Ultrahigh voltage DC line fault diagnosis method based on PCA and KDE

Abstract

The embodiment of the invention discloses a PCA and KDE-based ultra-high-voltage direct-current line fault diagnosis method, and relates to the technical field of ultra-high-voltage direct-current line fault diagnosis, and the method comprises the steps: obtaining the traveling wave data of a target ultra-high-voltage direct-current line, carrying out the standardization processing of the traveling wave data, and obtaining a fault diagnosis result of the target ultra-high-voltage direct-current line; projecting the data to each principal component through a preset principal component model, and outputting a projection data set; inputting the projection data set into each trained joint probability density model to obtain a probability density value; and outputting a fault diagnosis result based on the probability density value. Characteristic extraction is carried out on the traveling wave signal by utilizing principal component analysis, so that the dimension of data is reduced, and key characteristic information is captured; in addition, joint probability density distribution fitting is carried out on the key features through kernel density estimation, so that efficient comprehensive utilization of multiple fault features is realized, the internal relation among the multiple features can be fully mined, and the comprehensiveness and accuracy of fault diagnosis can be effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of ultra-high voltage direct current line fault diagnosis, and in particular to an ultra-high voltage direct current line fault diagnosis method based on PCA and KDE. Background Art

[0002] Ultra High Voltage Direct Current Transmission (UHVDC) has been widely used in modern power grids due to its advantages of long distance, large capacity and low loss. However, due to the complexity of the operating environment of UHVDC transmission lines and the diversity of fault types, the protection system for UHVDC transmission lines faces severe challenges in terms of reliability, accuracy and rapid response.

[0003] As the main protection of the line, the traveling wave protection is prone to failure to operate in the case of high resistance and remote faults, which will pose a certain threat to the safe and stable operation of the UHV DC system. Related technologies generally improve the traveling wave protection based on the difference in the line mode fault component voltage at the end of the line and its first peak time during different internal and external faults, or the difference in the relationship between the line mode component and the fault resistance during internal and external faults in the line area, but fail to solve the problem of incorrect protection operation. This is because the propagation of the fault traveling wave is affected by multiple factors such as transition resistance and fault distance, and a single feature cannot fully reflect the impact of these complex factors on the line.

[0004] Related technologies can also accurately characterize faults by extracting more fault features from fault data. Different fault features reflect fault information from different angles. The more features that can be used, the more accurately the actual condition of the fault can be described. Therefore, the complementary performance of multiple features can improve the reliability of protection, thereby significantly improving the accuracy of fault diagnosis. However, related technologies do not take into account the inherent connection between multiple different features, and only roughly stack the distribution of multiple features, which cannot accurately reflect the overall characteristics of the fault.

[0005] Therefore, there is currently a lack of a method that can comprehensively consider multiple characteristics and accurately, quickly, and in real time diagnose faults in UHV DC lines. Summary of the invention

[0006] The embodiment of the present application provides a UHV DC line fault diagnosis method based on PCA and KDE to solve the defects of the above-mentioned related technologies. The technical solution is as follows:

[0007] In a first aspect, an embodiment of the present application provides a UHV DC line fault diagnosis method based on PCA and KDE, comprising:

[0008] Acquire traveling wave data of the target UHV DC line;

[0009] Standardizing the traveling wave data, projecting the standardized traveling wave data onto each principal component through a preset principal component model, and outputting a projection data set consisting of the projection data on each principal component;

[0010] Inputting the projection data set into each trained joint probability density model respectively, and obtaining the probability density value output by each joint probability density model;

[0011] Outputting a fault diagnosis result of the target UHV DC line based on the probability density value;

[0012] Each of the joint probability density models corresponds to a fault type.

[0013] In an optional solution of the first aspect, the step of standardization processing specifically includes:

[0014] The data to be processed is standardized so that the mean value of each feature in the data to be processed is 0 and the variance is 1. The standardized data is calculated and the formula is applied:

[0015]

[0016] Wherein, X is the data to be processed, is the mean vector, σ is the standard deviation vector, and Z is the data after the standardization process.

[0017] In an optional solution of the first aspect, the principal component model is constructed based on the following steps, specifically including:

[0018] Get sample data for each fault type;

[0019] After the sample data is standardized, the corresponding covariance matrix is ​​calculated;

[0020] Performing eigenvalue decomposition on the covariance matrix to obtain a plurality of eigenvectors and an eigenvalue corresponding to each of the eigenvectors;

[0021] Arrange each of the eigenvectors in descending order according to the eigenvalue values, and determine that a preset number of eigenvectors ranked first are the main components;

[0022] Construct a principal component model for projecting data onto the determined principal components;

[0023] Y=ZV k ;

[0024] Wherein, Y is the projection data matrix after projection, Z is the data matrix after the standardization process, Ck is the principal component matrix.

