Self-inspection methods, systems, and storage media for power equipment based on field sensors

Abstract

This invention relates to the field of power equipment monitoring technology, specifically to a power equipment self-inspection method, system, and storage medium based on field sensors. The invention comprehensively acquires multi-parameter operating data of power equipment through multi-dimensional data acquisition and standardized processing, and standardizes the data using statistical features. Secondly, it extracts harmonic features from the upper and lower halves of the current data, enhancing the perception of nonlinear faults and abnormal states, and achieving accurate detection of potential faults such as harmonic pollution and equipment aging. A power equipment diagnostic model is constructed using a support vector machine model to obtain the probability of common faults. Single-equipment fault identifiers are aggregated into a fault identification matrix, and the linkage and similarity of faults within and between equipment chains are analyzed, effectively revealing fault propagation paths and system-level risks.

Description

Technical Field

[0001] This invention relates to the field of power equipment monitoring technology, specifically to a power equipment self-testing method, system, and storage medium based on field sensors. Background Technology

[0002] With the rapid development of smart grids and the Internet of Things for power, the requirements for the reliability and safety of power equipment are increasing. During operation, power equipment is affected by multiple factors, including electrical, mechanical, and thermodynamic factors, which may lead to faults such as equipment aging, insulation damage, overload, and harmonic pollution, thereby causing system paralysis or safety accidents.

[0003] Traditional power equipment monitoring methods mainly rely on regular manual inspections, basic parameter threshold alarms, and single fault diagnosis models, which have the following limitations: First, traditional methods usually only focus on the threshold exceedance of basic electrical parameters such as voltage and current, while ignoring the deep correlation of multi-dimensional information such as harmonics, vibration, and temperature, and cannot fully reflect the health status of the equipment.

[0004] Secondly, methods based on simple thresholds or rules are difficult to distinguish complex fault types and are not sensitive to early latent faults, making them prone to missed or false alarms.

[0005] Finally, power equipment typically operates in series chains or parallel groups, and faults may propagate between different devices, but existing technologies lack effective means to analyze the linkage between faults between devices.

[0006] To address the aforementioned issues, it is necessary to propose a self-testing method, system, and storage medium for power equipment based on field sensors. Summary of the Invention

[0007] The purpose of this invention is to solve the problems existing in the background art, and to propose a self-testing method, system and storage medium for power equipment based on field sensors.

[0008] The objective of this invention can be achieved through the following technical solutions:

[0009] The self-testing method, system, and storage medium for power equipment based on field sensors include a self-testing method for power equipment based on field sensors, a self-testing system for power equipment based on field sensors, and a storage medium for self-testing power equipment based on field sensors.

[0010] In a first aspect, the present invention provides a self-testing method for power equipment based on field sensors, comprising the following steps:

[0011] Step S100: Data acquisition and standardization processing;

[0012] Access the sensor group at the power equipment site to acquire power equipment operation data; standardize the collected power equipment operation data to obtain a standard dataset of operation data for each power equipment; extract the upper and lower whiskers of the input and output currents through statistical feature extraction.

[0013] In a preferred embodiment of the present invention, the collected power equipment operation data is standardized and converted into a standard data format. The specific process is as follows:

[0014] Obtain the mean, standard deviation, median, first quartile, third quartile, upper quartile, and lower quartile of the operating data for each device. Generate a standard dataset of power equipment operating data based on the mean, standard deviation, first quartile, and third quartile of the operating data for each device.

[0015] The upper and lower limits of the input and output currents of each power device are sent to step S200.

[0016] Send the standard dataset of operating data of each power device to step S300;

[0017] Step S200: Harmonic feature extraction;

[0018] Harmonic features are extracted based on the upper and lower whiskers of the input and output currents to obtain a harmonic feature dataset.

[0019] Fast Fourier Transform (FFT) is performed on the upper and lower whisker portions of the standard datasets of input and output currents for various power devices to convert them from time-domain signals to frequency-domain signals. Frequency-domain features of the input and output currents are then extracted from the frequency-domain signals, including:

[0020] The dominant frequency, harmonic amplitude, harmonic energy, and total harmonic distortion of the upper and lower parts of the input current standard dataset, the upper and lower parts of the output current standard dataset, and the lower part of the output current standard dataset.

[0021] In a preferred embodiment of the present invention, a harmonic feature dataset for each power device is generated, specifically a high-dimensional vector containing 32 elements, wherein the elements are as follows:

[0022] Elements 1-8: The main frequency of the upper part of the input current standard dataset, the harmonic amplitude of the frequency component three times the main frequency, the harmonic amplitude of the frequency component five times the main frequency, the harmonic amplitude of the frequency component seven times the main frequency, the harmonic energy of the third main frequency, the harmonic energy of the fifth main frequency, the harmonic energy of the seventh main frequency, and the total harmonic distortion.

[0023] Elements 9-16: The dominant frequency, harmonic amplitude of the frequency component three times the dominant frequency, harmonic amplitude of the frequency component five times the dominant frequency, harmonic amplitude of the frequency component seven times the dominant frequency, harmonic energy of the third dominant frequency, harmonic energy of the fifth dominant frequency, harmonic energy of the seventh dominant frequency, and total harmonic distortion in the lower part of the input current standard dataset.

