Generator fault diagnosis method based on vibration trend prediction and related product

Abstract

The invention discloses a generator fault diagnosis method based on vibration trend prediction and a related product, and belongs to the technical field of generator fault diagnosis. According to the generator fault diagnosis method based on vibration trend prediction provided by the invention, the historical vibration signal sequence of the generator is decomposed into a plurality of subsequences, and random function generation and residual function calculation are performed on each subsequence, so that feature information in the vibration signal can be carefully captured; therefore, the accuracy of fault diagnosis is improved. A neural network model is used for prediction, and a strategy of iterative loop and residual function minimization is combined, so that a real vibration trend can be gradually approached, and the diagnosis accuracy is improved; through iterative loop and multiple random function generation and selection of an optimal result, nonlinearity and aperiodicity of a vibration signal are overcome, and prediction stability and reliability are enhanced. The introduction of the residual error sequence has the effects of smoothing the vibration signal and reducing noise interference, and the prediction stability is improved.

Description

Technical Field

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

[0002] Vibration signal analysis is a key technology in motor fault diagnosis, revealing abnormalities within the motor. When a turn-to-turn short circuit occurs, the uneven current distribution causes changes in the motor's dynamic response, which manifests itself in the vibration signal. Therefore, monitoring and analyzing vibration signals is crucial for the early detection and diagnosis of motor faults.

[0003] However, the inherent non-periodic nature of vibration signals complicates their direct analysis and prediction. Non-periodicity means that the frequency and amplitude of vibration signals vary over time, without a fixed pattern. Non-linearity also means that the generation and propagation of vibration signals may be affected by a variety of nonlinear factors, making signal prediction and analysis even more difficult. Summary of the Invention

[0004] The purpose of the present invention is to provide a generator fault diagnosis method and related products based on vibration trend prediction, so as to overcome the problem of inaccurate prediction caused by the non-periodicity and nonlinearity of the generator vibration signal in the prior art.

[0005] The present invention solves the above technical problems through the following technical solutions: A generator fault diagnosis method based on vibration trend prediction includes the following steps: S1. Obtain the historical vibration signal sequence of the generator, divide the historical vibration signal sequence of the generator into a prediction set sequence and a validation set sequence according to a preset ratio; input the prediction set sequence into the trained neural network model for prediction, obtain the predicted value, and calculate the baseline prediction error SSE based on the predicted value and the actual value in the validation set sequence. S2. Decompose the prediction set sequence to obtain several subsequences IMF1~IMF K , IMF1 is the first subsequence obtained by decomposition, IMF K is the Kth subsequence obtained by decomposition, and K is the total number of subsequences obtained by decomposition; S3. Generate a random function for each subsequence M times through an iterative loop. Calculate a residual function based on the random function of each subsequence that conforms to the periodic function rule. Select the random function corresponding to the minimum value of the residual function of each subsequence as the subsequence of the Y-th prediction set. S4. Sum the minimum values ​​of the residual functions of the K subsequences to obtain the residual sequence Q; S5. Input the 1st to Kth subsequences and the residual sequence Q in the Yth prediction set into the trained neural network model for prediction to obtain the Yth prediction value; Y is the number of prediction sets generated based on the random function; S6. Based on the Y-th predicted value and the actual value in the validation set sequence, calculate the prediction error SSE(Y). SSE(Y) is the prediction error of the Y-th prediction set generated based on the random function. Determine whether the prediction error SSE(Y) is less than the baseline prediction error SSE. If the determination result is yes, then diagnose the generator fault based on the Y-th predicted value. If the determination result is no, then proceed to step S7. S7. Based on the correction formula, correct the 1st to Kth subsequences in the Y-th prediction set, let Y = Y + 1, and obtain the subsequences in the Y-th prediction set. Repeat steps S5 and S6.

[0006] A further improvement of the present invention is that step S3 specifically comprises: Step S3 is as follows: S31. Preset the number of cycles M, where M is at least 100,000. Let i=0, j=0, where i is the i-th subsequence among a plurality of subsequences, and j is the j-th cycle of the i-th subsequence. S32. Let i = i + 1; S33, let j=j+1, based on the subsequence IMF i , generate the corresponding random function f(t) ij ; S34. Determine the random function f(t). ij Does it conform to the law of periodic function? If it is, then based on the random function f(t) ij , calculate the random function f(t) ij The corresponding residual function Q(t) ij , execute step S35; if the judgment is no, directly execute step S35; S35, determine whether j is equal to M, if not, return to step S33; if yes, calculate the subsequence IMF i The value of all residual functions, the residual function with the smallest value is Q(t) min , Q(t) min The corresponding random function f(t) ij For the i-th subsequence in the Y-th prediction set, Y=1, execute step S36; S36, determine whether i is equal to K. If so, execute step S4; if not, set j = 0 and return to step S32; Step S4 specifically involves calculating the residual function Q(t) that minimizes all values ​​obtained in step S35. min The sum of these two values ​​yields the residual sequence Q.