[0025] In an optional solution of the first aspect, a joint probability density model corresponding to each fault type is obtained by training based on the following steps, including:

[0026] Acquire sample data of one fault type among the sample data of each fault type;

[0027] Standardizing the sample data of the one fault type, projecting the standardized sample data onto each principal component through a principal component model, and outputting a sample projection data set consisting of the sample projection data on each principal component;

[0028] Determine the kernel function and bandwidth of the joint probability density model, calculate the probability density estimate at each sample projection data by combining the kernel function, the bandwidth and the sample projection data set, and construct a joint probability density model of the corresponding fault type based on the functional relationship between each sample projection data and the corresponding probability density estimate;

[0029] Output the joint probability density model corresponding to each fault type.

[0030] In an optional solution of the first aspect, obtaining sample data of each fault type includes:

[0031] Select different line parameters to traverse and generate various fault conditions;

[0032] Generate a parameter matrix based on each of the fault conditions; wherein each row of the parameter matrix represents a line parameter combination corresponding to the fault condition, and each column of the parameter matrix represents a value of the corresponding line parameter;

[0033] Inputting the parameter matrix into the constructed UHV DC system digital twin model, and simulating the UHV DC system digital twin model to obtain fault parameters corresponding to each fault type;

[0034] The fault parameters during the simulation process are used as sample input, and the corresponding fault type is used as sample output to obtain sample data.

[0035] In an optional solution of the first aspect, outputting the fault diagnosis result of the target ultra-high voltage direct current line based on the probability density value includes:

[0036] Comparing the probability density value output by each of the joint probability density models with the corresponding fault diagnosis probability density threshold value;

[0037] When the value is greater than or equal to the corresponding fault diagnosis probability density threshold, a fault diagnosis indicator with a value of 1 is output, otherwise a fault diagnosis indicator with a value of 0 is output;

[0038] forming a fault diagnosis vector based on all the fault diagnosis indicators;

[0039] If the value of at most one fault diagnosis indicator in the fault diagnosis vector is 1, determining the corresponding fault type according to the fault diagnosis vector and outputting a fault diagnosis result;

[0040] Otherwise, determine the maximum value of the probability density value corresponding to each fault diagnosis indicator in the fault diagnosis vector, set the fault diagnosis indicator corresponding to the maximum value to 1, and set the other fault diagnosis indicators to 0 to obtain an updated fault diagnosis vector; determine the corresponding fault type according to the updated fault diagnosis vector, and output the fault diagnosis result.

[0041] In an optional solution of the first aspect, after acquiring the traveling wave data of the target ultra-high voltage direct current line, the method further includes:

[0042] When the average value of the absolute value of the line voltage difference of the target ultra-high voltage DC line is greater than the preset line voltage difference, it is determined that a fault signal exists in the target ultra-high voltage DC line, and the process proceeds to the step of normalizing the traveling wave data.

[0043] In a second aspect, the embodiment of the present application further provides a UHV DC line fault diagnosis device based on PCA and KDE, comprising:

[0044] A data acquisition module, used to obtain traveling wave data of the target UHV DC line;

[0045] A feature extraction module is used to perform standardization processing on the traveling wave data, project the standardized traveling wave data onto each principal component through a preset principal component model, and output a projection data set consisting of the projection data on each principal component;

[0046] A probability density estimation module, used to input the projection data set into each trained joint probability density model respectively, and obtain the probability density value output by each joint probability density model;

[0047] A fault diagnosis module, configured to output a fault diagnosis result of the target UHV DC line based on the probability density value;

[0048] Each of the joint probability density models corresponds to a fault type.

[0049] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method provided in the first aspect of the embodiment of the present application or any one of the implementation methods of the first aspect is implemented.

[0050] In a fourth aspect, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided by the first aspect of the embodiment of the present application or any one of the implementations of the first aspect.

