Hydro-generator ultrahigh frequency partial discharge fault diagnosis method and device

Abstract

The invention provides a hydro-generator ultrahigh frequency partial discharge fault diagnosis method and device, and relates to the technical field of partial discharge monitoring. The method comprises the following steps: discretizing an obtained ultrahigh frequency partial discharge signal to obtain a partial discharge pulse sequence; performing clustering analysis on an initial long sequence formed by each partial discharge pulse sequence, and identifying each cluster center; comparing the partial discharge pulse sequence corresponding to each cluster center with each typical partial discharge event to identify the cluster where the noise signal is located, and removing the partial discharge pulse sequence corresponding to the cluster where the noise signal is located from the initial long sequence to obtain a target long sequence; and taking the extracted characteristic parameters of each partial discharge pulse sequence in the target long sequence as model input, and predicting a typical partial discharge event type by using a partial discharge classification model. According to the method, the efficiency and precision of partial discharge fault diagnosis are improved through combination of clustering analysis taking the partial discharge pulse sequence as the sample point and typical partial discharge event identification based on machine learning.

Description

Technical Field

[0001] The present application belongs to the technical field of partial discharge monitoring, and specifically relates to a method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator, a device for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator, a computer device, and a machine-readable storage medium. Background Art

[0002] At present, the mainstream partial discharge (PD) test equipment in the market mainly adopts low-frequency PD test technology, with a frequency range of 10kHz~500kHz. When applied to hydro-turbine generators, the monitored test data fluctuates greatly due to the interference signal during the operation phase of the generator, resulting in problems such as difficulty in identifying PD patterns. In order to overcome the above-mentioned defects of low-frequency PD test technology, many scholars have conducted research on ultra-high frequency PD test technology for hydro-turbine generators. The frequency range of ultra-high frequency PD is 300MHz~3GHz, and existing research has shown that compared with low-frequency PD test technology, ultra-high frequency PD test technology can avoid some external interference, and the stability of test results is better than that of low-frequency PD test technology.

[0003] In the actual production process of hydro-turbine generators, the following three factors must be considered in the handling of partial discharge events: the first aspect is the determination of the partial discharge level and severity, which is related to whether the power plant needs to shut down to avoid serious damage to the equipment; the second aspect is the location of the partial discharge source. Accurately determining the location of the partial discharge is conducive to the formulation of subsequent repair strategies; the third aspect is the processing time of the partial discharge event. The processing time of the partial discharge event is related to the power plant's operating plan and resource income, and is a very important economic consideration.

[0004] Based on the above considerations, the pre-construction of a partial discharge case library can play an important role. The defects of typical partial discharge cases in the current partial discharge case library are mainly reflected in the representativeness, update speed and technical coverage. First, the existing cases are often constructed based on specific conditions or failure modes, which may not accurately reflect the partial discharge characteristics of different devices or environments in actual applications. In addition, how to use the constructed partial discharge case library to accurately identify partial discharge faults is also a technical difficulty that needs to be overcome urgently. Summary of the invention

[0005] The purpose of the embodiments of the present application is to provide a method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator, a device for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator, a computer device and a machine-readable storage medium, so as to realize rapid and accurate fault diagnosis based on the identification of typical partial discharge events.

[0006] To achieve the above-mentioned purpose, the first aspect of the present application provides a method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator, comprising: Acquire multiple ultra-high frequency partial discharge signals collected from the hydro-generator during the current monitoring period; Discretizing the ultra-high frequency partial discharge signal to obtain a partial discharge pulse sequence; Taking the partial discharge pulse sequence as a sample point, cluster analysis is performed on an initial long sequence composed of various partial discharge pulse sequences to identify each cluster center and its corresponding area; The partial discharge pulse sequences corresponding to the centers of each cluster are compared one by one with each typical partial discharge event in a pre-built partial discharge case library to identify the cluster where the noise signal is located, and the partial discharge pulse sequences corresponding to the cluster where the noise signal is located are removed from the initial long sequence to obtain a target long sequence; Extracting time domain and / or frequency domain characteristic parameters of each partial discharge pulse sequence in the target long sequence, using the characteristic parameters of each partial discharge pulse sequence as input of a pre-constructed partial discharge classification model, and using the partial discharge classification model to predict typical partial discharge event types generated by the hydro-generator in the current monitoring period; Locating the partial discharge source according to the typical partial discharge event type; Wherein, the partial discharge classification model is obtained by training an initial machine learning classifier.

[0007] In a specific embodiment of the present application, the characteristic parameter includes at least one of signal amplitude, pulse duration, pulse rise time, maximum frequency and minimum frequency.

[0008] In a specific embodiment of the present application, the partial discharge case library includes at least one of a first-level partial discharge case and a second-level partial discharge case, the first-level partial discharge case includes at least one of a first typical partial discharge event, a second typical partial discharge event, a third typical partial discharge event, a fourth typical partial discharge event, a fifth typical partial discharge event, a sixth typical partial discharge event, a seventh typical partial discharge event, and an eighth typical partial discharge event, the second-level partial discharge case includes at least one of a ninth typical partial discharge event, a tenth typical partial discharge event, and an eleventh typical partial discharge event, and the fault severity of the first-level partial discharge case is higher than the fault severity of the second-level partial discharge case; Among them, the first typical partial discharge event is a partial discharge event generated when the humidity of the environment in which the hydro-turbine generator is located is greater than the first threshold; the second typical partial discharge event is a partial discharge event generated when the temperature of the environment in which the hydro-turbine generator is located is greater than the second threshold; the third typical partial discharge event is a partial discharge event generated when the electromagnetic interference intensity of the environment in which the hydro-turbine generator is located is greater than the third threshold; the fourth typical partial discharge event is a partial discharge event generated after the aging of the insulating material in the hydro-turbine generator; the fifth typical partial discharge event is a partial discharge event generated by the hydro-turbine generator under mechanical stress; the sixth typical partial discharge event is a partial discharge event generated after the hydro-turbine generator has an insulation defect; the seventh typical partial discharge event is a partial discharge event generated when the hydro-turbine generator is extremely controlled; the eighth typical partial discharge event is a partial discharge event generated when the hydro-turbine generator is under a composite fault condition, and the composite fault condition includes more than one of the following: the humidity of the environment in which the hydro-turbine generator is located is greater than the first threshold, the temperature of the environment in which the hydro-turbine generator is located is greater than the second threshold, and the hydro-turbine generator is subjected to mechanical stress; the ninth typical partial discharge event is a partial discharge event corresponding to different partial discharge fault mechanisms; the tenth typical partial discharge event is a partial discharge event corresponding to different partial discharge generation locations; and the eleventh typical partial discharge event is a partial discharge event corresponding to different environmental conditions.

