Partial Discharge Monitoring and Location System and Method Based on Internet of Things Architecture

Abstract

The application discloses a partial discharge monitoring and positioning system and method based on an internet of things architecture, relates to the technical field of power monitoring, collects ideal operation parameters of power equipment, establishes an optimal operation contrast array, selects a plurality of ultrasonic wave emitting points and ultrasonic wave collecting points based on the power equipment to be detected, acquires a plurality of power equipment sound signal arrays, analyzes sound signals in the plurality of power equipment sound signal arrays based on the plurality of power equipment sound signal arrays, determines characteristic data in the sound signals, establishes a power equipment characteristic matrix, evaluates abnormal characteristics in the power equipment characteristic matrix based on the power equipment characteristic matrix and the optimal operation contrast array, determines an abnormal state of the power equipment, and marks an abnormal position of the abnormal state power equipment based on the abnormal state of the power equipment. The application has the advantages of accurately identifying abnormal discharge of the power equipment, reducing maintenance cost, and improving the stability of the power equipment.

Description

Technical Field

[0001] This invention relates to the field of power monitoring technology, specifically to a partial discharge monitoring and positioning system and method based on an Internet of Things (IoT) architecture. Background Technology

[0002] Partial discharge can damage the lifespan of line insulation materials. Due to partial discharge corrosion, it accelerates insulation aging and deterioration. By detecting partial discharge, these potential problems can be identified and addressed in a timely manner, preventing serious faults and avoiding large-scale power outages and economic losses.

[0003] Existing partial discharge detection technologies lack feature recognition and extraction for multi-point detection, making it difficult to accurately monitor and locate the partial discharge position of power equipment during multi-point detection in practical applications. Summary of the Invention

[0004] To address the aforementioned technical problems, this paper provides a partial discharge monitoring and positioning system and method based on an Internet of Things (IoT) architecture. This technical solution solves the problem that existing partial discharge detection technologies lack feature recognition and extraction for multi-point detection, making it difficult to accurately monitor and locate the partial discharge position of power equipment during multi-point detection in practical applications.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A partial discharge monitoring and location method based on an Internet of Things (IoT) architecture includes:

[0007] Collect ideal operating parameters for all power devices connected to the IoT architecture and establish an optimal operating comparison array;

[0008] Based on the power equipment to be tested, several ultrasonic emission points and ultrasonic collection points are selected to obtain several arrays of acoustic signals emitted by the power equipment.

[0009] Based on several arrays of acoustic signals emitted by power equipment, the acoustic signals in several arrays of acoustic signals emitted by power equipment are analyzed to determine the characteristic data in the acoustic signals and establish a characteristic matrix of power equipment.

[0010] Based on the power equipment feature matrix and the optimal operation comparison array, the abnormal features in the power equipment feature matrix are evaluated to determine the abnormal state of the power equipment.

[0011] Based on the abnormal state of power equipment, the abnormal discharge location of the abnormal power equipment is marked.

[0012] Preferably, the step of selecting several ultrasonic emission points and ultrasonic collection points based on the power equipment to be tested, and obtaining several arrays of acoustic signals emitted by the power equipment specifically includes:

[0013] The power equipment to be tested is marked, and several ultrasonic emission points and ultrasonic collection points are selected for the power equipment in the order of marking to obtain ultrasonic signal data for each power equipment.

[0014] Based on the ultrasonic signal data of each power device, the ultrasonic signal data is amplified.

[0015] Based on the amplified ultrasonic signal data, the continuous ultrasonic signal data is discretized into a series of pulses in the form of analog signals, and the value of the analog signal is represented by encoding the amplitude and time of the pulses, thereby realizing the conversion of ultrasonic signals into digital signals;

[0016] The ultrasonic signal of each power device is converted into a digital signal, and the digital signals are mapped according to the order of the power devices to establish several arrays of acoustic signals emitted by the power devices.

[0017] Preferably, the step of analyzing the acoustic signals in several groups of power equipment acoustic signal arrays, determining the characteristic data in the acoustic signals, and establishing a power equipment feature matrix specifically includes:

[0018] Based on several sets of acoustic signal arrays from power equipment, the acoustic data in the arrays are quantized to obtain a quantized array of acoustic signals from power equipment.