[0051] The beneficial effects brought about by the technical solutions provided by some embodiments of the present application include at least:

[0052] The embodiment of the present application provides a UHV DC line fault diagnosis method based on PCA and KDE, which uses principal component analysis (PCA) to extract features of traveling wave signals, thereby reducing the dimension of data and capturing key feature information; the present application also uses kernel density estimation (KDE) to fit the joint probability density distribution of key features, and constructs a multi-fault diagnosis model based on the joint probability density function of multiple types of faults, thereby achieving efficient and comprehensive utilization of multiple fault features, and can fully explore the intrinsic connections between multiple features, and use them as a whole to perform fault diagnosis. Compared with traditional methods, it has higher accuracy and reliability, and can effectively improve the comprehensiveness and precision of diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 It is one of the flow charts of a UHV DC line fault diagnosis method based on PCA and KDE provided in an embodiment of the present application;

[0055] Figure 2 This is a second flow chart of a UHV DC line fault diagnosis method based on PCA and KDE provided in an embodiment of the present application;

[0056] Figure 3 It is a schematic diagram of a fault point of a UHV DC line provided in an embodiment of the present application;

[0057] Figure 4 It is a structural schematic diagram of a UHV DC line fault diagnosis device based on PCA and KDE provided in an embodiment of the present application;

[0058] Figure 5 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0060] The terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or modules is not limited to the listed steps or modules, but may optionally include steps or modules that are not listed, or may optionally include other steps or modules that are inherent to these processes, methods, products or devices.

[0061] It should be noted that the terms "first\second" involved in the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that "first\second" can be interchanged with a specific order or sequence where permitted. It should be understood that the objects distinguished by "first\second" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those described or illustrated herein.

[0062] The present application is described in detail below with reference to specific embodiments.

[0063] Next, combine Figure 1 , introduces a UHV DC line fault diagnosis method based on PCA and KDE provided in the embodiment of the present application. For details, please refer to Figure 1 , Figure 1 FIG. 1 is a flow chart of a method for diagnosing UHV DC line faults based on PCA and KDE provided in an embodiment of the present application. Figure 1 As shown, the method comprises the following steps:

[0064] S101, obtaining traveling wave data of a target UHV DC line;

[0065] S102, performing standardization processing on the traveling wave data, projecting the standardized traveling wave data onto each principal component respectively through a preset principal component model, and outputting a projection data set consisting of the projection data on each principal component;

[0066] S103, inputting the projection data set into each trained joint probability density model respectively, and obtaining the probability density value output by each joint probability density model;

[0067] S104: outputting a fault diagnosis result of the target UHV DC line based on the probability density value.

[0068] It should be noted that in UHV DC lines, when an abnormality occurs in the line, the traveling wave stage usually lasts for 5-8ms. In this stage, the traveling wave signal is mainly composed of transient high-frequency electromagnetic waves generated by the fault point. Because it has not been affected by the control system, it retains more original fault characteristic information, so the fault type can be diagnosed.

[0069] It should be noted that the fault diagnosis performed in the embodiment of the present application can be understood as a process of detecting abnormal signals on the line and then determining the type of fault.

[0070] In some embodiments, in S101, traveling wave sensors may be installed at multiple sampling positions on the UHV DC line to collect traveling wave data at corresponding points, and traveling wave data may also be collected at multiple sampling points based on a preset sampling frequency.

[0071] Exemplarily, the sampling frequency can be set to tens to hundreds of kHz, and the sampling points can include key nodes such as converter stations and substations. Up to hundreds of sampling points can be set to accurately capture the traveling wave data of corresponding points on the UHV DC line. The embodiments of the present application are not limited to this.

[0072] It should be noted that related technologies generally only perform univariate analysis on the collected traveling wave features, and in univariate analysis, any common variation with other variables is explicitly ignored, which may cause important features to be ignored. In general, there are a large number of sampling points, and the dimension of the collected traveling wave data is high, resulting in the distance between data points being too close, which will affect the performance of classification and clustering algorithms. In addition, high-dimensional data usually contains more noise, which will significantly increase the complexity of the calculation, and may cause delays in diagnostic results when performing real-time fault diagnosis.

[0073] Therefore, in order to perform fault diagnosis more effectively, it is necessary to reduce the dimensionality of these data and extract the main information contained therein, so as to simplify the analysis process and improve the accuracy and efficiency of diagnosis.

[0074] In some embodiments, S102 can reduce the dimensionality of high-dimensional data by principal component analysis. Principal component analysis maps high-dimensional data to a low-dimensional subspace through linear transformation, extracts the main variation information in the data, thereby simplifying the analysis process, reducing the impact of noise, and improving the accuracy and efficiency of diagnosis.

