Hydro-generator fault diagnosis method and system based on artificial intelligence

Abstract

The invention discloses a hydro-generator fault diagnosis method and system based on artificial intelligence, and belongs to the technical field of hydro-generator fault diagnosis. An artificial intelligence model is adopted to construct a generator fault diagnosis model, and a training data set is adopted to train parameters of the generator fault diagnosis model; then, the real-time operation characteristic parameters corresponding to the hydro-generator are collected, and the trained generator fault diagnosis model is adopted to identify the real-time operation characteristic parameters to obtain a hydro-generator fault diagnosis result, so that the technical problems of low identification speed and low identification accuracy in the prior art are solved; and real-time online diagnosis of faults of the water turbine can be realized.

Description

Technical Field

[0001] The invention belongs to the technical field of hydraulic turbine fault diagnosis, and in particular relates to a hydraulic turbine generator fault diagnosis method and system based on artificial intelligence. Background Art

[0002] The full name of a water turbine is a hydro-generator, which is a rotating power machine that converts water energy into electrical energy. It is the core equipment of a hydroelectric power station. It drives the water turbine to rotate through water flow, which in turn drives the generator to generate electricity. During the operation of a water turbine, equipment failure may occur due to various reasons, affecting power generation efficiency and safety. Traditional fault diagnosis methods mainly rely on manual experience, and there are problems such as slow diagnosis speed and low accuracy. With the development of artificial intelligence technology, it is possible to apply it to turbine fault diagnosis. Summary of the invention

[0003] The present invention provides a hydraulic generator fault diagnosis method and system based on artificial intelligence, which are used to solve the technical problems of slow fault diagnosis speed and low accuracy caused by manual work in the prior art.

[0004] In one aspect, the present invention provides a hydraulic generator fault diagnosis method based on artificial intelligence, comprising:

[0005] Obtaining the operating characteristic parameters of the hydro-generator under various faults and the operating characteristic parameters under normal conditions, and constructing a training data set according to the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions;

[0006] An artificial intelligence model is used to construct a generator fault diagnosis model, and a training data set is used to train parameters of the generator fault diagnosis model to obtain a trained generator fault diagnosis model;

[0007] The real-time operating characteristic parameters corresponding to the hydro-generator are collected, and the trained generator fault diagnosis model is used to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result.

[0008] Furthermore, the method further includes: before using the operating characteristic parameters and the real-time operating characteristic parameters, normalizing the operating characteristic parameters and the real-time operating characteristic parameters.

[0009] Furthermore, the operating characteristic parameters include the operating speed, operating temperature, operating current, operating voltage and operating vibration frequency of the turbine.

[0010] Furthermore, a training data set is constructed based on the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions, including:

[0011] For any fault type, the operating characteristic parameters corresponding to the fault type are used as training data, and the fault type code is used as the training label;

[0012] For the normal state, the operating characteristic parameters under the normal state are encoded as training labels with normal types;

[0013] According to the training data and training labels, a training data set is obtained.

[0014] Furthermore, an artificial intelligence model is used to construct a generator fault diagnosis model, including: using a SVM model or a CNN model to construct a generator fault diagnosis model.

[0015] Furthermore, the parameters of the generator fault diagnosis model are trained using a training data set to obtain a trained generator fault diagnosis model, including:

[0016] A random initialization method is used to initialize the hyperparameters of the generator fault diagnosis model, and the initialized parameters are formed into vectors to obtain a parameter vector, and multiple different parameter vectors are obtained;

[0017] Taking the training data in the training data set as the input of the generator fault diagnosis model, taking the training labels in the training data set as the expected output of the generator fault diagnosis model, obtaining the error function value corresponding to each parameter vector, and obtaining the optimal parameter vector according to the error function value corresponding to each parameter vector;

[0018] Obtaining the adaptive inertia weight, and performing local information fusion on the parameter vector according to the adaptive inertia weight and the optimal parameter vector, to obtain the parameter vector after local information fusion;

[0019] According to the adaptive inertia weight, a local spiral search is performed on the parameter vector after local information fusion to obtain the parameter vector after the local spiral search;

[0020] A random flight vector is selected for each parameter vector after the local spiral search, and a global random search is performed on the parameter vector after the local spiral search according to the random flight vector to obtain a parameter vector after the global random search;

[0021] Local information fusion, local spiral search and global random search are repeatedly performed until the training end condition is met. The optimal parameter vector is re-determined according to the parameter vector after the last local spiral search. The parameters contained in the re-determined optimal parameter vector are used as the final hyperparameters of the generator fault diagnosis model to obtain the generator fault diagnosis model after training.