[0009] In a specific embodiment of the present application, the partial discharge classification model is obtained after training the initial machine learning classifier, including: The characteristic parameters extracted from the partial discharge signals when each typical partial discharge event occurs in the partial discharge case library are used as input parameters and the typical partial discharge event type is used as a label value to construct a training sample; The training samples are used to train an initial machine learning classifier to obtain a partial discharge classification model.

[0010] In a specific embodiment of the present application, locating the partial discharge source according to the typical partial discharge event type includes: If the predicted partial discharge event type is in the first level of partial discharge cases, a first control signal is issued to shut down the hydro-generator, and then a partial discharge source is determined according to the partial discharge event type; If the predicted partial discharge event type is in the second level partial discharge case, the partial discharge source is determined according to the partial discharge event type.

[0011] In a specific embodiment of the present application, the partial discharge pulse sequence is used as a sample point, and a cluster analysis is performed on the initial long sequence composed of various partial discharge pulse sequences to identify each cluster center and its corresponding area, including: The initial long sequence is composed of various partial discharge pulse sequences; Converting each partial discharge pulse sequence in the initial long sequence into the frequency domain to obtain each partial discharge spectrum; Extracting a low power ratio and a high power ratio of the partial discharge spectrum, and forming sample points by the low power ratio and the high power ratio; Perform spectral clustering analysis on each sample point to identify each cluster center and its corresponding area; Among them, the low power ratio is the ratio of the sum of the square of the modulus values ​​of each frequency point in the partial discharge spectrum within a preset low frequency interval to the sum of the square of the modulus values ​​of all frequency points in the preset frequency range in the partial discharge spectrum, and the high power ratio is the ratio of the sum of the square of the modulus values ​​of each frequency point in the preset high frequency interval of the partial discharge spectrum to the sum of the square of the modulus values ​​of all frequency points in the preset frequency range in the partial discharge spectrum. The preset low frequency interval and the preset high frequency interval are both within the preset frequency range, the lower boundary of the preset frequency range is not lower than the minimum frequency in the partial discharge spectrum, and the upper boundary of the preset frequency range is not higher than the maximum frequency in the partial discharge spectrum.

[0012] In a specific embodiment of the present application, the partial discharge pulse sequence is used as a sample point, and a cluster analysis is performed on an initial long sequence composed of various partial discharge pulse sequences to identify each cluster center and its corresponding area, and further includes: A numerical optimization algorithm is used to find the optimal preset low-frequency interval, preset high-frequency interval and preset frequency range to trigger the extraction of the low power ratio and high power ratio of the partial discharge spectrum, and the low power ratio and the high power ratio form sample points.

[0013] In a specific embodiment of the present application, a numerical optimization algorithm is used to find the optimal preset low frequency interval, preset high frequency interval and preset frequency range, including: The optimization objectives of the optimization problem are to maximize the minimum value of the Euclidean distance between cluster centers in spectral clustering, to maximize the minimum value of the Mahalanobis distance between cluster centers in spectral clustering, and to maximize the minimum value of the product of the Euclidean distance and the Mahalanobis distance between cluster centers in spectral clustering. Setting decision variables for the optimization problem, wherein the decision variables include a first difference, a second difference, a third difference, a fourth difference, and a fifth difference; Setting constraints of the optimization problem, wherein the constraints are set according to the value ranges of the upper and lower boundaries of the preset low-frequency interval and the upper and lower boundaries of the preset high-frequency interval; Find the optimal solution to the optimization problem and obtain the optimal decision variables; Calculating the optimal preset low-frequency interval, preset high-frequency interval and preset frequency range according to the optimal decision variables; Among them, the first difference is the difference between the lower boundary of the preset low-frequency interval and the lower boundary of the preset variation range, the second difference is the difference between the upper boundary of the preset low-frequency interval and the lower boundary of the preset low-frequency interval, the third difference is the difference between the lower boundary of the preset high-frequency interval and the lower boundary of the preset low-frequency interval, the upper boundary of the preset high-frequency interval is the maximum value of the sum of the upper boundary of the preset low-frequency interval and the fourth difference, and the lower boundary of the preset high-frequency interval and the sum of the fourth difference, the fifth difference is the difference between the upper boundary of the preset frequency range and the upper boundary of the preset high-frequency interval, and the preset variation range is the variation range when each element in the decision variable is updated.

[0014] In a specific embodiment of the present application, the constraint conditions are:

[0015]

[0016]

[0017] ; in, Indicates the lower boundary of the preset low-frequency range. Indicates the upper boundary of the preset low-frequency range. Indicates the lower boundary of the preset high-frequency range, Indicates the upper boundary of the preset high-frequency range. Indicates the upper boundary of the preset frequency range.

[0018] In a specific embodiment of the present application, the numerical optimization algorithm is a differential evolution algorithm.