[0019] Based on the quantization array of the acoustic signal emitted by the power equipment, a time-domain diagram is drawn, and the start point, peak, trough and end point of the waveform in the time-domain diagram are marked to determine the time-domain diagram of the power equipment.

[0020] Based on the time-domain graph of power equipment, the maximum amplitude value characteristic in the time-domain graph of power equipment is calculated using the maximum amplitude formula;

[0021] Based on the time-domain graph of power equipment, the root mean square amplitude value characteristics in the time-domain graph of power equipment are calculated using the root mean square amplitude formula.

[0022] Based on the time-domain graph of power equipment, the waveform kurtosis characteristics in the time-domain graph of power equipment are calculated using the waveform kurtosis formula;

[0023] Based on the time-domain diagram of power equipment, the waveform skewness characteristics in the time-domain diagram of power equipment are calculated using the waveform skewness formula.

[0024] Based on the maximum amplitude value characteristics, root mean square amplitude value characteristics, waveform kurtosis characteristics, and waveform skewness characteristics in the time-domain graph of power equipment, a feature matrix G is constructed, G = [g 11 …g ij …g nm ], where G is the characteristic matrix of power equipment, g ij Let be the j-th feature data of the i-th sampling point, n be the total number of sampling points, and m be the total number of features.

[0025] Preferably, the formula for the maximum amplitude is:

[0026] A max =max(|x(t)|)

[0027] In the formula, A max The maximum amplitude value characteristic in the time domain graph of the power equipment is denoted by x, where x is the amplitude value and t is the time.

[0028] The root mean square amplitude formula is:

[0029]

[0030] In the formula, B rms Let represent the root mean square amplitude value characteristic of the power equipment in the time domain graph, where n is the total number of sampling points in the time domain graph, and i is the i-th sampling point;

[0031] The waveform kurtosis formula is as follows:

[0032]

[0033] In the formula, K represents the waveform kurtosis characteristic in the time-domain graph of the power equipment, and x i Let i be the amplitude value of the i-th sampling point. This represents the average amplitude.

[0034] The waveform skewness formula is as follows:

[0035]

[0036] In the formula, S represents the waveform skewness characteristic in the time-domain graph of the power equipment, and V represents the standard deviation in the time-domain graph of the power equipment.

[0037] Preferably, the step of evaluating abnormal features in the power equipment feature matrix and determining the abnormal state of the power equipment based on the power equipment feature matrix and the optimal operation comparison array specifically includes:

[0038] Based on the power equipment feature matrix, several feature subsets of power equipment are selected according to the labeling order;

[0039] Based on the feature subsets of several power equipment, the correlation score within each feature subset of power equipment is calculated using the correlation coefficient formula, thereby obtaining an array of correlation scores within several feature subsets of power equipment.

[0040] The correlation score arrays within several power equipment feature subsets and the optimal operation control array are standardized to obtain a standardized power equipment feature subset array and a standardized optimal operation control array.

[0041] The standardized operating parameters of the standardized optimal operating control array are used as the optimal operating threshold.

[0042] Based on the standardized power equipment feature subset array, the maximum feature score is used as the segmentation threshold to divide the standardized power equipment feature subset array into two leaf sets, resulting in a left leaf tree and a right leaf tree;

[0043] Calculate the path length of each cotyledon in the left and right cotyledon trees respectively, and calculate the average path length of each cotyledon based on the path length;

[0044] Based on the path length of each sub-leaf and the average path length of each sub-leaf, the anomaly score is calculated using an exponential function formula.

[0045] The abnormal score is compared with the optimal operating threshold to determine whether the abnormal score exceeds the optimal operating threshold. If it does, it is determined to be an abnormal operation; otherwise, no output is made.

[0046] Preferably, the correlation coefficient formula is:

[0047] C=|ρ(g ij ,M)|

[0048] In the formula, C is the relevance score within each subset of power equipment features, and ρ is the density symbol;

[0049] Specifically, the path length of each coleoptile in the left and right coleoptile trees is calculated separately, and the average path length of each coleoptile is calculated based on the path lengths.

[0050]

[0051] In the formula, Let be the average length of the r-th leaf path, k be the leaf, h be the processing function, r be the leaf path, and Y be the total number of leaves in the leaf tree.