[0024] Elements 17-24: The main frequency, harmonic amplitude of the frequency component three times the main frequency, harmonic amplitude of the frequency component five times the main frequency, harmonic amplitude of the frequency component seven times the main frequency, harmonic energy of the third main frequency, harmonic energy of the fifth main frequency, harmonic energy of the seventh main frequency, and total harmonic distortion in the upper part of the standard dataset of output current.

[0025] Elements 25-32: The main frequency, harmonic amplitude of the frequency component three times the main frequency, harmonic amplitude of the frequency component five times the main frequency, harmonic amplitude of the frequency component seven times the main frequency, harmonic energy of the third main frequency, harmonic energy of the fifth main frequency, harmonic energy of the seventh main frequency, and total harmonic distortion in the lower part of the standard dataset of output current.

[0026] The generated harmonic feature datasets of each power device are sent to step S300.

[0027] Step S300: Historical data modeling and fault determination;

[0028] A power equipment diagnostic model is established based on historical data from standard datasets and harmonic feature datasets of various power equipment operation data. By inputting the collected real-time standard datasets and harmonic feature datasets into the power equipment diagnostic model, fault identifiers representing fault types are obtained.

[0029] The power equipment diagnostic model includes several support vector machine models.

[0030] In a preferred embodiment of the present invention, the total number of support vector machine models is the same as the total number of power devices. There is a one-to-one correspondence between support vector machines and power devices; that is, each power device has a unique support vector machine model corresponding to itself.

[0031] As a preferred embodiment of the present invention, each support vector machine model includes an input layer, a kernel function layer, and an output layer.

[0032] Specifically, some nodes in the input layer are responsible for receiving elements from the standard dataset and harmonic feature dataset.

[0033] Specifically, the kernel layer has several nodes, each containing a Gaussian radial basis function (RBF) kernel. The Gaussian RBF kernel elevates all the data received from the nodes of the input layer to a higher-dimensional space, extracts regression information about classification in the higher-dimensional space, and outputs the regression information to the output layer.

[0034] Specifically, the output layer contains several output nodes, and the output value of each node represents the probability of a specific fault type.

[0035] In a preferred embodiment of the present invention, the standard dataset and harmonic feature dataset of real-time operating data of various power equipment are input into a support vector machine model that uniquely corresponds to the fault type to obtain the probability of each fault type. Based on the probability of the fault type, a fault identifier representing the fault type is determined. The specific process is as follows:

[0036] Obtain the probability of each fault type for each power device. Extract the maximum value among the probabilities of each fault type and perform a numerical determination.

[0037] If the maximum value among the probabilities of identifying a fault type is greater than or equal to the maximum preset threshold, then the fault type corresponding to the maximum probability of that fault type for the power equipment is determined. At this time, the fault identifier corresponding to the fault type corresponding to the maximum probability of that fault type for the power equipment is output.

[0038] Send the fault identifiers of each power device to step S400;

[0039] Step S400: Inter-device collaborative self-test;

[0040] The fault identifiers of individual power devices are aggregated to obtain a fault identification matrix for the entire power device group. Based on the fault identification matrix, inter-device collaborative analysis is performed on the fault identifiers of multiple power devices. The operating status of each power device series link is determined by analyzing the data linkage of fault identifiers.

[0041] Obtain the fault identifiers of all power equipment and digitize them into a fault identification matrix. The specific process is as follows:

[0042] If a fault identifier for a power device cannot be identified, the fault identifier element for that power device will be output as 0.

[0043] The fault identifiers of each power device are converted into specific values ​​to obtain fault identification elements.

[0044] The specific value of the fault identification element is equal to the preset fault number corresponding to the fault identifier.

[0045] As a preferred embodiment of the present invention, a fault identification matrix for the entire power equipment group is generated based on the fault identification elements. The matrix contains several rows, each row corresponding to a series chain of equipment. Each element in each row corresponds to a fault identification element of a power equipment in the series chain of equipment.

[0046] Specifically, a device series chain refers to a group of electrical devices connected in a specific order to jointly perform a certain task, exhibiting a certain operating sequence or interdependence. Electrical devices not connected to other electrical devices generate an independent device series chain, with multiple rows corresponding to different device series chains, ultimately forming a fault identification matrix.

[0047] As a preferred embodiment of the present invention, inter-equipment collaborative analysis is performed on the fault identifiers of multiple power devices based on the fault identification matrix. The specific process is as follows:

[0048] Analyze the Pearson correlation coefficients between elements in different columns of the same row in the fault identification matrix to identify the fault linkage between different power devices in the same series chain;

[0049] Analyze the cosine similarity between different rows in the fault identification matrix to identify the fault similarity of different equipment series chains;

[0050] At preset time intervals, power equipment operation data is re-collected and input into step S100. Subsequently, the fault identifiers of each power equipment output in step S300 are obtained, and the fault identification matrix is ​​dynamically updated. The changes in elements in the fault identification matrix at each time interval are analyzed. When it is identified that the value of a certain element in the fault identification matrix of the previous time interval was 0, but the value in the fault identification matrix of the current time is not 0, the element is highlighted.

[0051] Record the device numbers corresponding to all highlighted elements in the fault identification matrix and connect them in series to obtain the fault propagation chain between power equipment.