[0019] A second aspect of the present application provides a hydro-generator ultra-high frequency partial discharge fault diagnosis device, comprising: A first acquisition module is used to acquire multiple ultra-high frequency partial discharge signals collected from the hydro-generator in the current monitoring period; A preprocessing module, used for discretizing the ultra-high frequency partial discharge signal to obtain a partial discharge pulse sequence; A cluster analysis module, used to perform cluster analysis on an initial long sequence composed of various partial discharge pulse sequences using the partial discharge pulse sequence as a sample point, and identify each cluster center and its corresponding area; A noise removal module is used to compare the partial discharge pulse sequence corresponding to each cluster center with each typical partial discharge event in a pre-built partial discharge case library one by one to identify the cluster where the noise signal is located, and remove the partial discharge pulse sequence corresponding to the cluster where the noise signal is located from the initial long sequence to obtain a target long sequence; A partial discharge event classification module is used to extract time domain and / or frequency domain characteristic parameters of each partial discharge pulse sequence in the target long sequence, use the characteristic parameters of each partial discharge pulse sequence as input of a pre-constructed partial discharge classification model, and use the partial discharge classification model to predict the typical partial discharge event type generated by the hydro-generator in the current monitoring period; A partial discharge source locating module, used for locating the partial discharge source according to the typical partial discharge event type; Wherein, the partial discharge classification model is obtained by training an initial machine learning classifier.

[0020] A third aspect of the present application provides a computer device, comprising: a memory configured to store instructions; and The processor is configured to call the instruction from the memory and implement the ultra-high frequency partial discharge fault diagnosis method of the hydro-turbine generator according to the first aspect of the present application when executing the instruction.

[0021] The fourth aspect of the present application provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator according to the first aspect of the present application.

[0022] The above technical solution is based on the construction of a partial discharge case library, uses partial discharge pulse sequences as sample points for cluster analysis to separate signals from noise, and uses a partial discharge classification model based on a machine learning classifier to predict typical partial discharge event types, thereby achieving rapid and accurate identification of typical partial discharge events, thereby improving the efficiency and accuracy of partial discharge fault diagnosis. Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present application, but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings: Figure 1 A schematic diagram of a flow chart of a method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to an embodiment of the present application is shown; Figure 2 A block diagram schematically shows a composition of a hydro-generator ultra-high frequency partial discharge fault diagnosis device according to an embodiment of the present application; Figure 3 The structural block diagram of a computer device according to an embodiment of the present application is schematically shown. DETAILED DESCRIPTION

[0024] The specific implementation of the embodiment of the present application is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present application, and is not used to limit the embodiment of the present application.

[0025] If there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0026] The present application provides a method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator, which mainly performs partial discharge fault diagnosis in the following manners: Acquire multiple ultra-high frequency partial discharge signals collected from the hydro-generator during the current monitoring period; Discretize the acquired ultra-high frequency partial discharge signal to obtain a partial discharge pulse sequence, and form an initial long sequence from each partial discharge pulse sequence; Taking each PD pulse sequence as a sample point, cluster analysis is performed on the initial long sequence to identify each cluster center and its corresponding area. The PD pulse sequences corresponding to the centers of each cluster are compared one by one with each typical PD event in the pre-built PD case library, the cluster where the noise signal is located is identified, and the PD pulse sequences corresponding to the cluster where the noise signal is located are removed from the initial long sequence to obtain the target long sequence; Extracting time domain and / or frequency domain characteristic parameters of each partial discharge pulse sequence in the target long sequence, using the characteristic parameters of each partial discharge pulse sequence as input of a pre-built partial discharge classification model, and using the partial discharge classification model to predict the typical partial discharge event type generated by the hydro-generator in the current monitoring period; The partial discharge source is located based on the predicted typical partial discharge event types.

[0027] In the present application, the partial discharge classification model takes the characteristic parameters of the partial discharge pulse sequence as input and the typical partial discharge event type of the partial discharge pulse sequence as output. The partial discharge classification model is obtained by training the initial machine learning classifier, and the initial machine learning classifier can be a classifier based on a machine learning algorithm or a classifier based on a neural network. It is known that after identifying the partial discharge event type of the current partial discharge signal, the partial discharge source can be located. For example, when the partial discharge event type is slot discharge, the current partial discharge fault position can be located in the slot part of the stator winding where the coupling device is located. Positioning the partial discharge source based on the typical partial discharge event type is not a technical improvement of the present application, so this application does not describe this part in detail.

[0028] It can be known that the time domain and frequency domain characteristics of the noise signal are different from the time domain and frequency domain characteristics of the partial discharge signal. Therefore, in the above embodiment, the cluster to which the noise signal belongs can be identified based on these differences.

[0029] Specifically, as an example, the difference between the amplitude of the noise signal and the maximum partial discharge amplitude of each typical partial discharge event in the partial discharge case library can be used to remove the noise signal. Specifically, the amplitude of the partial discharge pulse sequence corresponding to each cluster center is compared one by one with the maximum partial discharge amplitude of each typical partial discharge event in the pre-constructed partial discharge case library. If the difference between the amplitude of the partial discharge pulse sequence corresponding to a cluster center and all the maximum partial discharge amplitudes is greater than a first preset value, it can be determined that the partial discharge pulse sequence corresponding to the cluster center is not a real partial discharge pulse sequence, and the partial discharge pulse sequence corresponding to the cluster center is removed from the initial long sequence as a noise signal.

[0030] In some comparative embodiments, a fixed partial discharge amplitude threshold is determined in advance through artificial experience, and the partial discharge amplitude threshold is used for segmentation to separate the noise signal. In the above embodiment of the present application, a clustering analysis method is used to identify the noise signal, which is essentially an adaptive noise separation based on the actual situation of the current initial long sequence. Compared with the comparative embodiment, the recognition accuracy of the partial discharge signal is improved. In other comparative embodiments, a clustering method is used to classify partial discharge events. The clustering method is an unsupervised method, which shows its weakness in the recognition of a large number of partial discharge event types of hydro-turbine generators, and the recognition accuracy needs to be improved. However, in the above embodiment of the present application, partial discharge events are classified by a pre-constructed partial discharge classification model. The partial discharge classification model is constructed based on a machine learning algorithm and trained based on a training data set with real label values. The recognition accuracy and efficiency of a large number of classification types are significantly higher than the clustering method.