[0007] A further improvement of the present invention is that: the trained neural network model is an ELM (Extreme Learning Machine) model; The baseline prediction error SSE is as follows:

[0008] Where N is the number of subsequences in the validation set sequence; is the nth actual value in the validation set sequence, is the nth value in the predicted value.

[0009] A further improvement of the present invention is that: in step S1, the preset ratio is 9:1; in step S2, the VMD (Variational Mode Decomposition) method is used to decompose the prediction set sequence.

[0010] A further improvement of the present invention is: based on the subsequence IMF i Generate the corresponding random function f(t). ij Specifically: Let IMF i =

x 1i , x 2i , x 3i , ..., x 9Ni

t 1i , t 2i , t 3i , ..., t 9Ni

【rand(0.1x 1i ~1.9x 1i ), rand (0.1x 2i ~1.9x 2i ), rand (0.1x 3i ~1.9x 3i ),…,rand(0.1x 9Ni ~1.9x 9Ni ) Where rand() is a random function; x 1i , x 2i , x 3i ,…,x 9Ni Subsequence IMF i The 1st, 2nd, 3rd, ..., 9Nth values ​​of t 1i , t 2i , t 3i, ..., t 9Ni These are random functions f(t). ij The 1st, 2nd, 3rd, …, 9Nth values ​​of .

[0011] A further improvement of the present invention is that the law of the periodic function is specifically:

[0012] Among them, t wi For subsequence IMF i The w-th value of the random function; t (w+T)i is the subsequence IMF i The w+Tth value of the random function; T is the period; is the regularity parameter.

[0013] A further improvement of the present invention is that the residual function Q(t) ij Specifically: Q(t) ij = IMF i - f(t) ij =

x 1i -t 1i , x 2i -t 2i , x 3i -t 3i ,…,x 9Ni -t 9Ni

[0014] The residual sequence Q is specifically: Q=[x 11 -t 11 +x 12 -t 12 +x 13 -t 13 +…+x 1k -t 1k , x 21 -t 21 +x 22 -t 22 +x 23 -t 23 +…+x 2k -t 2k , x 31 -t 31 +x 32 -t 32 +x 33 -t 33 +…+x 3k -t 3k ,…,x9N1 -t 9N1 +x 9N2 -t 9N2 +x 9N3 -t 9N3 +…+x 9Nk -t 9Nk 】 Among them, x 11 -t 11 、x 12 -t 12 、x 13 -t 13 ,…,x 1k -t 1k They are subsequences IMF1, IMF2, IMF3, ..., IMF k The first value of the residual function with the smallest value; x 21 -t 21 、x 22 -t 22 、x 23 -t 23 ,…,x 2k -t 2k The subsequences are IMF1, IMF2, IMF3, ..., IMF. k The second value of the residual function with the smallest value; x 31 -t 31 、x 32 -t 32 、x 33 -t 33 ,…,x 3k -t 3k They are subsequences IMF1, IMF2, IMF3, ..., IMF k The third value of the residual function with the smallest corresponding value; x 9N1 -t 9N1 、x 9N2 -t 9N2 、x 9N3 -t 9N3 ,…,x 9Nk -t 9Nk They are subsequences IMF1, IMF2, IMF3, ..., IMF k The 9Nth value of the residual function with the smallest corresponding value.

[0015] A further improvement of the present invention is that the correction formula is specifically:

[0016] Where, f(t)' ij is the corrected random function; sin() is the sine function; ln is the natural logarithm.

[0017] The present invention also provides a computer device including a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement the steps of the generator fault diagnosis method based on vibration trend prediction as described above.

[0018] The present invention also provides a computer-readable storage medium storing at least one instruction, which, when executed by a processor, implements the steps of the generator fault diagnosis method based on vibration trend prediction as described above.