[0052] Secondly, the present invention provides a power equipment self-testing system based on field sensors, including a data acquisition module, a standardization processing module, a harmonic feature extraction module, a historical data modeling module, a fault identification and judgment module, and a collaborative analysis module.

[0053] The data acquisition module is responsible for accessing the field sensor group and collecting real-time operating data of the power equipment, including input voltage, output voltage, input current, output current, active power, reactive power, apparent power, current frequency, insulation resistance, vibration frequency, and equipment temperature, and then sending it to the standardization processing module.

[0054] The standardization processing module standardizes the collected power equipment operation data, transforming it into a standard dataset, and obtaining the mean, standard deviation, first quartile, third quartile, upper quartile, and lower quartile of the power equipment operation data.

[0055] The harmonic feature extraction module extracts harmonic features from the upper and lower halves of the input and output currents of power equipment to obtain information on the dominant frequency, harmonic amplitude, harmonic energy, and total harmonic distortion.

[0056] The historical data modeling module constructs a power equipment diagnostic model based on a standard dataset and harmonic characteristics. The fault identification probability is then obtained from the power equipment diagnostic model.

[0057] The fault identification and judgment module obtains fault identifiers based on the fault identification probability and determines the fault type of each power equipment.

[0058] The collaborative analysis module generates a fault identification matrix based on fault identifiers. It then performs inter-device collaborative analysis based on this matrix to identify the interconnectedness of faults between devices.

[0059] Thirdly, the present invention provides a power equipment self-test storage medium based on field sensors, including a central computer node and several computer sub-nodes. The number of computer sub-nodes is consistent with the number of power equipment, and each computer sub-node has a power equipment that is uniquely associated with itself.

[0060] Each computer sub-node stores a support vector machine model within its own power equipment diagnostic model, receiving power equipment operation data uniquely associated with it. Upon receiving the power equipment operation data, each sub-node begins data analysis, calculating the standard dataset and harmonic feature dataset, and inputting these into its own power equipment diagnostic model. All sub-nodes perform parallel computation, and then uniformly send the power equipment diagnostic model to the central computer node.

[0061] The central computer node receives the probability of fault types from all computer sub-nodes, performs data analysis, and obtains fault identification elements and fault identification matrices.

[0062] Compared with the prior art, the beneficial effects of the present invention are:

[0063] This invention comprehensively acquires multi-parameter operating data of power equipment, including electrical, mechanical, and thermodynamic parameters, through multi-dimensional data acquisition and standardized processing. Statistical features are used for data standardization, effectively improving data consistency and comparability, laying a reliable foundation for subsequent analysis. Secondly, it innovatively extracts harmonic features from the upper and lower halves of current data, enhancing the perception of nonlinear faults and abnormal states, and accurately capturing potential faults such as harmonic pollution and equipment aging. A support vector machine model is used to construct a power equipment diagnostic model, simultaneously outputting the probabilities of 12 common fault types, achieving parallel identification and high-precision judgment of multiple fault types. Finally, through an inter-equipment collaborative self-inspection mechanism, single-equipment fault identifiers are aggregated into a fault identification matrix. The linkage and similarity of faults within and between equipment chains are analyzed. Combined with dynamic updates and highlighting mechanisms, the fault propagation path and system-level risks are effectively revealed, improving the collaborative monitoring and early warning capabilities for the entire power equipment group's operating status. Attached Figure Description

[0064] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings:

[0065] Figure 1 This is a flowchart of the power equipment self-testing method based on field sensors proposed in the embodiments of the present invention;

[0066] Figure 2 This is a first schematic diagram of the support vector machine model proposed in the embodiments of the present invention;

[0067] Figure 3 This is a second schematic diagram of the support vector machine model proposed in the embodiments of the present invention;

[0068] Figure 4 This is a system block diagram of the power equipment self-testing system based on field sensors proposed in the embodiments of the present invention;

[0069] Figure 5 This is a schematic diagram of a power equipment self-test storage medium based on field sensors proposed in an embodiment of the present invention. Detailed Implementation

[0070] To facilitate a clear understanding of the technical means, creative features, objectives, and effects of this invention, the invention will be further described below in conjunction with specific illustrations.

[0071] like Figure 1 As shown, the self-testing method for power equipment based on field sensors includes the following steps:

[0072] Step S100: Data acquisition and standardization processing;

[0073] Access the sensor array at the power equipment site to acquire power equipment operation data; standardize the collected power equipment operation data and convert it to a standard data format to obtain a standard dataset of operation data for each power equipment; extract the upper and lower whiskers of the input and output currents through statistical feature extraction.

[0074] The collected power equipment operation data includes:

[0075] Input voltage, output voltage, input current, output current, active power, reactive power, apparent power, current frequency, insulation resistance, vibration frequency, and equipment temperature;

[0076] In a preferred embodiment of the present invention, the collected power equipment operation data is standardized and converted into a standard data format. The specific process is as follows:

[0077] Obtain the mean, standard deviation, median, first quartile, third quartile, upper quartile, and lower quartile of the input voltage, output voltage, input current, output current, active power, reactive power, apparent power, current frequency, insulation resistance, vibration frequency, and equipment temperature for each device.

[0078] A standard dataset of power equipment operating data is generated based on the mean, standard deviation, first quartile, and third quartile of the operating data of each device.