[0031] In summary, the above embodiments of the present application, based on the construction of a partial discharge case library, combine cluster analysis using partial discharge pulse sequences as sample points and a partial discharge classification model based on a machine learning classifier, thereby achieving rapid and accurate identification of partial discharge events, thereby improving the accuracy and efficiency of partial discharge fault diagnosis.

[0032] As an optional embodiment of the present application, the partial discharge case library includes at least one of the partial discharge cases of the first level and the partial discharge cases of the second level, the partial discharge cases of the first level include at least one of the first typical partial discharge event, the second typical partial discharge event, the third typical partial discharge event, the fourth typical partial discharge event, the fifth typical partial discharge event, the sixth typical partial discharge event, the seventh typical partial discharge event and the eighth typical partial discharge event, the partial discharge cases of the second level include at least one of the ninth typical partial discharge event, the tenth typical partial discharge event and the eleventh typical partial discharge event, and the fault severity of the partial discharge cases of the first level is higher than the fault severity of the partial discharge cases of the second level. Preferably, in a specific embodiment, the partial discharge case library includes the partial discharge cases of the first level and the partial discharge cases of the second level, the partial discharge cases of the first level include the first typical partial discharge event to the eighth typical partial discharge event, and the partial discharge cases of the second level include the ninth typical partial discharge event to the eleventh typical partial discharge event.

[0033] In the present application, the first typical partial discharge event is a partial discharge event generated when the humidity of the environment in which the hydro-turbine generator is located is greater than the first threshold value. The second typical partial discharge event is a partial discharge event generated when the temperature of the environment in which the hydro-turbine generator is located is greater than the second threshold value. The third typical partial discharge event is a partial discharge event generated when the electromagnetic interference intensity of the environment in which the hydro-turbine generator is located is greater than the third threshold value. The fourth typical partial discharge event is a partial discharge event generated after the aging of the insulating material in the hydro-turbine generator. The fifth typical partial discharge event is a partial discharge event generated by the hydro-turbine generator under mechanical stress. The sixth typical partial discharge event is a partial discharge event generated after the hydro-turbine generator has an insulation defect. The seventh typical partial discharge event is a partial discharge event generated when the hydro-turbine generator is extremely controlled. The eighth typical partial discharge event is a partial discharge event generated when the hydro-turbine generator is under a composite fault condition, and the composite fault condition includes more than one of the humidity of the environment in which the hydro-turbine generator is located is greater than the first threshold value, the temperature of the environment in which the hydro-turbine generator is located is greater than the second threshold value, and the hydro-turbine generator is subjected to mechanical stress. The ninth typical partial discharge event is a partial discharge event corresponding to different partial discharge fault mechanisms. The tenth typical partial discharge event is a partial discharge event corresponding to different partial discharge generation locations. The eleventh typical partial discharge event is the partial discharge event corresponding to different environmental conditions.

[0034] As an example, the first typical partial discharge event is a partial discharge case under extreme humidity conditions, specifically referring to: simulating partial discharge under high humidity environment, because moisture can significantly change the performance of insulating materials and increase the probability of partial discharge. Through the partial discharge simulation under such conditions, the maximum partial discharge amplitude or the maximum partial discharge amount under such conditions is obtained. The second typical partial discharge event is a partial discharge case under high temperature environment, specifically referring to: partial discharge simulation under conditions higher than the normal operating temperature, because high temperature may cause accelerated aging of insulating materials, thereby inducing partial discharge. Through the partial discharge simulation under such conditions, the maximum partial discharge amplitude or the maximum partial discharge amount under such conditions is obtained. The third typical partial discharge event is a partial discharge case under strong electromagnetic interference environment, specifically referring to: simulating partial discharge in an environment with strong electromagnetic interference. In specific applications, the complex electromagnetic interference environment in the actual power system is simulated to obtain the maximum partial discharge amplitude or the maximum partial discharge amount under such conditions. The fourth typical partial discharge event is a partial discharge case after the aging of the insulating material, which specifically refers to: using the aged insulating material or accelerated aging experiment to simulate the long-term operation effect of the hydro-turbine generator, examining the influence of insulation aging on partial discharge, and obtaining the maximum partial discharge amplitude or maximum partial discharge amount under this condition. The fifth typical partial discharge event is a partial discharge case caused by the hydro-turbine generator under mechanical stress, which specifically refers to: during the partial discharge test of the hydro-turbine generator, mechanical vibration or impact is applied to the hydro-turbine generator, because mechanical stress or impact may cause damage to the insulation structure, thereby generating partial discharge, and by examining the partial discharge caused by such insulation structure damage, the maximum partial discharge amplitude or maximum partial discharge amount under this condition is obtained. The sixth typical partial discharge event is a partial discharge case caused by different types of insulation defects, which specifically refers to: constructing insulation defect models with different types and sizes, such as bubbles, cracks, impurities, etc., examining the influence of these different insulation defects on partial discharge, and obtaining the maximum partial discharge amplitude or maximum partial discharge amount under this condition. The seventh typical partial discharge event is a partial discharge case under extreme operating conditions of the turbine generator, specifically referring to: simulating the extreme operating conditions of the turbine generator, such as rapid voltage increase, rapid voltage reduction and overload, etc., because these extreme operating conditions may lead to aggravation of the partial discharge phenomenon, and through the partial discharge simulation under these extreme operating conditions, the maximum partial discharge amplitude or maximum partial discharge amount under such conditions is obtained. The eighth typical partial discharge event is a partial discharge case under compound fault conditions, specifically referring to: conducting partial discharge simulation tests on the turbine generator under multiple extreme conditions at the same time, such as the simultaneous existence of high temperature, high humidity and mechanical stress, examining the comprehensive impact of compound fault conditions on partial discharge, and obtaining the maximum partial discharge amplitude or maximum partial discharge amount under such conditions.The ninth typical partial discharge event is the partial discharge case corresponding to different partial discharge fault mechanisms, specifically referring to: according to different fault mechanisms, the stator winding wire rods of the hydro-turbine generator are aged and defected, and different types of partial discharge fault cases including internal discharge, surface discharge, and corona discharge are designed. These fault cases not only include internal discharge phenomena caused by aging of insulating materials, but also involve surface discharge problems caused by surface wear or contamination of the wire rods and corona discharge phenomena caused by uneven electric field strength, and obtain the maximum partial discharge amplitude or maximum partial discharge amount corresponding to each partial discharge fault mechanism. The tenth typical partial discharge event is the partial discharge case corresponding to different partial discharge generation locations, specifically referring to: according to different partial discharge generation locations, simulate the defects of partial discharge occurring at different locations of the hydro-turbine generator, collect relevant data, such as: insulator surface, internal air gap, electrode tip, etc., to examine the influence of local discharge at different locations, and obtain the maximum partial discharge amplitude or maximum partial discharge amount corresponding to each partial discharge generation location. The eleventh typical partial discharge event is a partial discharge case under different environmental conditions, specifically referring to: simulating the defects of a hydro-turbine generator under different environmental conditions, including temperature, humidity, pressure, pollution level, background noise, interference from other equipment, etc., to obtain the corresponding maximum partial discharge amplitude or maximum partial discharge amount under each environmental condition.