[0019] Compared with the prior art, the present invention has the following positive effects: The generator fault diagnosis method based on vibration trend prediction provided by this invention can automatically adjust the subsequences in the prediction set according to the prediction error, and continuously optimize the prediction results through iterative correction, adapting to the fault diagnosis needs of different generators and different operating conditions. By decomposing the historical vibration signal sequence of the generator into multiple subsequences and performing random function generation and residual function calculation on each subsequence, the characteristic information in the vibration signal can be captured more meticulously, thereby improving the accuracy of fault diagnosis. Using a neural network model for prediction, combined with iterative loops and residual function minimization strategies, the method can gradually approximate the actual vibration trend, improving the accuracy of diagnosis. Through iterative loops and multiple random function generation, and selecting the optimal result, the nonlinearity and non-periodicity of the vibration signal are overcome, enhancing the stability and reliability of the prediction. The introduction of the residual sequence plays a role in smoothing the vibration signal and reducing noise interference, improving the stability of the prediction. Through vibration trend prediction, abnormal changes in the vibration signal can be captured in the early stage of generator fault occurrence, thereby achieving early warning of generator faults, helping to promptly detect and handle potential faults, and avoiding equipment damage and production interruption caused by further fault development. Attached Figure Description

[0020] The accompanying drawings are provided to further understand the invention and constitute a part of this invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0021] Figure 1 The figure is a flow chart of a generator fault diagnosis method based on vibration trend prediction according to the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This is an explanation of the present invention and not a limitation thereof.

[0023] See also Figure 1 A generator fault diagnosis method based on vibration trend prediction includes the following steps: S1. Obtain the historical vibration signal sequence of the generator, divide the historical vibration signal sequence of the generator into a prediction set sequence and a validation set sequence according to a preset ratio; input the prediction set sequence into the trained neural network model for prediction, obtain the predicted value, and calculate the baseline prediction error SSE based on the predicted value and the actual value in the validation set sequence. S2. Decompose the prediction set sequence to obtain several subsequences IMF1~IMF K , IMF1 is the first subsequence obtained by decomposition, IMF K is the Kth subsequence obtained by decomposition, and K is the total number of subsequences obtained by decomposition; S3. Through iterative loops, generate random functions M times for each subsequence. Based on the random function that conforms to the periodic function law of each subsequence, calculate the residual function. Select the random function corresponding to the minimum value of the residual function of each subsequence as the subsequence of the Y-th prediction set. S4. Sum the minimum values ​​of the residual functions of the K subsequences to obtain the residual sequence Q; S5. Input the 1st to Kth subsequences and the residual sequence Q in the Yth prediction set into the trained neural network model for prediction to obtain the Yth prediction value; Y is the number of prediction sets generated based on the random function; S6. Based on the Y-th predicted value and the actual value in the validation set sequence, calculate the prediction error SSE(Y). SSE(Y) is the prediction error of the Y-th prediction set generated based on the random function. Determine whether the prediction error SSE(Y) is less than the baseline prediction error SSE. If the determination result is yes, then diagnose the generator fault based on the Y-th predicted value. If the determination result is no, then proceed to step S7. S7. Based on the correction formula, correct the 1st to Kth subsequences in the Y-th prediction set, let Y = Y + 1, and obtain the subsequences in the Y-th prediction set. Repeat steps S5 and S6.

[0024] The generator fault diagnosis method based on vibration trend prediction provided by this invention can automatically adjust the subsequences in the prediction set according to the prediction error, and continuously optimize the prediction results through iterative correction, adapting to the fault diagnosis needs of different generators and different operating conditions. By decomposing the historical vibration signal sequence of the generator into multiple subsequences and performing random function generation and residual function calculation on each subsequence, the characteristic information in the vibration signal can be captured more meticulously, thereby improving the accuracy of fault diagnosis. Using a neural network model for prediction, combined with iterative loops and residual function minimization strategies, the method can gradually approximate the actual vibration trend, improving the accuracy of diagnosis. Through iterative loops and multiple random function generation, and selecting the optimal result, the nonlinearity and non-periodicity of the vibration signal are overcome, enhancing the stability and reliability of the prediction. The introduction of the residual sequence plays a role in smoothing the vibration signal and reducing noise interference, improving the stability of the prediction. Through vibration trend prediction, abnormal changes in the vibration signal can be captured in the early stage of generator fault occurrence, thereby achieving early warning of generator faults, helping to promptly detect and handle potential faults, and avoiding equipment damage and production interruption caused by further fault development.