[0079] The first quartile refers to the value in the bottom 25% of the dataset after the power equipment operation data is sorted in ascending order. 25% of the power equipment operation data of the same type are lower than the first quartile.

[0080] The third quartile refers to the value in the top 75% of the dataset after the power equipment operation data are sorted in ascending order, and 75% of the power equipment operation data of the same type are lower than the first quartile.

[0081] The term "above" refers to all data points that are greater than or equal to the third quartile plus 1.5 times the interquartile range.

[0082] The lower quartile refers to all data points that are less than or equal to the first quartile minus 1.5 times the interquartile range;

[0083] The interquartile range mentioned above refers to the difference between the third interquartile range and the first interquartile range.

[0084] Generate a standard dataset of operating data for each power device. The standard dataset includes the mean, standard deviation, first quartile, and third quartile of output voltage, input current, output current, active power, reactive power, apparent power, current frequency, insulation resistance, vibration frequency, and device temperature.

[0085] The upper and lower limits of the input and output currents of each power device are sent to step S200.

[0086] Send the standard dataset of operating data of each power device to step S300;

[0087] It should be noted that collecting data from the upper and lower whisker portions separately is to more accurately capture abnormal fluctuations and harmonic characteristics during power equipment operation. The data points contained in the upper and lower whisker portions often fall outside the normal data range, thus representing the performance of power equipment under abnormal or extreme operating conditions. Specifically, the upper whisker portion typically reflects the operating status of power equipment under excessive load or certain sudden events, such as system overload, equipment aging, or sudden current changes, usually accompanied by the generation of high-frequency harmonics. The lower whisker portion, on the other hand, may represent low-frequency fluctuations or anomalies during equipment operation, such as power instability, equipment failure, or data loss, and is usually related to low-frequency harmonics or noise components. By extracting features from the upper and lower whisker portions separately, more comprehensive harmonic information can be captured, thereby more effectively diagnosing potential equipment faults.

[0088] Step S200: Harmonic feature extraction;

[0089] Harmonic features are extracted based on the upper and lower whiskers of the input and output currents to obtain a harmonic feature dataset.

[0090] Fast Fourier Transform (FFT) is performed on the upper and lower whisker portions of the standard datasets of input and output currents for various power devices to convert them from time-domain signals to frequency-domain signals. Frequency-domain features of the input and output currents are then extracted from the frequency-domain signals, including:

[0091] The dominant frequency, harmonic amplitude, harmonic energy, and total harmonic distortion of the upper and lower parts of the input current standard dataset, the upper and lower parts of the output current standard dataset, and the lower part of the output current standard dataset.

[0092] The process of calculating the dominant frequency, harmonic amplitude, harmonic energy, and total harmonic distortion is as follows:

[0093] For a frequency domain signal that has undergone Fast Fourier Transform (FFT), the frequency point with the largest amplitude in the frequency domain signal is identified by the spectrum diagram, and the dominant frequency is obtained.

[0094] For a frequency domain signal subjected to Fast Fourier Transform (FFT), the amplitudes of frequency components that are integer multiples of the dominant frequency are obtained in the frequency domain signal to obtain the harmonic amplitudes. The harmonic amplitudes include the harmonic amplitudes of frequency components three times the dominant frequency, five times the dominant frequency, and seven times the dominant frequency.

[0095] Calculate the square of the amplitude of each harmonic component and integrate it with the power spectral density at the corresponding frequency point to obtain the harmonic energy, including the third, fifth, and seventh times the dominant frequency harmonic energy.

[0096] Calculate the squares of the harmonic amplitudes of the frequency components at three times, five times, and seven times the dominant frequency, and sum them. Take the square root of the sum and divide it by the amplitude corresponding to the dominant frequency to obtain the total harmonic distortion.

[0097] It should be noted that when electrical equipment malfunctions, it generates strong nonlinear harmonic currents. These nonlinear harmonic currents cause the equipment to produce high-order odd-numbered harmonics, with the third, fifth, and seventh harmonics being particularly prominent. Even-numbered harmonics, on the other hand, are typically caused by ideal sinusoidal waveforms and exist only under ideal load conditions. When electrical equipment is running, the amplitude of these even-numbered harmonics is usually small, therefore they offer limited help in diagnosing equipment failures or performance degradation.

[0098] In a preferred embodiment of the present invention, a harmonic feature dataset for each power device is generated, specifically a high-dimensional vector containing 32 elements, wherein the elements are as follows:

[0099] Elements 1-8: The main frequency of the upper part of the input current standard dataset, the harmonic amplitude of the frequency component three times the main frequency, the harmonic amplitude of the frequency component five times the main frequency, the harmonic amplitude of the frequency component seven times the main frequency, the harmonic energy of the third main frequency, the harmonic energy of the fifth main frequency, the harmonic energy of the seventh main frequency, and the total harmonic distortion.

[0100] Elements 9-16: The dominant frequency, harmonic amplitude of the frequency component three times the dominant frequency, harmonic amplitude of the frequency component five times the dominant frequency, harmonic amplitude of the frequency component seven times the dominant frequency, harmonic energy of the third dominant frequency, harmonic energy of the fifth dominant frequency, harmonic energy of the seventh dominant frequency, and total harmonic distortion in the lower part of the input current standard dataset.

