructing the penalty constraint condition using a Lagrangian operator to obtain an augmented Lagrangian expression, and performing alternating direction multiplier iteration on the augmented Lagrangian expression to obtain a decomposition coefficient sequence; The decomposition coefficient sequence is subjected to Hilbert transform processing to obtain an instantaneous frequency sequence, and the instantaneous frequency sequence is subjected to subsequence reconstruction to obtain a plurality of intrinsic mode function subsequences.

4. The generator fault diagnosis method based on vibration trend prediction according to claim 1, characterized in that: The regularization processing is performed on the subsequence of the intrinsic modal function, wherein the subsequence is subjected to dual-mode processing by a preset regular function to obtain a regularized vibration sequence, wherein the dual-mode processing includes: when the absolute value of the time error is less than 5% of the regular function value, the regular function value is equal to the product of the time error and the regular function period value for processing; when the absolute value of the time error is greater than or equal to 5% of the regular function value, the regular function value is equal to the sum of the time error and the regular function period value for processing, including: Performing sliding time window processing on the intrinsic mode function subsequence to obtain a time window sequence, and performing time error calculation on the time window sequence to obtain a time error sequence; Performing an absolute value operation on the time error sequence to obtain an error absolute value sequence, and performing a threshold comparison on the error absolute value sequence, marking the error sequence greater than the threshold as a first type of error, and marking the error sequence less than or equal to the threshold as a second type of error, to obtain an error classification result; Performing regular function value calculation on the error classification result to obtain an initial regular function value, and performing period analysis on the initial regular function value to obtain a regular function period value; Performing a product operation on the sequence of the absolute value of the error that is less than 5% of the regular function value to obtain a first pattern regular function value, and performing a sum operation on the sequence of the absolute value of the error that is greater than or equal to 5% of the regular function value to obtain a second pattern regular function value; The first mode regular function value and the second mode regular function value are subjected to sequence merging processing to obtain a dual-mode processing result, and the dual-mode processing result is subjected to normalization processing to obtain a regularized vibration sequence.

5. The generator fault diagnosis method based on vibration trend prediction according to claim 1, characterized in that: The predictive model training is performed on the regularized vibration sequence by using an extreme learning machine algorithm to obtain a vibration trend prediction model, wherein the prediction model includes an input layer, a hidden layer and an output layer, and the weight value of the input layer and the neuron threshold of the hidden layer are randomly generated, including: Performing data standardization processing on the regularized vibration sequence to obtain neural network input data, and performing data format conversion on the neural network input data through an extreme learning machine algorithm to obtain input layer data; Constructing a weight matrix for the input layer data to obtain input layer weight values, and performing random number assignment on the input layer weight values ​​to obtain a randomly generated weight matrix; Constructing hidden layer nodes for the randomly generated weight matrix to obtain hidden layer structural parameters, and randomly generating neuron thresholds for the hidden layer structural parameters to obtain hidden layer neuron thresholds; Performing activation function mapping on the hidden layer neuron threshold to obtain hidden layer output data, and performing output layer connection weight calculation on the hidden layer output data to obtain an output layer parameter matrix; A network topology is constructed for the output layer parameter matrix to obtain a vibration trend prediction model, wherein the vibration trend prediction model includes the input layer, the hidden layer and the output layer.

6. The generator fault diagnosis method based on vibration trend prediction according to claim 1, characterized in that: The vibration trend prediction model is adaptively adjusted through the error square sum evaluation function to obtain an optimized prediction model, wherein when the cycle number error ratio is greater than two, the regular function value is set to zero; when the cycle number error ratio is less than or equal to two, the regular function value is equal to the product of the previous regular function value and the natural logarithm of the error ratio, including: Calculating predicted values ​​of the vibration trend prediction model to obtain predicted sequence data, and comparing the predicted sequence data with real-time monitoring data to obtain an error sequence; Performing a square operation on the error sequence to obtain an error square sequence, and performing a sum calculation on the error square sequence to obtain an error square sum evaluation function value; Performing cycle count on the error square sum evaluation function value to obtain a cycle error sequence, and performing ratio calculation on the cycle error sequence to obtain a cycle error ratio; Performing a regular function value zeroing process on the data whose cycle number error ratio is greater than two to obtain a first-type regular function value, and performing a natural logarithm calculation on the data whose cycle number error ratio is less than or equal to two to obtain a logarithmic sequence; The last regular function value product operation is performed on the logarithmic sequence to obtain a second type of regular function value, and the first type of regular function value and the second type of regular function value are combined to obtain the optimized prediction model.

7. The generator fault diagnosis method based on vibration trend prediction according to claim 1, characterized in that: The optimized prediction model output result is compared and analyzed with the real-time monitoring data to obtain the generator fault diagnosis result, including: Performing sequence data prediction on the optimized prediction model to obtain a prediction output result, and collecting characteristic parameters on the real-time monitoring data to obtain a real-time characteristic sequence; Performing numerical deviation calculation on the predicted output result and the real-time feature sequence to obtain feature deviation data, and dividing the feature deviation data into threshold ranges to obtain a deviation interval sequence; Performing fault type matching on the deviation interval sequence to obtain initial fault features, and performing historical data verification on the initial fault features to obtain candidate fault types; Calculating the failure probability of the candidate failure types to obtain a failure probability distribution, and performing reliability analysis on the failure probability distribution to obtain a reliability index; A fault severity evaluation is performed on the reliability index to obtain a generator fault diagnosis result, wherein the generator fault diagnosis result includes a fault type, a fault probability and a fault severity.

8. A generator fault diagnosis system based on vibration trend prediction, used to implement the generator fault diagnosis method based on vibration trend prediction as described in any one of claims 1 to 7, characterized in that: The generator fault diagnosis system based on vibration trend prediction includes: The acquisition module is used to collect and process the vibration signal of the generator in real time to obtain historical vibration signal sequence data; A decomposition module, used for performing variational modal decomposition processing on the historical vibration signal sequence data to obtain a plurality of intrinsic modal function subsequences; A processing module is used to perform regular processing on the subsequence of the intrinsic mode function, wherein the subsequence is subjected to dual-mode processing by a preset regular function to obtain a regular vibration sequence, wherein the dual-mode processing includes: when the absolute value of the time error is less than 5% of the regular function value, the regular function value is equal to the product of the time error and the regular function period value for processing; when the absolute value of the time error is greater than or equal to 5% of the regular function value, the regular function value is equal to the sum of the time error and the regular function period value for processing; A training module is used to perform prediction model training on the regularized vibration sequence through an extreme learning machine algorithm to obtain a vibration trend prediction model, wherein the prediction model includes an input layer, a hidden layer and an output layer, and the weight value of the input layer and the neuron threshold of the hidden layer are randomly generated; An adjustment module is used to adaptively adjust the vibration trend prediction model through an error square sum evaluation function to obtain an optimized prediction model, wherein when the cycle number error ratio is greater than two, the regular function value is set to zero; when the cycle number error ratio is less than or equal to two, the regular function value is equal to the product of the previous regular function value and the natural logarithm of the error ratio; The comparison module is used to compare and analyze the output result of the optimized prediction model with the real-time monitoring data to obtain the generator fault diagnosis result.

Images