[0035] As in the above embodiment, by defining various extreme conditions to examine typical partial discharge events under these extreme conditions, and considering typical partial discharge events under general conditions, partial discharge events are visualized from multiple dimensions, and a more complete partial discharge case library than the traditional partial discharge case library is constructed, thereby supporting the effective separation of noise signals and useful signals based on cluster analysis, avoiding filtering out useful signals as noise signals, and thereby improving the accuracy of partial discharge fault diagnosis.

[0036] Figure 1 The flowchart of the method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator according to an embodiment of the present application is schematically shown. Figure 1 As shown, the hydro-generator ultra-high frequency partial discharge fault diagnosis method provided in the embodiment of the present application includes steps 102 to 112. It can be understood that the above-mentioned hydro-generator ultra-high frequency partial discharge fault diagnosis method may include all steps from step 102 to step 112, or may only include some of the steps.

[0037] Step 102, obtaining a plurality of ultra-high frequency partial discharge signals collected from the hydro-generator in the current monitoring period.

[0038] Step 104, preprocessing the acquired ultra-high frequency partial discharge signal. The preprocessing includes filtering, discretization and normalization, etc. Partial discharge pulse sequences are obtained by discretization, and the initial long sequence is composed of each partial discharge pulse sequence.

[0039] Step 106 , using the partial discharge pulse sequence as a sample point, performing cluster analysis on the initial long sequence, and identifying each cluster center and its corresponding area.

[0040] Step 108, comparing the PD pulse sequences corresponding to the centers of each cluster with each typical PD event in the pre-built PD case library one by one, identifying the cluster where the noise signal is located, and removing the PD pulse sequences corresponding to the cluster where the noise signal is located from the initial long sequence to obtain the target long sequence.

[0041] Step 110, extracting time domain and / or frequency domain characteristic parameters of each partial discharge pulse sequence in the target long sequence, using the characteristic parameters of each partial discharge pulse sequence as input of a pre-constructed partial discharge classification model, and using the partial discharge classification model to predict typical partial discharge event types generated by the hydro-generator in the current monitoring period.

[0042] Specifically, the partial discharge classification model is obtained by training the initial machine learning classifier.

[0043] As an optional embodiment of the present application, the partial discharge classification model is obtained by training the initial machine learning classifier using the constructed partial discharge case library, specifically, the following steps are included: The characteristic parameters extracted from the partial discharge signals of each typical partial discharge event in the partial discharge case library are used as input parameters, and the typical partial discharge event type is used as a label value to construct a training sample; The initial machine learning classifier is trained using the training samples to obtain a partial discharge classification model.

[0044] As an example, the characteristic parameter includes at least one of amplitude, pulse duration, pulse rise time, maximum frequency, and minimum frequency.

[0045] Step 112: locating the partial discharge source according to the predicted typical partial discharge event type.

[0046] As an optional embodiment of the present application, localizing the partial discharge source according to the predicted partial discharge event type specifically includes the following steps: If the predicted partial discharge event type is in the first level of partial discharge cases, a first control signal is issued to shut down the hydro-generator, and then a partial discharge source is determined according to the partial discharge event type; If the predicted partial discharge event type is in the second level partial discharge case, the partial discharge source is determined according to the partial discharge event type.

[0047] As an optional embodiment of the present application, a spectral clustering method is used to perform clustering analysis of the initial long sequence, specifically, the following steps are included: Step 202, converting each partial discharge pulse sequence of the initial long sequence into the frequency domain to obtain each partial discharge spectrum; Step 204, extracting the low power ratio and the high power ratio of the partial discharge spectrum, and forming sample points by the low power ratio and the high power ratio; Step 206: Perform spectral clustering analysis on each sample point to identify each cluster center and its corresponding area.

[0048] Specifically, the low power ratio indicates the ratio of the sum of the squared modulus values ​​of each frequency point in the preset low frequency interval in the partial discharge spectrum to the sum of the squared modulus values ​​of all frequency points in the preset frequency range in the partial discharge spectrum. The high power ratio indicates the ratio of the sum of the squared modulus values ​​of each frequency point in the preset high frequency interval in the partial discharge spectrum to the sum of the squared modulus values ​​of all frequency points in the preset frequency range in the partial discharge spectrum. The preset low frequency interval and the preset high frequency interval are both within the above-mentioned preset frequency range, the lower boundary of the preset frequency range is not lower than the minimum frequency in the partial discharge spectrum, the upper boundary of the preset frequency range is not higher than the maximum frequency in the partial discharge spectrum, and the upper boundary frequency of the preset low frequency interval is lower than the upper boundary frequency of the preset high frequency interval.

[0049] It should be understood that a sample point is composed of a low power ratio and a high power ratio, specifically: in spectral clustering, a high power ratio and a low power ratio of a partial discharge pulse sequence are represented by a two-dimensional graph. In frequency domain conversion, Fourier transform can be used.