[0052] The formula for the exponential function is:

[0053]

[0054] In the formula, C 异This is an abnormal score.

[0055] Furthermore, a partial discharge monitoring and location system based on an Internet of Things (IoT) architecture is proposed to implement the aforementioned partial discharge monitoring and location method based on an IoT architecture, including:

[0056] The standard parameter acquisition module is used to collect ideal operating parameters of power equipment and establish an optimal operating comparison array.

[0057] The acoustic emission data acquisition module is used to select several ultrasonic emission points and ultrasonic collection points based on the electrical equipment under test, and acquire several arrays of acoustic emission signals from the electrical equipment.

[0058] The acoustic emission data analysis module is electrically connected to the acoustic emission data acquisition module. The acoustic emission data analysis module is used to analyze the acoustic emission signals in several sets of acoustic emission signal arrays of power equipment based on several sets of acoustic emission signal arrays of power equipment, determine the characteristic data in the acoustic emission signals, and establish the characteristic matrix of power equipment.

[0059] The anomaly assessment module is electrically connected to the acoustic emission data analysis module. The anomaly assessment module is used to assess the abnormal features in the power equipment feature matrix based on the power equipment feature matrix and the optimal operation comparison array, and to determine the abnormal state of the power equipment.

[0060] The marking module is electrically connected to the anomaly assessment module. The marking module is used to mark the abnormal discharge location of the power equipment based on the abnormal state of the power equipment.

[0061] Optionally, the acoustic data acquisition module includes:

[0062] The acquisition unit marks the power equipment to be tested, selects several ultrasonic emission points and ultrasonic collection points for the power equipment in the order of the marking, and acquires ultrasonic signal data for each power equipment.

[0063] The amplification processing unit amplifies the ultrasonic signal data of each power device.

[0064] The signal conversion unit discretizes the continuous ultrasonic signal data into a series of pulses in the form of analog signals based on the amplified ultrasonic signal data, and encodes the value of the analog signal by the amplitude and time of the pulses, thereby realizing the conversion of ultrasonic signals into digital signals.

[0065] The acoustic signal array unit converts the ultrasonic signal of each power device into a digital signal, maps the digital signals according to the order of the power devices, and establishes several acoustic signal arrays for power devices.

[0066] Optionally, the acoustic emission data analysis module includes:

[0067] The quantization unit, based on several sets of acoustic signal arrays of power equipment, performs quantization processing on the acoustic data in the arrays to obtain a quantized array of acoustic signals of power equipment.

[0068] The time-domain plotting unit, based on the quantization array of the acoustic signal emitted by the power equipment, plots the time-domain plot, marking the start point, peak, trough, and end point of the waveform in the time-domain plot to determine the time-domain plot of the power equipment;

[0069] The maximum amplitude characteristic unit is based on the time-domain diagram of power equipment and uses the maximum amplitude formula to calculate the maximum amplitude value characteristic in the time-domain diagram of power equipment.

[0070] The root mean square amplitude characteristic element is based on the time-domain diagram of power equipment and uses the root mean square amplitude formula to calculate the root mean square amplitude value characteristics in the time-domain diagram of power equipment.

[0071] The waveform kurtosis feature unit calculates the waveform kurtosis feature in the time domain diagram of power equipment using the waveform kurtosis formula, based on the time domain diagram of power equipment.

[0072] The waveform skewness feature unit calculates the waveform skewness features in the time domain diagram of power equipment using the waveform skewness formula.

[0073] The feature matrix unit constructs the power equipment feature matrix G based on the maximum amplitude value feature, the root mean square amplitude value feature, the waveform kurtosis feature, and the waveform skewness feature in the power equipment time domain graph.

[0074] Optionally, the anomaly assessment module includes:

[0075] The filtering unit, based on the power equipment feature matrix, filters out a subset of features of several power equipment according to the label order;

[0076] The correlation calculation unit, based on the feature subsets of several power equipment, uses the correlation coefficient formula to calculate the correlation score within each feature subset of power equipment, thereby obtaining an array of correlation scores within several feature subsets of power equipment.

[0077] The standardization unit standardizes the correlation score array within several power equipment feature subsets and the optimal operation reference array to obtain a standardized power equipment feature subset array and a standardized optimal operation reference array.