[0022] Furthermore, the adaptive inertia weight is obtained, and the parameter vector is locally fused according to the adaptive inertia weight and the optimal parameter vector to obtain the parameter vector after local information fusion, including:

[0023] Based on the current number of training times, the adaptive inertia weight is obtained as:

[0024]

[0025] Wherein, ω represents the adaptive inertia weight, cos represents the cosine function, π represents the pi, t represents the current training times, and T represents the preset maximum training times;

[0026] According to the adaptive inertia weight and the optimal parameter vector, the parameter vector is locally fused, and the parameter vector after local information fusion is obtained as follows:

[0027]

[0028] in, represents the i-th parameter vector in the t-th training process, i=1,2,…,L, L represents the total number of parameter vectors, Represents the parameter vector after local information fusion represents the parameter increment corresponding to the i-th parameter vector in the t+1-th training process, α1 represents the first local information fusion coefficient, α2 represents the second local information fusion coefficient, and α3 represents the third local information fusion coefficient. represents the parameter increment corresponding to the i-th parameter vector during the t-th training process, represents the jth adjacent parameter vector in the neighborhood of the ith parameter vector during the tth training process, represents the optimal parameter vector during the t-th training process, Represents the worst parameter vector during the t-th training process;

[0029] The neighborhood radius of the neighborhood range is:

[0030]

[0031] Among them, r represents the neighborhood radius, X max represents the upper bound vector of the parameter, X min represents the parameter lower bound vector, S(X max -X min ) represents the Euclidean distance between the upper bound vector of the parameter and the lower bound vector of the parameter.

[0032] Further, according to the adaptive inertia weight, a local spiral search is performed on the parameter vector after local information fusion to obtain the parameter vector after the local spiral search, including:

[0033] According to the optimal parameter vector, a local spiral search is performed on the parameter vector after local information fusion, and the spiral search amount is obtained as: in, represents the optimal parameter vector, represents the parameter vector after the mth local information fusion in the tth training process, π represents pi, e represents a natural constant, b represents a constant parameter of the local spiral search, l represents a random number between (0,1), m = 1, 2, ..., L, L represents the total number of parameter vectors;

[0034] According to the adaptive inertia weight and the spiral search amount, a local spiral search is performed on the parameter vector after local information fusion, and the parameter vector after the local spiral search is obtained as follows:

[0035]

[0036] in, represents the parameter vector after the local spiral search ω represents the adaptive inertia weight.

[0037] Further, a random flight vector is selected for each parameter vector after the local spiral search, and a global random search is performed on the parameter vector after the local spiral search according to the random flight vector to obtain a parameter vector after the global random search, including:

[0038] The first random search parameter and the second random search parameter are obtained as follows:

[0039] v~N(0,σ v 2 )

[0040] u~N(0,σ u 2 )

[0041] Where v represents the first random search parameter, u represents the second random search parameter, v~N(0,σ v 2 ) indicates that the first random search parameter has a mean of 0 and a variance of σ v 2 The normal distribution of v represents the first normal distribution factor, and σ v =1;u~N(0,σ u 2 ) indicates that the second random search parameter has a mean of 0 and a variance of σ u 2 The normal distribution of u represents the second normal distribution factor;

[0042] The second normal distribution factor is:

[0043]

[0044] Among them, Γ represents the gamma function, λ represents the parameter factor, π represents pi, and sin represents the sine function;

[0045] For the parameter vector after the local spiral search, a different other parameter vector is randomly matched to obtain a random flight vector;

[0046] According to the random flight vector, the first random search parameter and the second random search parameter, a global random search is performed on the parameter vector after the local spiral search, and the parameter vector after the global random search is obtained as follows:

[0047]

[0048] in, represents a random flight vector, represents the parameter vector after the nth local spiral search during the tth training process, represents the parameter vector after global random search rand represents a random number between (0,1), sign represents the sign function, and β represents the search step control factor.