[0050] As an example, the low power ratio is expressed by: (Formula 1); In formula 1, PRL Indicates low power ratio, Indicates the preset low frequency range. is the lower boundary of the preset low-frequency range, is the upper boundary of the preset low-frequency range. Indicates the mid-frequency point of the partial discharge spectrum The modulus value, Indicates the upper boundary of the preset frequency range, that is, the maximum value in the preset frequency range. Indicates the lower boundary of the preset frequency range, that is, the minimum value in the preset frequency range.

[0051] As an example, the high power ratio is expressed by:

[0052] In formula 2, PRH Indicates high power ratio, Indicates the preset high frequency range, is the lower boundary of the preset high-frequency range, It is the upper boundary of the preset high frequency range.

[0053] It can be seen that, as in the above embodiment, in order to achieve accurate spectral clustering of each partial discharge pulse sequence, the preset low-frequency interval, the preset high-frequency interval and the preset frequency range need to be accurately divided according to the partial discharge spectrum, and inaccurate division may lead to overlap between the corresponding areas of each cluster center obtained by spectral clustering.

[0054] To overcome the above defects, in an improved embodiment of the present application, step 203 is further included between step 202 and step 204: Step 203: using a numerical optimization algorithm to find the optimal preset low frequency interval, preset high frequency interval and preset frequency range.

[0055] By optimizing the preset low-frequency interval, the preset high-frequency interval and the preset frequency range, the overlap between the clusters is avoided, thereby improving the accuracy of signal-noise separation.

[0056] As an example, using a numerical optimization algorithm to find the optimal preset low frequency interval, preset high frequency interval and preset frequency range specifically includes the following steps: The optimization objectives of the optimization problem are to maximize the minimum value of the Euclidean distance between cluster centers in spectral clustering, to maximize the minimum value of the Mahalanobis distance between cluster centers in spectral clustering, and to maximize the minimum value of the product of the Euclidean distance and the Mahalanobis distance between cluster centers in spectral clustering. Setting decision variables of the optimization problem, the decision variables including a first difference, a second difference, a third difference, a fourth difference and a fifth difference; Setting constraints for the optimization problem, where the constraints are set according to a value range of an upper boundary of a preset low-frequency interval, a lower boundary of a preset low-frequency interval, an upper boundary of a preset high-frequency interval, and a lower boundary of a preset high-frequency interval; Find the optimal solution to the optimization problem and obtain the optimal decision variables; The optimal preset low-frequency interval, the optimal preset high-frequency interval and the optimal preset frequency range are calculated according to the optimal decision variables.

[0057] Specifically, the first difference is the difference between the lower boundary of the preset low-frequency interval and the lower boundary of the preset variation range, the preset variation range is the variation range when each element in the decision variable is updated, the second difference is the difference between the upper boundary of the preset low-frequency interval and the lower boundary of the preset low-frequency interval, the third difference is the difference between the lower boundary of the preset high-frequency interval and the lower boundary of the preset low-frequency interval, the upper boundary of the preset high-frequency interval is the maximum value of the sum of the upper boundary of the preset low-frequency interval and the fourth difference, the lower boundary of the preset high-frequency interval and the sum of the fourth difference, and the fifth difference is the difference between the upper boundary of the preset frequency range and the upper boundary of the preset high-frequency interval.

[0058] As an example, the constraints of the above optimization problem can be expressed as follows:

[0059]

[0060]

[0061] (Formula 3); As an example, the decision variables of the above optimization problem can be expressed as follows: (Formula 4); In formula 4, represents the decision variable, 、 、 、 and denotes the elements of decision variables, among which, represents the first difference, represents the second difference, represents the third difference, represents the fourth difference, Represents the fifth difference.

[0062] The relationship expressions between each decision variable element and the preset low frequency interval, the preset high frequency interval and the preset frequency range are as follows:

[0063]

[0064] (Formula 5);

[0065]

[0066] In formula five, is the lower boundary of the preset variation range, which can be expressed as , The upper boundary of the preset variation range.

[0067] As an example, the preset change range is the dynamic change range when the elements in the decision variable are updated, which can be: , and wait.

[0068] As an example, the optimization objective of the optimization problem can be expressed as follows:

[0069]

[0070] (Formula 6); in, represents the first optimization objective, represents the second optimization objective, represents the third optimization objective, Max() represents the maximum value function, Represents the Euclidean distance between cluster centers ED The minimum value of Represents the Mahalanobis distance between cluster centers MD The minimum value of and are exponential factors reflecting the relative importance of each distance.

[0071] As a preferred embodiment, the numerical optimization algorithm adopts a differential evolution algorithm.

[0072] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0073] Corresponding to the ultra-high frequency partial discharge fault diagnosis method for a hydro-turbine generator in the above embodiment, Figure 2 The block diagram of the composition of the ultra-high frequency partial discharge fault diagnosis device 400 for a hydro-turbine generator according to an embodiment of the present application is schematically shown. For the sake of convenience of explanation, only the parts related to the embodiment of the present application are shown.

[0074] like Figure 2 As shown, the hydro-generator ultra-high frequency partial discharge fault diagnosis method device 400 includes: A first acquisition module 410 is used to acquire a plurality of ultra-high frequency partial discharge signals collected from the hydro-generator in a current monitoring period; A preprocessing module 420, configured to discretize the ultra-high frequency partial discharge signal to obtain a partial discharge pulse sequence; A cluster analysis module 430 is used to perform cluster analysis on an initial long sequence composed of various partial discharge pulse sequences using the partial discharge pulse sequence as a sample point, and identify each cluster center and its corresponding area; The noise removal module 440 is used to compare the partial discharge pulse sequence corresponding to each cluster center with each typical partial discharge event in the pre-built partial discharge case library one by one to identify the cluster where the noise signal is located, and remove the partial discharge pulse sequence corresponding to the cluster where the noise signal is located from the initial long sequence to obtain a target long sequence; The partial discharge event classification module 450 is used to extract the time domain and / or frequency domain characteristic parameters of each partial discharge pulse sequence in the target long sequence, use the characteristic parameters of each partial discharge pulse sequence as input of a pre-built partial discharge classification model, and use the partial discharge classification model to predict the typical partial discharge event type generated by the hydro-generator in the current monitoring period; A partial discharge source locating module 460, configured to locate a partial discharge source according to the typical partial discharge event type; Wherein, the partial discharge classification model is obtained by training an initial machine learning classifier.