[0075] It should be noted that when performing principal component analysis on multiple input data at the same time, it is essentially constructing a new coordinate system composed of principal components so that the projection of the input data on these principal components can maximize the explanation of the overall variance. Principal components of the same order explain the same variance in each input data, and these principal components jointly describe the main variation characteristics of each input data. In this case, if all input data correspond to the same type of fault, each principal component can be regarded as a projection of the fault feature in a specific direction, and all significant principal components jointly define the fault feature space.

[0076] Each of the joint probability density models corresponds to a fault type.

[0077] Specifically, in S102, the input traveling wave data needs to be standardized first, so that the mean value of each feature in the data to be processed is 0 and the variance is 1, and the standardized data is calculated and the formula is applied:

[0078]

[0079] Among them, X is the data to be processed, is the mean vector, σ is the standard deviation vector, and Z is the standardized data.

[0080] Furthermore, the standardized data is input into a preset principal component model, and the standardized traveling wave data is projected onto each principal component through the principal component model, and the projection data on each principal component is output to obtain a projection data set composed of all projection data.

[0081] Among them, the preset principal component model can be pre-constructed based on the sample data of each fault type, and the principal components of the sample data are determined during the construction process, so that when actually performing fault diagnosis, the input data can be directly projected onto the determined principal components through the preset principal component model.

[0082] Therefore, through step S102, the normalized traveling wave data can be projected onto each principal component to obtain the data after dimension reduction, that is, the projection data.

[0083] Further, step S103 is executed to process each projection data in the projection data set respectively through a plurality of trained joint probability density models, and each joint probability density model outputs a corresponding probability density value respectively.

[0084] It can be understood that one joint probability density model corresponds to one fault type, and the probability density value on the same fault type is calculated by combining the joint probability density model of a single fault type with the input data.

[0085] Further, in S104, fault diagnosis may be performed according to the probability density values ​​output by all joint probability density models, and a fault diagnosis result of the target UHV DC line may be output.

[0086] Specifically, the probability density value may be compared with a corresponding probability density threshold, and the corresponding fault type may be determined according to the comparison result.

[0087] It should be noted that each joint probability density model is trained based on a training sample of a fault type during the training phase. The probability density distribution output by the joint probability density model reflects the probability of normal data without faults falling into different density areas. In high-density areas, the probability of normal data points appearing is higher; in low-density areas, the probability of normal data points appearing is lower. In order to determine whether new data is abnormal, a density threshold can be set. Generally, a lower percentile of the normal data density value can be selected as the probability density threshold.

[0088] In some embodiments, the specific step of outputting the fault diagnosis result of the target UHV DC line based on the probability density value in S104 includes:

[0089] Comparing the probability density value output by each of the joint probability density models with the corresponding fault diagnosis probability density threshold value;

[0090] When the value is greater than or equal to the corresponding fault diagnosis probability density threshold, a fault diagnosis indicator with a value of 1 is output, otherwise a fault diagnosis indicator with a value of 0 is output;

[0091] forming a fault diagnosis vector based on all the fault diagnosis indicators;

[0092] If the value of at most one fault diagnosis indicator in the fault diagnosis vector is 1, determining the corresponding fault type according to the fault diagnosis vector and outputting a fault diagnosis result;

[0093] Otherwise, determine the maximum value of the probability density value corresponding to each fault diagnosis indicator in the fault diagnosis vector, set the fault diagnosis indicator corresponding to the maximum value to 1, and set the other fault diagnosis indicators to 0 to obtain an updated fault diagnosis vector; determine the corresponding fault type according to the updated fault diagnosis vector, and output the fault diagnosis result.

[0094] In a specific embodiment, the fault types include positive grounding fault, negative grounding fault, bipolar grounding fault and fault outside the line area, and the trained joint probability density model includes a positive grounding fault model corresponding to the positive grounding fault, a negative grounding fault model corresponding to the negative grounding fault and a bipolar grounding fault model corresponding to the bipolar grounding fault. Figure 2 As shown, the specific steps of outputting the fault diagnosis result of the target UHV DC line based on the probability density value include:

[0095] S201, a probability density value X is calculated based on a positive pole grounding fault model, a probability density value Y is calculated based on a negative pole grounding fault model, and a probability density value Z is calculated based on a bipolar grounding fault model.

[0096] S202, respectively comparing the probability density values ​​with corresponding fault diagnosis probability density thresholds;

[0097] When the probability density value is greater than or equal to the corresponding fault diagnosis probability density threshold, a fault diagnosis indicator with a value of 1 is output, otherwise a fault diagnosis indicator with a value of 0 is output;

[0098] Specifically, the fault diagnosis indicator corresponding to the positive pole grounding fault is recorded as x, the fault diagnosis probability density threshold corresponding to the positive pole grounding fault is recorded as δ1, the fault diagnosis indicator corresponding to the negative pole grounding fault is recorded as y, the fault diagnosis probability density threshold corresponding to the negative pole grounding fault is recorded as δ2, the fault diagnosis indicator corresponding to the bipolar grounding fault is recorded as z, and the fault diagnosis probability density threshold corresponding to the bipolar grounding fault is recorded as δ3.