[0025] Specifically, step S3 is as follows: S31. Preset the number of cycles M, where M is at least 100,000. Let i=0, j=0, where i is the i-th subsequence among a plurality of subsequences, and j is the j-th cycle of the i-th subsequence. S32. Let i = i + 1; S33, let j=j+1, based on the subsequence IMF i , generate the corresponding random function f(t) ij ; S34. Determine the random function f(t). ij Does it conform to the law of periodic function? If it is, then based on the random function f(t) ij , calculate the random function f(t) ij The corresponding residual function Q(t) ij , execute step S35; if the judgment is no, directly execute step S35; S35, determine whether j is equal to M, if not, return to step S33; if yes, calculate the subsequence IMF i The value of all residual functions, the residual function with the smallest value is Q(t) min , Q(t) min The corresponding random function f(t) ij For the i-th subsequence in the Y-th prediction set, Y=1, execute step S36; S36. Determine if i is equal to K. If yes, proceed to step S4; if no, set j=0 and return to step S32. Step S4 specifically involves calculating the residual function Q(t) that minimizes all values ​​obtained in step S35. min The sum of , we get the residual sequence Q.

[0026] In step S34, regardless of the random function f(t) ij Whether it conforms to the law of periodic function, step S35 must be executed to enter the next loop; M is at least equal to 100000. Since the number of loops M is set large enough, it is assumed that the random function f(t) will exist within M loops. ij Conforms to the cyclical law.

[0027] Specifically, the trained neural network model is called an ELM model; The baseline prediction error SSE is as follows:

[0028] Where N is the number of subsequences in the validation set sequence; is the nth actual value in the validation set sequence, is the nth value in the predicted value.

[0029] Specifically, in step S1, the preset ratio is 9:1; in step S2, the VMD method is used to decompose the prediction set sequence.

[0030] Specifically, based on subsequence IMF i Generate the corresponding random function f(t). ij Specifically: Let IMF i =

x 1i , x 2i , x 3i ,…,x 9Ni

t 1i , t 2i , t 3i , ..., t 9Ni

【rand(0.1x 1i ~1.9x 1i ), rand (0.1x 2i ~1.9x 2i ), rand (0.1x 3i ~1.9x 3i ),…,rand(0.1x 9Ni~1.9x 9Ni ) Where rand() is a random function; x 1i , x 2i , x 3i ,…,x 9Ni Subsequence IMF i The 1st, 2nd, 3rd, ..., 9Nth values ​​of t 1i , t 2i , t 3i , ..., t 9Ni These are random functions f(t). ij The 1st, 2nd, 3rd, …, 9Nth values ​​of .

[0031] Specifically, the law of periodic functions is:

[0032] Among them, t wi is the subsequence IMF i The w-th value of the random function; t (w+T)i is the subsequence IMF i The w+Tth value of the random function; T is the period; is the regularity parameter.

[0033] Specifically, the residual function Q(t) ij Specifically: Q(t) ij = IMF i - f(t) ij =

x 1i -t 1i , x 2i -t 2i , x 3i -t 3i ,…,x 9Ni -t 9Ni

[0034] The residual sequence Q is specifically: Q=[x 11 -t 11 +x 12 -t 12 +x 13 -t 13 +…+x 1k -t 1k , x 21 -t 21 +x 22 -t 22 +x 23-t 23 +…+x 2k -t 2k , x 31 -t 31 +x 32 -t 32 +x 33 -t 33 +…+x 3k -t 3k ,…,x 9N1 -t 9N1 +x 9N2 -t 9N2 +x 9N3 -t 9N3 +…+x 9Nk -t 9Nk 】 Among them, x 11 -t 11 、x 12 -t 12 、x 13 -t 13 ... x 1k -t 1k They are subsequences IMF1, IMF2, IMF3, ..., IMF k The first value of the residual function with the smallest corresponding value; x 21 -t 21 、x 22 -t 22 、x 23 -t 23 ... x 2k -t 2k The subsequences are IMF1, IMF2, IMF3, ..., IMF. k The second value of the residual function with the smallest corresponding value; x 31 -t 31 、x 32 -t 32 、x 33 -t 33 ,…,x 3k -t 3k They are subsequences IMF1, IMF2, IMF3, ..., IMF k The third value of the residual function with the smallest corresponding value; x 9N1 -t 9N1 、x 9N2 -t 9N2 、x 9N3 -t 9N3 ... x 9Nk -t 9Nk The subsequences are IMF1, IMF2, IMF3, ..., IMF. kThe 9Nth value of the residual function with the smallest value.