[0078] The threshold unit uses the standardized operating parameters of the standardized optimal operating control array as the optimal operating threshold.

[0079] The segmentation unit, based on the standardized power equipment feature subset array, uses the maximum feature score as the segmentation threshold to divide the standardized power equipment feature subset array into two leaf sets, obtaining the left leaf tree and the right leaf tree;

[0080] The path length calculation unit calculates the path length of each cotyledon in the left and right cotyledon trees respectively, and calculates the average path length of each cotyledon based on the path length.

[0081] The anomaly score calculation unit calculates the anomaly score based on the path length of each sub-leaf and the average path length of each sub-leaf, using an exponential function formula.

[0082] The anomaly detection unit compares the anomaly score with the optimal operating threshold to determine whether the anomaly score exceeds the optimal operating threshold. If it does, the operation is considered abnormal; otherwise, no output is made.

[0083] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0084] This invention proposes a partial discharge monitoring and positioning scheme based on an Internet of Things (IoT) architecture. By collecting and analyzing ultrasonic signals from ultrasonic detection devices set up at multiple locations, the operating status of power equipment is evaluated, the location of abnormal discharge in power equipment is accurately identified, maintenance costs are reduced, and the stability of power equipment is improved. Attached Figure Description

[0085] Figure 1 This is a flowchart of the partial discharge monitoring and location method based on the Internet of Things architecture proposed in this solution;

[0086] Figure 2 This is a flowchart of the method for obtaining the acoustic signal array of power equipment in this scheme;

[0087] Figure 3 This is a flowchart illustrating the method for establishing the feature matrix of power equipment in this scheme;

[0088] Figure 4 This is a flowchart of the method for determining abnormal states of power equipment in this scheme. Detailed Implementation

[0089] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0090] Reference Figure 1 As shown, a partial discharge monitoring and location method based on an Internet of Things (IoT) architecture includes:

[0091] Collect ideal operating parameters for all power devices connected to the IoT architecture and establish an optimal operating comparison array;

[0092] Based on the power equipment to be tested, several ultrasonic emission points and ultrasonic collection points are selected to obtain several arrays of acoustic signals emitted by the power equipment.

[0093] Based on several arrays of acoustic signals emitted by power equipment, the acoustic signals in several arrays of acoustic signals emitted by power equipment are analyzed to determine the characteristic data in the acoustic signals and establish a characteristic matrix of power equipment.

[0094] Based on the power equipment feature matrix and the optimal operation comparison array, the abnormal features in the power equipment feature matrix are evaluated to determine the abnormal state of the power equipment.

[0095] Based on the abnormal state of power equipment, the abnormal discharge location of the abnormal power equipment is marked.

[0096] This scheme collects ideal operating parameters of power equipment to form an optimal operating control array. Subsequently, several ultrasonic emission and collection points were selected on the power equipment to be tested, and corresponding acoustic signal arrays were obtained. By analyzing these acoustic signal arrays, key acoustic signal features were extracted, and a feature matrix of the power equipment was established. Next, this feature matrix was compared with the optimal operating control array to evaluate abnormal features, thereby determining whether the power equipment was in an abnormal state. Finally, by screening out power equipment in abnormal states and marking the abnormal discharge locations, accurate location of abnormal states was achieved. In this way, rapid monitoring and abnormal state identification of power equipment are realized, thereby improving the safety and reliability of power equipment operation.

[0097] Reference Figure 2 As shown, the step of selecting several ultrasonic emission points and ultrasonic collection points based on the power equipment under test to obtain several arrays of acoustic signals from the power equipment specifically includes:

[0098] The power equipment to be tested is marked, and several ultrasonic emission points and ultrasonic collection points are selected for the power equipment in the order of marking to obtain ultrasonic signal data for each power equipment.

[0099] Based on the ultrasonic signal data of each power device, the ultrasonic signal data is amplified.

[0100] Based on the amplified ultrasonic signal data, the continuous ultrasonic signal data is discretized into a series of pulses in the form of analog signals, and the value of the analog signal is represented by encoding the amplitude and time of the pulses, thereby realizing the conversion of ultrasonic signals into digital signals;

[0101] The ultrasonic signal of each power device is converted into a digital signal, and the digital signals are mapped according to the order of the power devices to establish several arrays of acoustic signals emitted by the power devices.