[0099] S203, constructing a fault diagnosis vector based on each fault diagnosis indicator;

[0100] Specifically, the fault diagnosis vector can be recorded as (x, y, z), where x, y, z∈{0, 1}.

[0101] S204: In the fault diagnosis vector, if at most one element has a value of 1, that is, x+y+z≤1, the corresponding fault type can be directly determined based on the fault diagnosis vector.

[0102] Exemplarily, in the three-dimensional vector, at most one value is 1, including the following situations:

[0103] If x=1 and other values ​​are 0, the three-dimensional vector is represented as [1,0,0], and the fault type can be determined to be a positive grounding fault;

[0104] If y=1 and other values ​​are 0, the three-dimensional vector is represented as [0,1,0], and the fault type can be determined to be a negative grounding fault;

[0105] If z = 1 and the other values ​​are 0, the three-dimensional vector is represented as [0, 0, 1], and the fault type can be determined to be a bipolar grounding fault;

[0106] If x, y, and z are all 0, the three-dimensional vector is represented as [0, 0, 0], and the fault type can be determined to be an out-of-line fault, indicating that the fault is not located in the current line section.

[0107] S205, in the fault diagnosis vector, if the value of more than one element is 1, that is, x+y+z>1, determine the maximum value of the probability density value corresponding to each fault diagnosis indicator in the fault diagnosis vector, set the fault diagnosis indicator corresponding to the maximum value to 1, and set the other fault diagnosis indicators to 0, to obtain an updated fault diagnosis vector;

[0108] It can be expressed as the following formula:

[0109]

[0110] The step S204 is executed according to the updated fault diagnosis vector to determine the corresponding fault type and output the fault diagnosis result.

[0111] In some embodiments, after S101, in order to reduce the risk of false triggering due to noise interference and abnormal values, the absolute value average of the line voltage difference can be used as a criterion for protection startup. Specifically, it can be determined whether the corresponding traveling wave data is an abnormal signal caused by false triggering according to the absolute value average of the line voltage difference. The formula can be applied:

[0112]

[0113] When the above inequality is satisfied, that is, the average value of the absolute value of the line voltage difference of the target ultra-high voltage DC line is greater than the preset line voltage difference Δu, it is determined that there is a fault signal in the target ultra-high voltage DC line, and the process goes to S102 to perform standardization on the traveling wave data. For the subsequent execution process, please refer to the description in S102-S104, which will not be repeated here.

[0114] Where u is the line voltage and n is the number of sample values.

[0115] In some embodiments, before executing the UHV DC line fault diagnosis method of S101 to S104, the principal component model in S102 and the joint probability density model in S103 may be obtained by training with training samples.

[0116] Specifically, the steps of obtaining training samples include:

[0117] First, different line parameters can be selected to traverse and generate a variety of fault conditions;

[0118] Among them, different line parameters include but are not limited to different wiring methods, power levels, voltage levels, transition resistances, fault durations and fault location parameters, etc. One or more of these parameters can be selected and combined to construct a fault operating condition, which is not limited in the embodiments of the present application.

[0119] For example, Figure 3 The figure is a schematic diagram of a fault point of a UHV DC line provided by an exemplary embodiment of the present application. Figure 3 In the example UHV DC line model, the fault points include but are not limited to the following:

[0120] F1: Pole 1 line ground fault;

[0121] F2: Pole 2 line ground fault;

[0122] F3: Bipolar line ground fault;

[0123] F4: Ground fault in the valve area of ​​the rectifier station;

[0124] F5: Rectifier station valve short circuit fault;

[0125] F6: Rectifier station AC bus grounding fault;

[0126] F7: Rectifier station busbar grounding fault;

[0127] F8: ground fault in the inverter station valve area;

[0128] F9: Inverter station valve short circuit fault;

[0129] F10: Inverter station AC bus grounding fault;

[0130] F11: Inverter station busbar grounding fault.

[0131] Each fault point corresponds to a fault type, and a fault condition can be constructed based on each fault type and corresponding line parameters such as the value of the fault distance and the value of the transition resistance.