[0101] Elements 17-24: The main frequency, harmonic amplitude of the frequency component three times the main frequency, harmonic amplitude of the frequency component five times the main frequency, harmonic amplitude of the frequency component seven times the main frequency, harmonic energy of the third main frequency, harmonic energy of the fifth main frequency, harmonic energy of the seventh main frequency, and total harmonic distortion in the upper part of the standard dataset of output current.

[0102] Elements 25-32: The main frequency, harmonic amplitude of the frequency component three times the main frequency, harmonic amplitude of the frequency component five times the main frequency, harmonic amplitude of the frequency component seven times the main frequency, harmonic energy of the third main frequency, harmonic energy of the fifth main frequency, harmonic energy of the seventh main frequency, and total harmonic distortion in the lower part of the standard dataset of output current.

[0103] The generated harmonic feature datasets of each power device are sent to step S300.

[0104] Step S300: Historical data modeling and fault determination;

[0105] A power equipment diagnostic model is established based on historical data from standard datasets and harmonic feature datasets of various power equipment operation data. By inputting the collected real-time standard datasets and harmonic feature datasets into the power equipment diagnostic model, fault identifiers representing fault types are obtained.

[0106] The power equipment diagnostic model includes several support vector machine models.

[0107] like Figure 2 As shown, the total number of support vector machine (SVM) models is the same as the total number of power devices. There is a one-to-one correspondence between SVMs and power devices; that is, each power device has a unique SVM model corresponding to itself.

[0108] like Figure 3 As shown, each support vector machine model contains an input layer, a kernel function layer, and an output layer.

[0109] Specifically, the input layer has 44+32 nodes.

[0110] The first 44 nodes are responsible for receiving the mean, standard deviation, first quartile, and third quartile of the input voltage, output voltage, input current, output current, active power, reactive power, apparent power, current frequency, insulation resistance, vibration frequency, and equipment temperature, respectively.

[0111] The last 32 nodes are each responsible for receiving 32 elements from the harmonic feature dataset.

[0112] Specifically, the kernel layer has several nodes, each containing a Gaussian radial basis function (RBF) kernel. The RBF kernel's function is to increase the dimensionality of all data received from the original 44+32 nodes of the input layer from 65 dimensions to a higher dimensional space, making data that was inseparable in the lower dimensional space separable in the higher dimensional space. The kernel layer extracts regression information about classification in the higher dimensional space and outputs this regression information to the output layer.

[0113] Specifically, the output layer contains 12 output nodes, and the output value of each node represents the probability of a specific fault type. That is, the 12 nodes of the output layer correspond to 12 different fault conditions.

[0114] The probabilities of the fault types output by the 12 nodes in the output layer include:

[0115] The output of the first node: the probability of an overvoltage fault;

[0116] The output of the second node: the probability of a low voltage fault;

[0117] The output of the third node: the probability of overload failure;

[0118] The output of the 4th node: the probability of an abnormal current fault;

[0119] The output of the 5th node: the probability of an overheating failure;

[0120] The output of the 6th node: the probability of an insulation failure;

[0121] The output of the 7th node: the probability of vibration abnormality fault;

[0122] The output of the 8th node: the probability of an abnormal current frequency fault;

[0123] Output of the 9th node: Probability of reactive power anomaly fault;

[0124] Output of the 10th node: Probability of active power abnormality fault;

[0125] The output of the 11th node: the probability of harmonic contamination fault;

[0126] The output of the 12th node: the probability of equipment aging and failure;

[0127] In a preferred embodiment of the present invention, the standard dataset and harmonic feature dataset of real-time operating data of various power equipment are input into a support vector machine model that uniquely corresponds to the fault type to obtain the probability of each fault type. Based on the probability of the fault type, a fault identifier representing the fault type is determined. The specific process is as follows:

[0128] Obtain the probability of each fault type for each power device. Extract the maximum value among the probabilities of each fault type and perform a numerical determination.

[0129] If the maximum probability of identifying a fault type is greater than or equal to the maximum preset threshold of 90%, then the fault type corresponding to the maximum probability of that fault type for the power equipment is determined. At this point, the fault identifier corresponding to the fault type corresponding to the maximum probability of that fault type for the power equipment is output.

[0130] For example, by inputting the standard dataset and harmonic feature dataset of power equipment 1 into the support vector machine model that uniquely corresponds to it, the probabilities of the following fault types are obtained:

[0131] Probability of overvoltage fault: 0.0605;

[0132] Probability of low voltage fault: 0.1535;

[0133] Probability of overload failure: 0.1182;

[0134] Probability of abnormal current fault: 0.0967;

[0135] Probability of failure due to overheating: 0.0252;

[0136] Probability of insulation failure: 0.0252;

[0137] Probability of abnormal vibration failure: 0.0094;

[0138] Probability of abnormal current frequency fault: 0.1399;

[0139] The probability of reactive power abnormality fault: 0.0971;

[0140] The probability of an abnormal active power failure is 0.1144.

[0141] The probability of harmonic pollution fault: 0.0033;

[0142] The probability of equipment failure due to aging: 0.9566;

[0143] Among them, the maximum probability of each fault type is: the probability of equipment aging fault is 0.9566; compared with the preset threshold of 0.9, since the probability of equipment aging fault is greater than the preset threshold, the fault identifier of power equipment 1 is determined to be: equipment aging fault.