[0049] On the other hand, the present invention provides a hydraulic generator fault diagnosis system based on artificial intelligence, comprising: a parameter acquisition module, a model training module and a fault diagnosis module;

[0050] The parameter acquisition module is used to acquire the operating characteristic parameters of the hydro-generator under various faults and the operating characteristic parameters under normal conditions, and to construct a training data set according to the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions;

[0051] The model training module is used to construct a generator fault diagnosis model using an artificial intelligence model, and to train parameters of the generator fault diagnosis model using a training data set to obtain a trained generator fault diagnosis model;

[0052] The fault diagnosis module is used to collect the real-time operating characteristic parameters corresponding to the hydro-generator, and use the trained generator fault diagnosis model to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result.

[0053] The present invention provides an artificial intelligence-based hydro-generator fault diagnosis method and system, which constructs a generator fault diagnosis model by adopting an artificial intelligence model, and trains the parameters of the generator fault diagnosis model by using a training data set, then collects the real-time operating characteristic parameters corresponding to the hydro-generator, and uses the trained generator fault diagnosis model to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result. This not only solves the technical problems of slow recognition speed and low recognition accuracy in the prior art, but also can realize real-time online diagnosis of turbine faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0055] Figure 1 A flow chart of a hydraulic generator fault diagnosis method based on artificial intelligence provided by an embodiment of the present invention.

[0056] Figure 2 A schematic structural diagram of a hydraulic generator fault diagnosis system based on artificial intelligence provided by an embodiment of the present invention.

[0057] The above drawings have shown clear embodiments of the present invention, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present invention in any way, but to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0058] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0059] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0060] like Figure 1 As shown, an embodiment of the present invention provides a hydraulic generator fault diagnosis method based on artificial intelligence, comprising:

[0061] S11, obtaining the operating characteristic parameters of the hydro-generator under various faults and the operating characteristic parameters under normal conditions, and constructing a training data set according to the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions;

[0062] Optionally, the operating characteristic parameters may include the operating speed, operating temperature, operating current, operating voltage and operating vibration frequency of the turbine. However, it is worth noting that the above parameters are only preferred implementations of the embodiments of the present invention, and other parameters may also be used as operating characteristic parameters to make the fault diagnosis result more accurate.

[0063] The operating characteristic parameters of a turbine in various fault states and normal states are different. These sample data can be collected in advance, and then the sample data can be learned through artificial intelligence to ultimately achieve fault diagnosis.

[0064] S12, using an artificial intelligence model to build a generator fault diagnosis model, and using a training data set to train parameters of the generator fault diagnosis model to obtain a trained generator fault diagnosis model;

[0065] Optionally, an artificial intelligence model is used to construct a generator fault diagnosis model, including: using an SVM (Support Vector Machine) model or a CNN (Convolutional Neural Network) model to construct a generator fault diagnosis model. However, it is worth noting that the above two artificial intelligence models are preferred selection methods for the embodiments of the present invention, and other artificial intelligence models can also be used to construct a generator fault diagnosis model.

[0066] After the generator fault diagnosis model is constructed, it is also necessary to train the generator fault diagnosis model to learn the correlation between the operating characteristic parameters and the fault, so as to realize the online diagnosis of the turbine fault. For example, the particle swarm algorithm or the gradient descent method can be used to train the generator fault diagnosis model.

[0067] S13, collecting the real-time operating characteristic parameters corresponding to the hydro-generator, and using the trained generator fault diagnosis model to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result.

[0068] The data type and data structure of the real-time operation characteristic parameters should be the same as the operation characteristic parameters used during training, so as to ensure that the turbine fault can be identified normally.

[0069] The fault diagnosis result of the hydro-turbine generator can be various fault types or normal state types. When the fault diagnosis result of the hydro-turbine generator can be a certain fault type, an early warning message can be generated to enable the staff to deal with the abnormal situation in time.

[0070] The present invention provides an artificial intelligence-based hydro-generator fault diagnosis method and system, which constructs a generator fault diagnosis model by adopting an artificial intelligence model, and trains the parameters of the generator fault diagnosis model by using a training data set, then collects the real-time operating characteristic parameters corresponding to the hydro-generator, and uses the trained generator fault diagnosis model to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result. This not only solves the technical problems of slow recognition speed and low recognition accuracy in the prior art, but also can realize real-time online diagnosis of turbine faults.