[0132] Furthermore, a parameter matrix may be generated based on each fault condition; wherein each row of the parameter matrix represents a line parameter combination corresponding to the fault condition, and each column of the parameter matrix represents a value of a corresponding line parameter.

[0133] Further, the parameter matrix is ​​input into the constructed UHV DC system digital twin model, and the fault parameters corresponding to each fault type are obtained based on the simulation of the UHV DC system digital twin model;

[0134] Specifically, the fault parameters obtained at the time of the fault include but are not limited to current and voltage recording data at both ends of the UHV DC line.

[0135] Finally, the fault parameters obtained during the simulation process can be used as the input of the sample, and the corresponding fault type can be used as the output of the sample to obtain sample data.

[0136] It should be noted that a digital twin model of the UHV DC system can be constructed in the power system electromagnetic transient simulation software PSCAD / EMTDC, so that a large number of training samples can be generated through simulation, greatly expanding the number of samples, and achieving comprehensive coverage of various fault conditions, so that the model can learn more intrinsic connections between features, avoiding the defect of poor model accuracy due to insufficient number of samples when training only based on historical data.

[0137] Specifically, a principal component model can be constructed based on sample data, which includes the following steps:

[0138] First, sample data for each fault type constructed in the above steps may be obtained.

[0139] Furthermore, after standardizing the sample data, the corresponding covariance matrix is ​​calculated. The specific application formula is:

[0140]

[0141] Among them, Z is the data matrix after standardization, n is the number of samples, and C is the covariance matrix. The covariance matrix can reflect the linear correlation between various features.

[0142] Further, the covariance matrix is ​​subjected to eigenvalue decomposition to obtain a plurality of eigenvectors and an eigenvalue corresponding to each of the eigenvectors, and the formula is applied:

[0143] CV = VΛ;

[0144] Among them, V is the eigenvector matrix, Λ is a diagonal matrix, and the diagonal elements of the diagonal matrix are eigenvalues.

[0145] Specifically, the eigenvalue represents the amount of variation explained by each principal component, and the eigenvector represents the direction of each principal component.

[0146] Next, each of the eigenvectors is arranged in descending order according to the numerical value of the eigenvalue, and a preset number of eigenvectors at the top of the order are determined as principal components.

[0147] For example, the first three eigenvectors may be selected as principal components, and the embodiment of the present application does not limit the number of principal components.

[0148] Finally, a principal component model is constructed to project the data onto the determined principal components;

[0149] Each principal component model can be expressed as the following formula, which can project the data onto the determined principal components, including:

[0150] Y=ZV k ;

[0151] Wherein, Y is the projection data matrix after projection, Z is the data matrix after the standardization process, V k is the principal component matrix.

[0152] It should be noted that the principal component model can be constructed based on the sample data of each fault type, and the principal component model corresponding to each fault type can be constructed. For example, if there are two fault types, the principal component model 1 can be constructed based on the sample data of fault type 1, and the principal component model 2 can be constructed based on the sample data of fault type 2.

[0153] Optionally, the sample data of each fault type may be used as a whole to construct the same principal component model, which is not limited in the embodiments of the present application.

[0154] In some embodiments, a joint probability density model corresponding to each fault type may be further obtained based on the sample data and the constructed principal component model training, including the following steps:

[0155] First, sample data of one fault type among the sample data of each fault type may be obtained; it can be understood that the sample data of a single fault type is only used to train a joint probability density model of the corresponding fault type.

[0156] Furthermore, the sample data of the one fault type is standardized, and the standardized sample data is projected onto each principal component through a principal component model, and a sample projection data set consisting of the sample projection data on each principal component is output.

[0157] Furthermore, the kernel function and bandwidth of the joint probability density model are determined, and the probability density estimate at each sample projection data is calculated by combining the kernel function, the bandwidth and the sample projection data set. Based on the functional relationship between each sample projection data and the corresponding probability density estimate, a joint probability density model of the corresponding fault type is constructed.

[0158] For example, given a dataset containing n samples {x1, x2, ..., x n}, the mathematical expression of the joint probability density function of each sample with d variables is:

[0159]

[0160] Among them, K is a multidimensional kernel function, and the commonly used multidimensional Gaussian kernel function is:

[0161]

[0162] h is the bandwidth parameter, which is used to control the width of the kernel function and affect the degree of smoothing. Generally, the bandwidth parameter can be selected based on the empirical formula of sample standard deviation and sample size, or by cross-validation method.

[0163] d is the dimension of the data.

[0164] Finally, the joint probability density model corresponding to each fault type is output.