[0144] Send the fault identifiers of each power device to step S400;

[0145] Step S400: Inter-device collaborative self-test;

[0146] The fault identifiers of individual power devices are aggregated to obtain a fault identification matrix for the entire power device group. Based on the fault identification matrix, inter-device collaborative analysis is performed on the fault identifiers of multiple power devices. The operating status of each power device series link is determined by analyzing the data linkage of fault identifiers.

[0147] Obtain the fault identifiers of all power equipment and digitize them into a fault identification matrix. The specific process is as follows:

[0148] If a fault identifier for a power device cannot be identified, the fault identifier element for that power device will be output as 0.

[0149] If a fault identifier for a power device is identified as an overvoltage fault, then the fault identifier element for that power device will be output as 1.

[0150] If a fault identifier for a certain power equipment is identified as a low voltage fault, then the fault identifier element for that power equipment is output as 2.

[0151] If an overload fault is identified as a fault identifier for a certain power equipment, then the fault identification element for that power equipment will be output as 3.

[0152] If a fault identifier for a certain power equipment is identified as an abnormal current fault, then the fault identification element for that power equipment will be output as 4.

[0153] If the fault identifier of a certain power equipment is identified as an overheating fault, then the fault identification element of the power equipment is output as 5.

[0154] If the fault identifier of a certain power equipment is identified as an insulation fault, then the fault identification element of that power equipment is output as 6.

[0155] If the fault identifier of a certain power equipment is identified as an abnormal vibration fault, then the fault identification element of the power equipment is output as 7.

[0156] If a fault identifier for a certain power equipment is identified as an abnormal current frequency fault, then the fault identification element for that power equipment will be output as 8.

[0157] If a fault identifier for a power device is identified as an abnormal reactive power fault, then the fault identification element for that power device will be output as 9.

[0158] If a fault identifier for a power device is identified as an active power abnormality fault, then the fault identification element for that power device will be output as 10.

[0159] If a fault identifier for a power device is identified as a harmonic pollution fault, then the fault identification element for that power device will be output as 11.

[0160] If the fault identifier of a certain power equipment is identified as an aging fault, then the fault identification element of that power equipment is output as 12.

[0161] Generate a fault identification matrix for the entire power equipment group, which contains several rows. Each row corresponds to a series chain of equipment, and each element in each row corresponds to a fault identification element of a power equipment in the series chain.

[0162] Specifically, a device series chain refers to a group of electrical devices connected in a specific order to jointly perform a certain task, exhibiting a certain operating sequence or interdependence. Electrical devices not connected to other electrical devices generate an independent device series chain, with multiple rows corresponding to different device series chains, ultimately forming a two-dimensional fault identification matrix.

[0163] As a preferred embodiment of the present invention, inter-equipment collaborative analysis is performed on the fault identifiers of multiple power devices based on the fault identification matrix. The specific process is as follows:

[0164] Analyze the Pearson correlation coefficients between elements in different columns of the same row in the fault identification matrix to identify the fault linkage between different power devices in the same series chain;

[0165] Analyze the cosine similarity between different rows in the fault identification matrix to identify the fault similarity of different equipment series chains;

[0166] At preset time intervals, power equipment operation data is re-collected and input into step S100. Subsequently, the fault identifiers of each power equipment output in step S300 are obtained, and the fault identification matrix is ​​dynamically updated. The changes in elements in the fault identification matrix at each time interval are analyzed. When it is identified that the value of a certain element in the fault identification matrix of the previous time interval was 0, but the value in the fault identification matrix of the current time is not 0, the element is highlighted.

[0167] Record the device numbers corresponding to all highlighted elements in the fault identification matrix and connect them in series to obtain the fault propagation chain between power equipment.

[0168] like Figure 4 As shown, the power equipment self-inspection system based on field sensors includes a data acquisition module, a standardization processing module, a harmonic feature extraction module, a historical data modeling module, a fault identification and judgment module, and a collaborative analysis module.

[0169] The data acquisition module is responsible for accessing the field sensor group and collecting real-time operating data of the power equipment, including input voltage, output voltage, input current, output current, active power, reactive power, apparent power, current frequency, insulation resistance, vibration frequency, and equipment temperature, and then sending it to the standardization processing module.

[0170] The standardization processing module standardizes the collected power equipment operation data, transforming it into a standard dataset, and obtaining the mean, standard deviation, first quartile, third quartile, upper quartile, and lower quartile of the power equipment operation data.

[0171] The harmonic feature extraction module extracts harmonic features from the upper and lower halves of the input and output currents of power equipment to obtain information on the dominant frequency, harmonic amplitude, harmonic energy, and total harmonic distortion.

[0172] The historical data modeling module constructs a power equipment diagnostic model based on a standard dataset and harmonic characteristics. The fault identification probability is then obtained from the power equipment diagnostic model.

[0173] The fault identification and judgment module obtains fault identifiers based on the fault identification probability and determines the fault type of each power equipment.

[0174] The collaborative analysis module generates a fault identification matrix based on fault identifiers. It then performs inter-device collaborative analysis based on this matrix to identify the interconnectedness of faults between devices.