[0071] In the embodiment of the present invention, the method further includes: before using the operation characteristic parameters and the real-time operation characteristic parameters, normalizing the operation characteristic parameters and the real-time operation characteristic parameters.

[0072] By normalizing the operating characteristic parameters and real-time operating characteristic parameters, the complexity of the data can be reduced, and the artificial intelligence model can recognize the data more accurately.

[0073] In an embodiment of the present invention, a training data set is constructed according to the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions, including:

[0074] For any fault type, the operating characteristic parameters corresponding to the fault type are used as training data, and the fault type code is used as the training label;

[0075] For the normal state, the operating characteristic parameters under the normal state are encoded as training labels with normal types;

[0076] According to the training data and training labels, a training data set is obtained.

[0077] The output of the generator fault diagnosis model should be the number of fault types plus one. Each type of fault and normal state corresponds to an output. The training label represents the fault type or normal state type. The training of the generator fault diagnosis model can be completed by learning the data relationship between the training data and the training label.

[0078] In an embodiment of the present invention, a training data set is used to train the parameters of the generator fault diagnosis model to obtain a trained generator fault diagnosis model, including:

[0079] A random initialization method is used to initialize the hyperparameters of the generator fault diagnosis model, and the initialized parameters are formed into vectors to obtain a parameter vector, and multiple different parameter vectors are obtained;

[0080] For example, hyperparameters have upper and lower limits, and they can be randomly initialized between the upper and lower limits to achieve hyperparameter initialization.

[0081] Taking the training data in the training data set as the input of the generator fault diagnosis model, taking the training labels in the training data set as the expected output of the generator fault diagnosis model, obtaining the error function value corresponding to each parameter vector, and obtaining the optimal parameter vector according to the error function value corresponding to each parameter vector;

[0082] For example, the error function value corresponding to each parameter vector can be obtained through the cross entropy error function, and the parameter vector with the smallest error function value is determined to be the optimal parameter vector.

[0083] Obtaining the adaptive inertia weight, and performing local information fusion on the parameter vector according to the adaptive inertia weight and the optimal parameter vector, to obtain the parameter vector after local information fusion;

[0084] According to the adaptive inertia weight, a local spiral search is performed on the parameter vector after local information fusion to obtain the parameter vector after the local spiral search;

[0085] A random flight vector is selected for each parameter vector after the local spiral search, and a global random search is performed on the parameter vector after the local spiral search according to the random flight vector to obtain a parameter vector after the global random search;

[0086] Local information fusion, local spiral search and global random search are repeatedly performed until the training end condition is met. The optimal parameter vector is re-determined according to the parameter vector after the last local spiral search. The parameters contained in the re-determined optimal parameter vector are used as the final hyperparameters of the generator fault diagnosis model to obtain the generator fault diagnosis model after training.

[0087] The training end condition may be: when the current training times are greater than or equal to a preset maximum training times, the training end condition is met.

[0088] Although existing particle swarm optimization algorithms or gradient descent algorithms can realize hyperparameter training, they are very easy to fall into local optimum, resulting in a low final fault recognition rate. Therefore, an embodiment of the present invention proposes an improved training method to improve the fault recognition accuracy of a turbine.

[0089] In the embodiment of the present invention, the adaptive inertia weight is obtained, and the parameter vector is locally fused according to the adaptive inertia weight and the optimal parameter vector to obtain the parameter vector after local information fusion, including:

[0090] Based on the current number of training times, the adaptive inertia weight is obtained as:

[0091]

[0092] Wherein, ω represents the adaptive inertia weight, cos represents the cosine function, π represents the pi, t represents the current training times, and T represents the preset maximum training times;

[0093] In the early stages of algorithm iteration, the weight value is large, the algorithm can search a larger space, and the algorithm focuses more on global search; in the later stages of algorithm iteration, the weight value is small, and the algorithm performs high-precision local search in a smaller range.