[0165] In some embodiments, in combination with the sample data, the probability density distribution of the corresponding fault type can be calculated based on the trained joint probability density model;

[0166] The probability density distribution can reflect the probability of normal data in each probability density area, and then the fault diagnosis probability density threshold of each fault type can be determined based on the probability density distribution. For example, a probability density of 1% can be selected as the fault diagnosis probability density threshold for judging abnormal points.

[0167] The following are device embodiments of the present application, which can be used to execute the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0168] See next Figure 4, is a structural diagram of a UHV DC line fault diagnosis device based on PCA and KDE provided by an exemplary embodiment of the present application. The device can be implemented as all or part of a terminal through software, hardware, or a combination of both, and can also be integrated on a server as an independent module. A UHV DC line fault diagnosis device based on PCA and KDE in an embodiment of the present application can be applied to a terminal or a cloud. The device 40 includes a data acquisition module 401, a feature extraction module 402, a probability density estimation module 403, and a fault diagnosis module 404, wherein:

[0169] The data acquisition module 401 is used to obtain the traveling wave data of the target UHV DC line;

[0170] The feature extraction module 402 is used to perform standardization processing on the traveling wave data, project the standardized traveling wave data onto each principal component through a preset principal component model, and output a projection data set consisting of the projection data on each principal component;

[0171] The probability density estimation module 403 is used to input the projection data set into each trained joint probability density model respectively, and obtain the probability density value output by each joint probability density model;

[0172] The fault diagnosis module 404 is used to output the fault diagnosis result of the target UHV DC line based on the probability density value;

[0173] It should be noted that the device 40 provided in the above embodiment only uses the division of the above functional modules as an example when executing the UHV DC line fault diagnosis method based on PCA and KDE. In actual applications, the above functional distribution can be completed by different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device provided in the above embodiment and the embodiment of the UHV DC line fault diagnosis method based on PCA and KDE belong to the same concept. The embodiment and implementation process thereof are detailed in the method embodiment, which will not be repeated here.

[0174] An embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any of the above-mentioned method embodiments when executing the program.

[0175] See also Figure 5 , which is a structural block diagram of an electronic device provided in an embodiment of the present application.

[0176] like Figure 5 As shown, the electronic device 500 includes: a processor 501 and a memory 502 .

[0177] In the embodiment of the present application, the processor 501 is the control center of the computer system, which can be a processor of a physical machine or a processor of a virtual machine. The processor 501 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 501 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array).

[0178] The processor 501 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also called a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state.

[0179] The memory 502 may include one or more computer-readable storage media, which may be non-transitory. The memory 502 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments of the present application, the non-transitory computer-readable storage medium in the memory 502 is used to store at least one instruction, which is used to be executed by the processor 501 to implement the method in the embodiment of the present application.

[0180] In some embodiments, the electronic device 500 further includes: a peripheral device interface 503 and at least one peripheral device 504. The processor 501, the memory 502 and the peripheral device interface 503 can be connected via a bus or a signal line. Each peripheral device 504 can be connected to the peripheral device interface 503 via a bus, a signal line or a circuit board. Specifically, the peripheral device 504 includes: a display screen, a camera and an audio circuit. The peripheral device interface 503 can be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 501 and the memory 502.

[0181] In some embodiments of the present application, the processor 501, the memory 502, and the peripheral device interface 503 are integrated on the same chip or circuit board; in some other embodiments of the present application, any one or two of the processor 501, the memory 502, and the peripheral device interface 503 can be implemented on a separate chip or circuit board. This embodiment of the present application does not specifically limit this.

[0182] The electronic device structure block diagram shown in the embodiment of the present application does not constitute a limitation on the electronic device 500. The electronic device 500 may include more or fewer components than shown in the figure, or combine certain components, or adopt a different component arrangement.

[0183] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method of any of the above embodiments are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0184] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution can be essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0185] 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 UHV DC line fault diagnosis method based on PCA and KDE, characterized in that: include: Acquire traveling wave data of the target UHV DC line; Standardizing the traveling wave data, projecting the standardized traveling wave data onto each principal component through a preset principal component model, and outputting a projection data set consisting of the projection data on each principal component; Inputting the projection data set into each trained joint probability density model respectively, and obtaining the probability density value output by each joint probability density model; Outputting a fault diagnosis result of the target UHV DC line based on the probability density value; Each of the joint probability density models corresponds to a fault type.