[0175] like Figure 5 As shown, the power equipment self-test storage medium based on field sensors includes a central computer node and several computer sub-nodes. The number of computer sub-nodes is the same as the number of power equipment, and each computer sub-node has a power equipment that is uniquely associated with itself.

[0176] Each computer sub-node stores a support vector machine model within its own power equipment diagnostic model, receiving power equipment operation data uniquely associated with it. Upon receiving the power equipment operation data, each sub-node begins data analysis, calculating the standard dataset and harmonic feature dataset, and inputting these into its own power equipment diagnostic model. All sub-nodes perform parallel computation, and then uniformly send the power equipment diagnostic model to the central computer node.

[0177] The central computer node receives the probability of fault types from all computer sub-nodes, performs data analysis, and obtains fault identification elements and fault identification matrices.

[0178] It should be understood that the terms “comprising” and “including” used in this disclosure and claims indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0179] It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the disclosure. As used in this disclosure and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this disclosure and claims means any combination and all possible combinations of one or more of the associated listed items, and includes such combinations;

[0180] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for power equipment self-checking based on field sensors, characterized in that, The method comprises the following steps: Step S100, data collection and standardization processing; Accessing a sensor group at a power equipment site to obtain power equipment operation data; performing standardization processing on the collected power equipment operation data to obtain a standard data set of each power equipment operation data; and obtaining upper and lower whiskers of input end current and output end current through statistical feature extraction; Step S200, harmonic feature extraction; Performing harmonic feature extraction according to the upper and lower whiskers of the input end current and the output end current to obtain a harmonic feature data set; Step S300, historical data modeling and fault determination; Establishing a power equipment diagnosis model of historical data based on the standard data set and the harmonic feature data set of each power equipment operation data; and inputting the collected real-time standard data set and harmonic feature data set into the power equipment diagnosis model to obtain a fault identifier representing a fault type; The specific process of obtaining the fault identifier representing the fault type is as follows: Inputting the real-time obtained standard data set and harmonic feature data set of each power equipment operation data into a support vector machine model corresponding to itself to obtain a probability of each fault type, and determining based on the probability of the fault type; If the maximum value in the probability of the fault type is greater than or equal to a maximum preset threshold, it is determined that the power equipment has the fault type corresponding to the maximum probability of the fault type; at this time, the fault identifier corresponding to the fault type corresponding to the maximum probability of the fault type is output. Step S400, inter-device cooperative self-checking; Aggregating the fault identifiers of each power equipment to obtain a fault identification matrix of the entire power equipment group; Performing inter-device cooperative analysis on the fault identifiers of the multiple power equipments based on the fault identification matrix; and determining the running state of a series chain of each power equipment through data linkage analysis of the fault identifiers.

2. The on-site sensor-based power equipment self-checking method according to claim 1, characterized in that, The specific process of performing standardization processing on the collected power equipment operation data is as follows: Obtaining the mean, standard deviation, median, first quartile, third quartile, upper whisker and lower whisker of the operation data of each equipment; and generating a standard data set of the power equipment operation data according to the mean, standard deviation, first quartile and third quartile of the operation data of each equipment.

3. The on-site sensor-based power equipment self-checking method according to claim 1, characterized in that, The specific process of performing harmonic feature extraction is as follows: Performing fast Fourier transform (FFT) on the upper whisker part and the lower whisker part in the standard data set of the input end current and the output end current of each power equipment, respectively, to convert them from time domain signals to frequency domain signals, and extracting frequency domain features of the input end current and the output end current from the frequency domain signals, including: Dominant frequency, harmonic amplitude, harmonic energy and total harmonic distortion of the upper whisker part in the input end current standard data set, the lower whisker part in the input end current standard data set, the upper whisker part in the output end current standard data set and the lower whisker part in the output end current standard data set.

4. The on-site sensor-based power equipment self-checking method according to claim 1, characterized by, The harmonic feature data set is specifically: A high-dimensional vector containing 32 elements, wherein the elements are respectively: The first to eighth elements are the fundamental frequency, the frequency component harmonic amplitude of the triple fundamental frequency, the frequency component harmonic amplitude of the five times fundamental frequency, the frequency component harmonic amplitude of the seven times fundamental frequency, the triple fundamental frequency harmonic energy, the five times fundamental frequency harmonic energy, the seven times fundamental frequency harmonic energy and the total harmonic distortion of the upper whisker part in the input end current standard data set; The ninth to sixteenth elements are the fundamental frequency, the frequency component harmonic amplitude of the triple fundamental frequency, the frequency component harmonic amplitude of the five times fundamental frequency, the frequency component harmonic amplitude of the seven times fundamental frequency, the triple fundamental frequency harmonic energy, the five times fundamental frequency harmonic energy, the seven times fundamental frequency harmonic energy and the total harmonic distortion of the lower whisker part in the input end current standard data set; The seventeenth to twenty-fourth elements are the fundamental frequency, the frequency component harmonic amplitude of the triple fundamental frequency, the frequency component harmonic amplitude of the five times fundamental frequency, the frequency component harmonic amplitude of the seven times fundamental frequency, the triple fundamental frequency harmonic energy, the five times fundamental frequency harmonic energy, the seven times fundamental frequency harmonic energy and the total harmonic distortion of the upper whisker part in the output end current standard data set; The twenty-fifth to thirty-second elements are the fundamental frequency, the frequency component harmonic amplitude of the triple fundamental frequency, the frequency component harmonic amplitude of the five times fundamental frequency, the frequency component harmonic amplitude of the seven times fundamental frequency, the triple fundamental frequency harmonic energy, the five times fundamental frequency harmonic energy, the seven times fundamental frequency harmonic energy and the total harmonic distortion of the lower whisker part in the output end current standard data set.