[0094] According to the adaptive inertia weight and the optimal parameter vector, the parameter vector is locally fused, and the parameter vector after local information fusion is obtained as follows:

[0095]

[0096] in, represents the i-th parameter vector in the t-th training process, i=1,2,…,L, L represents the total number of parameter vectors, Represents the parameter vector after local information fusion represents the parameter increment corresponding to the i-th parameter vector in the t+1-th training process, α1 represents the first local information fusion coefficient, α2 represents the second local information fusion coefficient, and α3 represents the third local information fusion coefficient. represents the parameter increment corresponding to the i-th parameter vector during the t-th training process, represents the jth adjacent parameter vector in the neighborhood of the ith parameter vector during the tth training process, represents the optimal parameter vector during the t-th training process, Represents the worst parameter vector during the t-th training process;

[0097] The neighborhood radius of the neighborhood range is:

[0098]

[0099] Among them, r represents the neighborhood radius, X max represents the upper bound vector of the parameter, X min represents the parameter lower bound vector, S(X max -X min ) represents the Euclidean distance between the upper bound vector of the parameter and the lower bound vector of the parameter.

[0100] The local information fusion of parameter vectors provided in the embodiment of the present invention can fuse the information in the local area, making it less likely to have repeated searches. At the same time, it refers to the optimal position and the worst position, so that more unfamiliar area searches can be achieved, the search efficiency of the algorithm can be improved, and the possibility of global search can be improved.

[0101] In the embodiment of the present invention, according to the adaptive inertia weight, a local spiral search is performed on the parameter vector after local information fusion to obtain the parameter vector after the local spiral search, including:

[0102] According to the optimal parameter vector, a local spiral search is performed on the parameter vector after local information fusion, and the spiral search amount is obtained as: in, represents the optimal parameter vector, represents the parameter vector after the mth local information fusion in the tth training process, π represents pi, e represents a natural constant, b represents a constant parameter of the local spiral search, l represents a random number between (0,1), m = 1, 2, ..., L, L represents the total number of parameter vectors;

[0103] According to the adaptive inertia weight and the spiral search amount, a local spiral search is performed on the parameter vector after local information fusion, and the parameter vector after the local spiral search is obtained as follows:

[0104]

[0105] in, represents the parameter vector after the local spiral search ω represents the adaptive inertia weight.

[0106] The local spiral search provided by the embodiment of the present invention can enable all parameter vectors to perform a spiral search around the optimal position, thereby improving the solution search efficiency in the early stage of the algorithm and improving the search accuracy of the algorithm in the later stage of the algorithm.

[0107] In an embodiment of the present invention, a random flight vector is selected for each parameter vector after the local spiral search, and a global random search is performed on the parameter vector after the local spiral search according to the random flight vector to obtain the parameter vector after the global random search, including:

[0108] The first random search parameter and the second random search parameter are obtained as follows:

[0109] v~N(0,σ v 2 )

[0110] u~N(0,σ u 2 )

[0111] Where v represents the first random search parameter, u represents the second random search parameter, v~N(0,σ v 2 ) indicates that the first random search parameter has a mean of 0 and a variance of σ v 2 The normal distribution of v represents the first normal distribution factor, and σ v =1;u~N(0,σ u 2 ) indicates that the second random search parameter has a mean of 0 and a variance of σ u 2 The normal distribution of u represents the second normal distribution factor;

[0112] The second normal distribution factor is:

[0113]

[0114] Among them, Γ represents the gamma function, λ represents the parameter factor, π represents pi, and sin represents the sine function;

[0115] For the parameter vector after the local spiral search, a different other parameter vector is randomly matched to obtain a random flight vector;

[0116] According to the random flight vector, the first random search parameter and the second random search parameter, a global random search is performed on the parameter vector after the local spiral search, and the parameter vector after the global random search is obtained as follows:

[0117]

[0118] in, represents a random flight vector, represents the parameter vector after the nth local spiral search during the tth training process, represents the parameter vector after global random search rand represents a random number between (0,1), sign represents the sign function, and β represents the search step control factor.

[0119] The global random search provided by the embodiment of the present invention can enable the algorithm to have sufficient ability to jump out of the local optimal solution, improve the global search ability of the algorithm, and assist the parameter vector in performing more unknown searches.

[0120] like Figure 2 As shown, the embodiment of the present invention provides a hydraulic generator fault diagnosis system based on artificial intelligence, including: a parameter acquisition module 21, a model training module 22 and a fault diagnosis module 23;

[0121] The parameter acquisition module 21 is used to acquire the operating characteristic parameters of the hydro-generator under various faults and the operating characteristic parameters under normal conditions, and to construct a training data set according to the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions;

[0122] The model training module 22 is used to construct a generator fault diagnosis model using an artificial intelligence model, and to train the parameters of the generator fault diagnosis model using a training data set to obtain a trained generator fault diagnosis model;

[0123] The fault diagnosis module 23 is used to collect the real-time operating characteristic parameters corresponding to the hydro-generator, and use the trained generator fault diagnosis model to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result.