[0035] Specifically, the correction formula is:

[0036] Where, f(t)' ij is the modified random function; sin() is the sine function; ln is the natural logarithm.

[0037] To further verify the advantages of this method, the method of this application and the traditional prediction method were used to predict the historical vibration signals of the generator. The traditional prediction method is to decompose the historical vibration signal prediction set sequence through VMD and then directly input it into the ELM model for prediction. In the experiment of this application, MAPE (Mean Absolute Percentage Error) was selected as the evaluation standard for each model. The prediction results (i.e., MAPE values) are shown in Table 1: Table 1 Prediction Results

[0038] It can be seen that the prediction accuracy of the generator vibration signal using this method is higher and more reliable.

[0039] Based on the same inventive concept, an embodiment of the present application provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of a generator fault diagnosis method based on vibration trend prediction are implemented. The memory may include internal memory, such as a high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industrial standard architecture bus, a peripheral component interconnect standard bus, an extended industrial standard architecture bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the programs may include program code, and the program code includes computer operating instructions. The memory may include internal memory and non-volatile memory, and provides instructions and data to the processor.

[0040] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the steps of the generator fault diagnosis method based on vibration trend prediction. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include RAM (Random Access Memory) and / or cache memory, etc. The non-volatile memory may include ROM (Read-Only Memory), hard disk, flash memory, optical disk, magnetic disk, etc.

[0041] Those skilled in the art will understand that embodiments of the present invention can be provided as methods or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM (Compact Disc Read-Only Memory), optical storage, etc.) containing computer-usable program code.

[0042] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as 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 apparatus or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0043] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer device or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0044] These computer program instructions can also be loaded onto a computer device or other programmable data processing device so that a series of operating steps are executed on the computer device or other programmable device to produce a process implemented by the computer device, thereby providing instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0045] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0046] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A generator fault diagnosis method based on vibration trend prediction, characterized in that: The following steps are involved: S1. Obtain the historical vibration signal sequence of the generator, and divide the historical vibration signal sequence of the generator into a prediction set sequence and a verification set sequence according to a preset ratio; input the prediction set sequence into the trained neural network model for prediction to obtain a prediction value, and calculate the benchmark prediction error SSE based on the prediction value and the actual value in the verification set sequence; S2. Decompose the prediction set sequence to obtain several subsequences IMF1~IMF K , IMF1 is the first subsequence obtained by decomposition, IMF K is the Kth subsequence obtained by decomposition, and K is the total number of subsequences obtained by decomposition; S3, through iterative loops, generate random functions for each subsequence M times in turn, and calculate the residual function based on the random function of each subsequence that conforms to the law of periodic functions; select the random function corresponding to the minimum value of the residual function of each subsequence as the subsequence of the Yth prediction set; S4, summing up the minimum values ​​of the residual functions of the K subsequences to obtain a residual sequence Q; S5, input the 1st to Kth subsequences and the residual sequence Q in the Yth prediction set into the trained neural network model for prediction, and obtain the Yth prediction value; Y is the number of prediction sets generated based on the random function; S6. Based on the Yth predicted value and the actual value in the validation set sequence, calculate the prediction error SSE(Y), where SSE(Y) is the prediction error of the Yth prediction set generated based on the random function; determine whether the prediction error SSE(Y) is less than the reference prediction error SSE. If the determination result is yes, diagnose the generator fault based on the Yth predicted value; if the determination result is no, execute step S7; S7. Based on the correction formula, correct the 1st to Kth subsequences in the Yth prediction set, set Y=Y+1, obtain the subsequences in the Yth prediction set, and repeat steps S5 and S6.

2. A generator fault diagnosis method based on vibration trend prediction according to claim 1, characterized in that: Step S3 is specifically as follows: S31, preset the number of cycles M, M is at least equal to 100000, let i=0, j=0, i is the i-th subsequence among a plurality of subsequences, j is the j-th cycle of the i-th subsequence; S32, let i=i+1; S33, let j = j + 1, based on the subsequence IMF i , generate the corresponding random function f(t) ij ; S34. Determine the random function f(t) ij Does it conform to the law of periodic function? If it is judged to be yes, then based on the random function f(t) ij , calculate the random function f(t) ij The corresponding residual function Q(t) ij , execute step S35; if the judgment is no, directly execute step S35; S35, determine whether j is equal to M, if not, return to step S33; If the judgment is yes, calculate the subsequence IMF i The value of all residual functions, the residual function with the smallest value is Q(t) min , Q(t) min The corresponding random function f(t) ij For the i-th subsequence in the Y-th prediction set, Y=1, execute step S36; S36, determine whether i is equal to K, if it is, execute step S4; if it is not, set j=0 and return to step S32; Step S4 is specifically: calculating the residual function Q(t) with the minimum value among all values ​​obtained in step S35 min The sum of , we get the residual sequence Q.