5. The on-site sensor-based power equipment self-checking method according to claim 1, characterized by, The power equipment diagnosis model specifically is: The power equipment diagnosis model comprises a plurality of support vector machine models; The total number of the support vector machine models is consistent with the total number of the power equipment; the support vector machine model and the power equipment have a one-to-one correspondence; that is, each power equipment has a support vector machine model corresponding thereto; Each support vector machine model comprises an input layer, a kernel function layer and an output layer; Some nodes of the input layer are responsible for receiving the elements in the standard data set and the harmonic feature data set; The kernel function layer has a plurality of nodes, each of which comprises a Gaussian radial basis kernel; the Gaussian radial basis kernel raises all the data received by each node of the input layer to a higher dimensional space, extracts regression information about classification in the high dimensional space, and outputs the regression information to the output layer; The output layer comprises a plurality of output nodes, and the output value of each node represents the probability of a specific fault type.

6. The on-site sensor-based power equipment self-checking method according to claim 1, characterized by, The specific process of obtaining the fault identification matrix of the entire power equipment group is: Obtain the fault identification symbols of all the power equipment, convert the fault identification symbols of each power equipment into specific numerical values to obtain fault identification elements; The specific numerical value of the fault identification element is equal to the preset fault number corresponding to the fault identification symbol; If the fault identification symbol of a power equipment cannot be identified, the fault identification element of the power equipment is output as 0; Generate the fault identification matrix of the entire power equipment group according to the fault identification elements, which comprises a plurality of rows, each row corresponds to a device series chain, and each element in each row corresponds to the fault identification element of a power equipment in the device series chain; Perform inter-device collaborative analysis on the fault identification symbols of the plurality of power equipment based on the fault identification matrix.

7. The on-site sensor-based power equipment self-checking method according to claim 6, characterized by, The specific process of performing inter-device collaborative analysis is: Pearson correlation coefficient between elements in the same row and different columns in the fault identification matrix is analyzed to identify the fault linkage between different power devices in the same device series chain; Cosine similarity between different rows in the fault identification matrix is analyzed to identify the fault similarity of different device series chains; Every preset time interval, the power device operation data is re-collected and input into step S100, and then the fault identification symbol of each power device output by step S300 is obtained, and the fault identification matrix is dynamically updated; the element change in the fault identification matrix of each time interval is analyzed, and when it is identified that the value of a certain element is 0 in the fault identification matrix of the previous time interval, but the value in the fault identification matrix at the current time is not 0, the element is highlighted; The device numbers corresponding to all highlighted elements in the fault identification matrix are recorded and connected in series to obtain the fault conduction chain between the power devices.

8. A field sensor based power equipment self-checking system for implementing the field sensor based power equipment self-checking method of any one of claims 1-7, characterized by, The system comprises a data collection module, a standardization processing module, a harmonic feature extraction module, a historical data modeling module, a fault identification and determination module, and a collaborative analysis module. The data collection module is responsible for accessing the on-site sensor group, collecting the operation data of the power device in real time, including the input end voltage, the output end voltage, the input end current, the output end current, the active power, the reactive power, the apparent power, the current frequency, the insulation resistance, the vibration frequency, and the device temperature, and sending them to the standardization processing module. The standardization processing module performs standardization processing on the collected power device operation data, converts it into a standard data set, and obtains the mean, standard deviation, first quartile, third quartile, upper whisker, and lower whisker of the power device operation data. The harmonic feature extraction module extracts the harmonic features of the upper whisker and lower whisker parts of the input end current and output end current of the power device, and obtains the main frequency, harmonic amplitude, harmonic energy, and total harmonic distortion information. The historical data modeling module constructs a power device diagnosis model according to the standard data set and the harmonic features; and obtains a fault identification probability according to the power device diagnosis model. The fault identification and determination module obtains a fault identification symbol according to the fault identification probability, and determines the fault type of each power device. The collaborative analysis module obtains a fault identification matrix according to the fault identification symbol; and performs inter-device collaborative analysis according to the fault identification matrix to identify the fault linkage between devices.

9. A storage medium for a field sensor based power equipment self-checking for implementing the field sensor based power equipment self-checking method of any one of claims 1 to 7, characterized in that ； The system comprises one central computer node and a plurality of computer sub-nodes, the number of computer sub-nodes is consistent with the number of power devices, and each computer sub-node has one power device uniquely corresponding to itself; Each computer sub-node saves a support vector machine model in the power device diagnosis model, and receives the power device operation data of the power device uniquely corresponding to itself; each computer sub-node starts data analysis after receiving the power device operation data, calculates the standard data set and the harmonic feature data set, and inputs them into the power device diagnosis model; all computer sub-nodes are calculated in parallel, and then the power device diagnosis model is uniformly sent to the central computer node; The central computer node receives the probabilities of failure types from all computer sub-nodes, and performs data analysis to obtain a failure identification element and a failure identification matrix.

Images