[0124] An artificial intelligence-based hydro-generator fault diagnosis system provided in an embodiment of the present invention can execute the above-mentioned method and technical solution, and its principles and beneficial effects are similar, which will not be repeated here.

[0125] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0126] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0127] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0128] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0129] A person of ordinary skill in the art can understand that all or part of the steps in realizing the above-mentioned facts and methods can be completed by instructing the relevant hardware through a program, and the program involved or the program described can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are derived, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0130] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A hydraulic generator fault diagnosis method based on artificial intelligence, characterized in that: include: Obtaining the operating characteristic parameters of the hydro-generator under various faults and the operating characteristic parameters under normal conditions, and constructing a training data set according to the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions; An artificial intelligence model is used to construct a generator fault diagnosis model, and a training data set is used to train parameters of the generator fault diagnosis model to obtain a trained generator fault diagnosis model; The real-time operating characteristic parameters corresponding to the hydro-generator are collected, and the trained generator fault diagnosis model is used to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result.

2. The artificial intelligence-based hydro-generator fault diagnosis method according to claim 1 is characterized in that: Also includes: Before using the operating characteristic parameters and the real-time operating characteristic parameters, the operating characteristic parameters and the real-time operating characteristic parameters are normalized.

3. The artificial intelligence-based hydraulic generator fault diagnosis method according to claim 1 is characterized in that: The operating characteristic parameters include the operating speed, operating temperature, operating current, operating voltage and operating vibration frequency of the turbine.

4. The artificial intelligence-based hydraulic generator fault diagnosis method according to claim 1 is characterized in that: The training data set is constructed based on the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions, including: For any fault type, the operating characteristic parameters corresponding to the fault type are used as training data, and the fault type code is used as the training label; For the normal state, the operating characteristic parameters under the normal state are encoded as training labels with normal types; According to the training data and training labels, a training data set is obtained.

5. The artificial intelligence-based hydro-generator fault diagnosis method according to claim 1 is characterized in that: An artificial intelligence model is used to construct a generator fault diagnosis model, including: using a SVM model or a CNN model to construct a generator fault diagnosis model.

6. The artificial intelligence-based hydraulic generator fault diagnosis method according to claim 4 is characterized in that: The parameters of the generator fault diagnosis model are trained using a training data set to obtain a trained generator fault diagnosis model, including: A random initialization method is used to initialize the hyperparameters of the generator fault diagnosis model, and the initialized parameters are formed into vectors to obtain a parameter vector, and multiple different parameter vectors are obtained; The training data in the training data set is used as the input of the generator fault diagnosis model, and the training labels in the training data set are used as the expected output of the generator fault diagnosis model, and the error function value corresponding to each parameter vector is obtained, and the optimal parameter vector is obtained according to the error function value corresponding to each parameter vector; Obtaining the adaptive inertia weight, and performing local information fusion on the parameter vector according to the adaptive inertia weight and the optimal parameter vector, to obtain the parameter vector after local information fusion; According to the adaptive inertia weight, a local spiral search is performed on the parameter vector after local information fusion to obtain the parameter vector after the local spiral search; A random flight vector is selected for each parameter vector after the local spiral search, and a global random search is performed on the parameter vector after the local spiral search according to the random flight vector to obtain a parameter vector after the global random search; Local information fusion, local spiral search and global random search are repeatedly performed until the training end condition is met. The optimal parameter vector is re-determined according to the parameter vector after the last local spiral search. The parameters contained in the re-determined optimal parameter vector are used as the final hyperparameters of the generator fault diagnosis model to obtain the generator fault diagnosis model after training.