2. The method for UHV DC line fault diagnosis based on PCA and KDE according to claim 1, characterized in that: The steps of the standardization process specifically include: The data to be processed is standardized so that the mean value of each feature in the data to be processed is 0 and the variance is 1. The standardized data is calculated and the formula is applied: Wherein, X is the data to be processed, is the mean vector, σ is the standard deviation vector, and Z is the data after the standardization process.

3. A UHV DC line fault diagnosis method based on PCA and KDE according to claim 1 or 2, characterized in that: The principal component model is constructed based on the following steps, which specifically include: Get sample data for each fault type; After the sample data is standardized, the corresponding covariance matrix is ​​calculated; Performing eigenvalue decomposition on the covariance matrix to obtain a plurality of eigenvectors and an eigenvalue corresponding to each of the eigenvectors; Arrange each of the eigenvectors in descending order according to the eigenvalue values, and determine that a preset number of eigenvectors ranked first are the main components; Construct a principal component model for projecting data onto the determined principal components; Y=ZV k ; Wherein, Y is the projection data matrix after projection, Z is the data matrix after the standardization process, V k is the principal component matrix.

4. The method for UHV DC line fault diagnosis based on PCA and KDE according to claim 3, characterized in that: The joint probability density model corresponding to each fault type is trained based on the following steps, including: Acquire sample data of one fault type among the sample data of each fault type; Standardizing the sample data of the one fault type, projecting the standardized sample data onto each principal component through a principal component model, and outputting a sample projection data set consisting of the sample projection data on each principal component; Determine the kernel function and bandwidth of the joint probability density model, calculate the probability density estimate at each sample projection data by combining the kernel function, the bandwidth and the sample projection data set, and construct a joint probability density model of the corresponding fault type based on the functional relationship between each sample projection data and the corresponding probability density estimate; Output the joint probability density model corresponding to each fault type.

5. The method for UHV DC line fault diagnosis based on PCA and KDE according to claim 4, characterized in that: The obtaining of sample data for each fault type includes: Select different line parameters to traverse and generate various fault conditions; Generate a parameter matrix based on each of the fault conditions; wherein each row of the parameter matrix represents a line parameter combination corresponding to the fault condition, and each column of the parameter matrix represents a value of the corresponding line parameter; Inputting the parameter matrix into the constructed UHV DC system digital twin model, and simulating the UHV DC system digital twin model to obtain fault parameters corresponding to each fault type; The fault parameters during the simulation process are used as sample input, and the corresponding fault type is used as sample output to obtain sample data.

6. The method for UHV DC line fault diagnosis based on PCA and KDE according to claim 1, characterized in that: Outputting the fault diagnosis result of the target UHV DC line based on the probability density value includes: Comparing the probability density value output by each of the joint probability density models with the corresponding fault diagnosis probability density threshold value; When the value is greater than or equal to the corresponding fault diagnosis probability density threshold, a fault diagnosis indicator with a value of 1 is output, otherwise a fault diagnosis indicator with a value of 0 is output; forming a fault diagnosis vector based on all the fault diagnosis indicators; If the value of at most one fault diagnosis indicator in the fault diagnosis vector is 1, determining the corresponding fault type according to the fault diagnosis vector and outputting a fault diagnosis result; Otherwise, determine the maximum value of the probability density value corresponding to each fault diagnosis indicator in the fault diagnosis vector, set the fault diagnosis indicator corresponding to the maximum value to 1, and set the other fault diagnosis indicators to 0 to obtain an updated fault diagnosis vector; determine the corresponding fault type according to the updated fault diagnosis vector, and output the fault diagnosis result.

7. The method for UHV DC line fault diagnosis based on PCA and KDE according to claim 1, characterized in that: After obtaining the traveling wave data of the target UHV DC line, the method further includes: When the average value of the absolute value of the line voltage difference of the target ultra-high voltage DC line is greater than the preset line voltage difference, it is determined that a fault signal exists in the target ultra-high voltage DC line, and the process proceeds to the step of normalizing the traveling wave data.

8. A UHV DC line fault diagnosis device based on PCA and KDE, characterized in that: include: A data acquisition module, used to obtain traveling wave data of the target UHV DC line; A feature extraction module is used to perform standardization processing on the traveling wave data, project the standardized traveling wave data onto each principal component through a preset principal component model, and output a projection data set consisting of the projection data on each principal component; A probability density estimation module, used to input the projection data set into each trained joint probability density model respectively, and obtain the probability density value output by each joint probability density model; A fault diagnosis module, configured to output a fault diagnosis result of the target UHV DC line based on the probability density value; Each of the joint probability density models corresponds to a fault type.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images