3. A generator fault diagnosis method based on vibration trend prediction according to claim 2, characterized in that: The trained neural network model is the ELM model; The baseline prediction error SSE is specifically: Where N is the number of subsequences in the validation set sequence; is the nth actual value in the validation set sequence, is the nth value in the prediction.

4. A generator fault diagnosis method based on vibration trend prediction according to claim 3, characterized in that: In step S1, the preset ratio is 9:1; in step S2, the VMD method is used to decompose the prediction set sequence.

5. A generator fault diagnosis method based on vibration trend prediction according to claim 4, characterized in that: Subsequence-based IMF i , generate the corresponding random function f(t) ij Specifically: IMF i =【x 1i , x 2i , x 3i , …, x 9Ni 】, IMF i is a number of subsequences IMF1~IMF K The ith subsequence in , then f(t) ij =【t 1i , t 2i , t 3i ,…,t 9Ni 】=【rand(0.1x 1i ~1.9x 1i ), rand (0.1x 2i ~1.9x 2i ), rand (0.1x 3i ~1.9x 3i ), …, rand (0.1x 9Ni ~1.9x 9Ni )]; Among them, rand() is a random function; x 1i , x 2i , x 3i , …, x 9Ni The subsequences IMF are i The 1st, 2nd, 3rd, ..., 9Nth values ​​of t 1i , t 2i , t 3i ,…,t 9Ni They are random functions f(t) ij The 1st, 2nd, 3rd, …, 9Nth values ​​of .

6. A generator fault diagnosis method based on vibration trend prediction according to claim 5, characterized in that: The specific rules of periodic functions are: Among them, t wi IMF is the subsequence i The w-th value of the random function; t (w+T)i is the subsequence IMF i The w+Tth value of the random function; T is the period; is the regularity parameter.

7. The generator fault diagnosis method based on vibration trend prediction according to claim 6, characterized in that: Residual function Q(t) ij Specifically: Q(t) ij = IMF i - f(t) ij = 【x 1i -t 1i ,x 2i -t 2i ,x 3i -t 3i ,…,x 9Ni -t 9Ni 】 The value A of the residual function is specifically: The residual sequence Q is specifically: Q=【x 11 -t 11 +x 12 -t 12 +x 13 -t 13 +…+x 1k -t 1k ,x 21 -t 21 +x 22 -t 22 +x 23 -t 23 +…+x 2k -t 2k ,x 31 -t 31 +x 32 -t 32 +x 33 -t 33 +…+x 3k -t 3k ,…,x 9N1 -t 9N1 +x 9N2 -t 9N2 +x 9N3 -t 9N3 +…+x 9Nk -t 9Nk 】 Among them, x 11 -t 11 、x 12 -t 12 、x 13 -t 13 , …, x 1k -t 1k They are subsequences IMF1, IMF2, IMF3, …, IMF k The first value of the residual function with the smallest value; x 21 -t 21 、x 22 -t 22 、x 23 -t 23 , …, x 2k -t 2k They are subsequences IMF1, IMF2, IMF3, …, IMF k The second value of the residual function with the smallest value; x 31 -t 31 、x 32 -t 32 、x 33 -t 33 , …, x 3k -t 3k They are subsequences IMF1, IMF2, IMF3, …, IMF k The third value of the residual function with the smallest value; x 9N1 -t 9N1 、x 9N2 -t 9N2 、x 9N3 -t 9N3 , …, x 9Nk -t 9Nk They are subsequences IMF1, IMF2, IMF3, …, IMF k The corresponding value is the 9Nth value of the residual function with the smallest value.

8. A generator fault diagnosis method based on vibration trend prediction according to claim 7, characterized in that: The specific correction formula is: Where, f(t)' ij is the modified random function; sin() is the sine function; ln is the natural logarithm.

9. A computer device, characterized in that: It comprises a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement the steps of the generator fault diagnosis method based on vibration trend prediction as claimed in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one instruction, and when the at least one instruction is executed by the processor, the steps of the generator fault diagnosis method based on vibration trend prediction as described in any one of claims 1 to 8 are implemented.

Images