[0075] As an embodiment of the present application, the ultra-high frequency partial discharge fault diagnosis device 400 for a hydro-turbine generator can be implemented as follows: Figure 1 The illustrated embodiments and other related method embodiments.

[0076] The process of each module of the ultra-high frequency partial discharge fault diagnosis method device 400 of the hydro-turbine generator provided in the embodiment of the present application realizing its respective function can be specifically referred to the aforementioned Figure 1 The description of the illustrated embodiment and other related method embodiments will not be repeated here.

[0077] It should be noted that the information interaction, execution process and other contents between the above modules are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0078] Figure 3 A schematic diagram of a computer device according to an embodiment of the present application is shown. In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 3As shown. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05 and a memory (not shown in the figure) connected through a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, a method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator is implemented. The display screen A04 of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device A05 of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse.

[0079] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0080] In one embodiment, the hydro-turbine generator ultra-high frequency partial discharge fault diagnosis device 400 provided in the present application can be implemented in the form of a computer program. The computer program can be used in Figure 3 The computer device shown in the figure is run. The memory of the computer device can store various program modules constituting the ultra-high frequency partial discharge fault diagnosis device 400 for a hydro-turbine generator. The computer program composed of various program modules enables the processor to execute the steps of the ultra-high frequency partial discharge fault diagnosis method for a hydro-turbine generator in various embodiments of the present application described in this specification.

[0081] In one embodiment, the present application further provides a machine-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator in the above embodiment is implemented.

[0082] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0083] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application 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.) that include computer-usable program codes.

[0084] It should also be noted that the terms "include", "comprising" or any other variation thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. 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 method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator, characterized in that: include: Acquire multiple ultra-high frequency partial discharge signals collected from the hydro-generator during the current monitoring period; Discretizing the ultra-high frequency partial discharge signal to obtain a partial discharge pulse sequence; Taking the partial discharge pulse sequence as a sample point, cluster analysis is performed on an initial long sequence composed of various partial discharge pulse sequences to identify each cluster center and its corresponding area; The partial discharge pulse sequences corresponding to the centers of each cluster are compared one by one with each typical partial discharge event in a pre-built partial discharge case library to identify the cluster where the noise signal is located, and the partial discharge pulse sequences corresponding to the cluster where the noise signal is located are removed from the initial long sequence to obtain a target long sequence; Extracting time domain and / or frequency domain characteristic parameters of each partial discharge pulse sequence in the target long sequence, using the characteristic parameters of each partial discharge pulse sequence as input of a pre-constructed partial discharge classification model, and using the partial discharge classification model to predict typical partial discharge event types generated by the hydro-generator in the current monitoring period; Locating the partial discharge source according to the typical partial discharge event type; The partial discharge classification model is obtained by training an initial machine learning classifier.

2. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 1, characterized in that: The characteristic parameter includes at least one of signal amplitude, pulse duration, pulse rise time, maximum frequency and minimum frequency.

3. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 1, characterized in that: The partial discharge case library includes at least one of a first-level partial discharge case and a second-level partial discharge case, the first-level partial discharge case includes at least one of a first typical partial discharge event, a second typical partial discharge event, a third typical partial discharge event, a fourth typical partial discharge event, a fifth typical partial discharge event, a sixth typical partial discharge event, a seventh typical partial discharge event, and an eighth typical partial discharge event, the second-level partial discharge case includes at least one of a ninth typical partial discharge event, a tenth typical partial discharge event, and an eleventh typical partial discharge event, and the fault severity of the first-level partial discharge case is higher than the fault severity of the second-level partial discharge case; Among them, the first typical partial discharge event is a partial discharge event generated when the humidity of the environment in which the hydro-turbine generator is located is greater than the first threshold; the second typical partial discharge event is a partial discharge event generated when the temperature of the environment in which the hydro-turbine generator is located is greater than the second threshold; the third typical partial discharge event is a partial discharge event generated when the electromagnetic interference intensity of the environment in which the hydro-turbine generator is located is greater than the third threshold; the fourth typical partial discharge event is a partial discharge event generated after the aging of the insulating material in the hydro-turbine generator; the fifth typical partial discharge event is a partial discharge event generated by the hydro-turbine generator under mechanical stress; the sixth typical partial discharge event is a partial discharge event generated after the hydro-turbine generator has an insulation defect; the seventh typical partial discharge event is a partial discharge event generated when the hydro-turbine generator is extremely controlled; the eighth typical partial discharge event is a partial discharge event generated when the hydro-turbine generator is under a composite fault condition, and the composite fault condition includes more than one of the following: the humidity of the environment in which the hydro-turbine generator is located is greater than the first threshold, the temperature of the environment in which the hydro-turbine generator is located is greater than the second threshold, and the hydro-turbine generator is subjected to mechanical stress; the ninth typical partial discharge event is a partial discharge event corresponding to different partial discharge fault mechanisms; the tenth typical partial discharge event is a partial discharge event corresponding to different partial discharge generation locations; and the eleventh typical partial discharge event is a partial discharge event corresponding to different environmental conditions.

4. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 1 or 3, characterized in that: The partial discharge classification model is obtained after training the initial machine learning classifier, including: The characteristic parameters extracted from the partial discharge signals when each typical partial discharge event occurs in the partial discharge case library are used as input parameters and the typical partial discharge event type is used as a label value to construct a training sample; The training samples are used to train an initial machine learning classifier to obtain a partial discharge classification model.

5. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 3, characterized in that: The partial discharge source is located according to the typical partial discharge event types, including: If the predicted partial discharge event type is in the first level of partial discharge cases, a first control signal is issued to shut down the hydro-generator, and then a partial discharge source is determined according to the partial discharge event type; If the predicted partial discharge event type is in the second level partial discharge case, the partial discharge source is determined according to the partial discharge event type.

6. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 1, characterized in that: Taking the partial discharge pulse sequence as a sample point, cluster analysis is performed on the initial long sequence composed of various partial discharge pulse sequences to identify each cluster center and its corresponding area, including: The initial long sequence is composed of various partial discharge pulse sequences; Converting each partial discharge pulse sequence in the initial long sequence into the frequency domain to obtain each partial discharge spectrum; Extracting a low power ratio and a high power ratio of the partial discharge spectrum, and forming sample points by the low power ratio and the high power ratio; Perform spectral clustering analysis on each sample point to identify each cluster center and its corresponding area; Among them, the low power ratio is the ratio of the sum of the square of the modulus values ​​of each frequency point in the partial discharge spectrum within a preset low frequency interval to the sum of the square of the modulus values ​​of all frequency points in the preset frequency range in the partial discharge spectrum, and the high power ratio is the ratio of the sum of the square of the modulus values ​​of each frequency point in the preset high frequency interval of the partial discharge spectrum to the sum of the square of the modulus values ​​of all frequency points in the preset frequency range in the partial discharge spectrum. The preset low frequency interval and the preset high frequency interval are both within the preset frequency range, the lower boundary of the preset frequency range is not lower than the minimum frequency in the partial discharge spectrum, and the upper boundary of the preset frequency range is not higher than the maximum frequency in the partial discharge spectrum.

7. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 6, characterized in that: Taking the partial discharge pulse sequence as a sample point, clustering analysis is performed on the initial long sequence composed of each partial discharge pulse sequence to identify each cluster center and its corresponding area, and also includes: A numerical optimization algorithm is used to find the optimal preset low-frequency interval, preset high-frequency interval and preset frequency range to trigger the extraction of the low power ratio and high power ratio of the partial discharge spectrum, and the low power ratio and the high power ratio form sample points.

8. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 7, characterized in that: The numerical optimization algorithm is used to find the optimal preset low frequency interval, preset high frequency interval and preset frequency range, including: The optimization objectives of the optimization problem are to maximize the minimum value of the Euclidean distance between cluster centers in spectral clustering, to maximize the minimum value of the Mahalanobis distance between cluster centers in spectral clustering, and to maximize the minimum value of the product of the Euclidean distance and the Mahalanobis distance between cluster centers in spectral clustering. Setting decision variables for the optimization problem, wherein the decision variables include a first difference, a second difference, a third difference, a fourth difference, and a fifth difference; Setting constraints of the optimization problem, wherein the constraints are set according to the value ranges of the upper and lower boundaries of the preset low-frequency interval and the upper and lower boundaries of the preset high-frequency interval; Find the optimal solution to the optimization problem and obtain the optimal decision variables; Calculating the optimal preset low-frequency interval, preset high-frequency interval and preset frequency range according to the optimal decision variables; Among them, the first difference is the difference between the lower boundary of the preset low-frequency interval and the lower boundary of the preset variation range, the second difference is the difference between the upper boundary of the preset low-frequency interval and the lower boundary of the preset low-frequency interval, the third difference is the difference between the lower boundary of the preset high-frequency interval and the lower boundary of the preset low-frequency interval, the upper boundary of the preset high-frequency interval is the maximum value of the sum of the upper boundary of the preset low-frequency interval and the fourth difference, and the lower boundary of the preset high-frequency interval and the sum of the fourth difference, the fifth difference is the difference between the upper boundary of the preset frequency range and the upper boundary of the preset high-frequency interval, and the preset variation range is the variation range when each element in the decision variable is updated.

9. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-generator according to claim 8, characterized in that: The constraints are: ； in, Indicates the lower boundary of the preset low-frequency range. Indicates the upper boundary of the preset low-frequency range. Indicates the lower boundary of the preset high-frequency range, Indicates the upper boundary of the preset high-frequency range. Indicates the upper boundary of the preset frequency range.

10. The method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator according to any one of claims 7 to 9, characterized in that: The numerical optimization algorithm is a differential evolution algorithm.

11. A hydro-generator ultra-high frequency partial discharge fault diagnosis device, characterized in that: include: A first acquisition module is used to acquire multiple ultra-high frequency partial discharge signals collected from the hydro-generator in the current monitoring period; A preprocessing module, used for discretizing the ultra-high frequency partial discharge signal to obtain a partial discharge pulse sequence; A cluster analysis module, used to perform cluster analysis on an initial long sequence composed of various partial discharge pulse sequences using the partial discharge pulse sequence as a sample point, and identify each cluster center and its corresponding area; A noise removal module is used to compare the partial discharge pulse sequence corresponding to each cluster center with each typical partial discharge event in a pre-built partial discharge case library one by one to identify the cluster where the noise signal is located, and remove the partial discharge pulse sequence corresponding to the cluster where the noise signal is located from the initial long sequence to obtain a target long sequence; A partial discharge event classification module is used to extract time domain and / or frequency domain characteristic parameters of each partial discharge pulse sequence in the target long sequence, use the characteristic parameters of each partial discharge pulse sequence as input of a pre-constructed partial discharge classification model, and use the partial discharge classification model to predict the typical partial discharge event type generated by the hydro-generator in the current monitoring period; A partial discharge source locating module, used for locating the partial discharge source according to the typical partial discharge event type; The partial discharge classification model is obtained by training an initial machine learning classifier.

12. A computer device, characterized in that: include: a memory configured to store instructions; as well as A processor is configured to call the instruction from the memory and implement the hydro-turbine generator ultra-high frequency partial discharge fault diagnosis method according to any one of claims 1 to 10 when executing the instruction.

13. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores instructions for causing a machine to execute the method for diagnosing ultra-high frequency partial discharge faults of a hydro-turbine generator according to any one of claims 1 to 10.

Images