7. The artificial intelligence-based hydro-generator fault diagnosis method according to claim 6 is characterized in that: Obtain the adaptive inertia weight, and perform local information fusion on the parameter vector according to the adaptive inertia weight and the optimal parameter vector to obtain the parameter vector after local information fusion, including: Based on the current number of training times, the adaptive inertia weight is obtained as: Wherein, ω represents the adaptive inertia weight, cos represents the cosine function, π represents the pi, t represents the current training times, and T represents the preset maximum training times; According to the adaptive inertia weight and the optimal parameter vector, the parameter vector is locally fused, and the parameter vector after local information fusion is obtained as follows: in, represents the i-th parameter vector in the t-th training process, i=1,2,…,L, L represents the total number of parameter vectors, Represents the parameter vector after local information fusion represents the parameter increment corresponding to the i-th parameter vector in the t+1-th training process, α1 represents the first local information fusion coefficient, α2 represents the second local information fusion coefficient, and α3 represents the third local information fusion coefficient. represents the parameter increment corresponding to the i-th parameter vector during the t-th training process, represents the jth adjacent parameter vector in the neighborhood of the ith parameter vector during the tth training process, represents the optimal parameter vector during the t-th training process, Represents the worst parameter vector during the t-th training process; The neighborhood radius of the neighborhood range is: Among them, r represents the neighborhood radius, X max represents the upper bound vector of the parameter, X min represents the parameter lower bound vector, S(X max -X min ) represents the Euclidean distance between the upper bound vector of the parameter and the lower bound vector of the parameter.

8. The artificial intelligence-based hydro-generator fault diagnosis method according to claim 7 is characterized in that: According to the adaptive inertia weight, a local spiral search is performed on the parameter vector after local information fusion to obtain the parameter vector after the local spiral search, including: According to the optimal parameter vector, a local spiral search is performed on the parameter vector after local information fusion, and the spiral search amount is obtained as: in, represents the optimal parameter vector, represents the parameter vector after the mth local information fusion in the tth training process, π represents pi, e represents a natural constant, b represents a constant parameter of the local spiral search, l represents a random number between (0,1), m = 1, 2, ..., L, L represents the total number of parameter vectors; According to the adaptive inertia weight and the spiral search amount, a local spiral search is performed on the parameter vector after local information fusion, and the parameter vector after the local spiral search is obtained as follows: in, represents the parameter vector after the local spiral search ω represents the adaptive inertia weight.

9. The artificial intelligence-based fault diagnosis method for a hydro-generator according to claim 8 is characterized in that: A random flight vector is selected for each parameter vector after the local spiral search, and a global random search is performed on the parameter vector after the local spiral search according to the random flight vector to obtain the parameter vector after the global random search, including: The first random search parameter and the second random search parameter are obtained as follows: v~N(0,σ v 2 ) u~N(0,σ u 2 ) Where v represents the first random search parameter, u represents the second random search parameter, v~N(0,σ v 2 ) indicates that the first random search parameter has a mean of 0 and a variance of σ v 2 The normal distribution of v represents the first normal distribution factor, and σ v =1;u~N(0,σ u 2 ) indicates that the second random search parameter has a mean of 0 and a variance of σ u 2 The normal distribution of u represents the second normal distribution factor; The second normal distribution factor is: Among them, Γ represents the gamma function, λ represents the parameter factor, π represents pi, and sin represents the sine function; For the parameter vector after the local spiral search, a different other parameter vector is randomly matched to obtain a random flight vector; According to the random flight vector, the first random search parameter and the second random search parameter, a global random search is performed on the parameter vector after the local spiral search, and the parameter vector after the global random search is obtained as follows: in, represents a random flight vector, represents the parameter vector after the nth local spiral search during the tth training process, represents the parameter vector after global random search rand represents a random number between (0,1), sign represents the sign function, and β represents the search step control factor.

10. A hydraulic generator fault diagnosis system based on artificial intelligence, characterized in that: include: Parameter acquisition module, model training module and fault diagnosis module; The parameter acquisition module is used to acquire the operating characteristic parameters of the hydro-generator under various faults and the operating characteristic parameters under normal conditions, and to construct a training data set according to the operating characteristic parameters under various faults and the operating characteristic parameters under normal conditions; The model training module is used to construct a generator fault diagnosis model using an artificial intelligence model, and to train parameters of the generator fault diagnosis model using a training data set to obtain a trained generator fault diagnosis model; The fault diagnosis module is used to collect the real-time operating characteristic parameters corresponding to the hydro-generator, and use the trained generator fault diagnosis model to identify the real-time operating characteristic parameters to obtain the hydro-generator fault diagnosis result.

Images