[0102] Understandably, due to the complexity of electrical equipment, ultrasonic signals are susceptible to environmental interference when using ultrasonic testing. Therefore, when using signal amplifiers, multiple amplifiers should be used to amplify the ultrasonic signals step by step according to their different characteristics.

[0103] It is understandable that when performing digital signal conversion on ultrasonic signals, the collected ultrasonic signals should be filtered to reduce signal distortion and insufficient sampling during conversion.

[0104] Reference Figure 3 As shown, the analysis of acoustic signals in several sets of power equipment acoustic signal arrays to determine the characteristic data in the acoustic signals and establish a power equipment feature matrix specifically includes:

[0105] Based on several sets of acoustic signal arrays from power equipment, the acoustic data in the arrays are quantized to obtain a quantized array of acoustic signals from power equipment.

[0106] Based on the quantization array of the acoustic signal emitted by the power equipment, a time-domain diagram is drawn, and the start point, peak, trough and end point of the waveform in the time-domain diagram are marked to determine the time-domain diagram of the power equipment.

[0107] Based on the time-domain graph of power equipment, the maximum amplitude value characteristic in the time-domain graph of power equipment is calculated using the maximum amplitude formula;

[0108] Based on the time-domain graph of power equipment, the root mean square amplitude value characteristics in the time-domain graph of power equipment are calculated using the root mean square amplitude formula.

[0109] Based on the time-domain graph of power equipment, the waveform kurtosis characteristics in the time-domain graph of power equipment are calculated using the waveform kurtosis formula;

[0110] Based on the time-domain diagram of power equipment, the waveform skewness characteristics in the time-domain diagram of power equipment are calculated using the waveform skewness formula.

[0111] Based on the maximum amplitude value characteristics, root mean square amplitude value characteristics, waveform kurtosis characteristics, and waveform skewness characteristics in the time-domain graph of power equipment, a feature matrix G is constructed, G = [g 11 …g ij …g nm ], where G is the characteristic matrix of power equipment, g ij Let be the j-th feature data of the i-th sampling point, n be the total number of sampling points, and m be the total number of features.

[0112] The formula for the maximum amplitude is:

[0113] A max =max(|x(t)|)

[0114] In the formula, A max The maximum amplitude value characteristic in the time domain graph of the power equipment is denoted by x, where x is the amplitude value and t is the time.

[0115] The root mean square amplitude formula is:

[0116]

[0117] In the formula, B rms Let represent the root mean square amplitude value characteristic of the power equipment in the time domain graph, where n is the total number of sampling points in the time domain graph, and i is the i-th sampling point;

[0118] The waveform kurtosis formula is as follows:

[0119]

[0120] In the formula, K represents the waveform kurtosis characteristic in the time-domain graph of the power equipment, and x i Let i be the amplitude value of the i-th sampling point. This represents the average amplitude.

[0121] The waveform skewness formula is as follows:

[0122]

[0123] In the formula, S represents the waveform skewness characteristic in the time-domain graph of the power equipment, and V represents the standard deviation in the time-domain graph of the power equipment.

[0124] This scheme first quantizes the acoustic signal emitted by the power equipment, converting it into a numerical data array. Then, by plotting a time-domain graph, the start point, peaks, troughs, and end points of the waveform are marked, forming a time-domain image of the power equipment. Next, using the maximum amplitude formula, root mean square amplitude formula, waveform kurtosis formula, and waveform skewness formula, the maximum amplitude value, root mean square amplitude value, waveform kurtosis, and waveform skewness in the time-domain graph are calculated, respectively. Finally, these features are combined into a feature matrix, which contains the key features of the time-domain graph of the power equipment's acoustic signal, providing a foundation for subsequent analysis and processing.

[0125] Reference Figure 4 As shown, the process of evaluating abnormal features in the power equipment feature matrix and determining abnormal power equipment states based on the power equipment feature matrix and the optimal operation comparison array specifically includes:

[0126] Based on the power equipment feature matrix, several feature subsets of power equipment are selected according to the labeling order;

[0127] Based on the feature subsets of several power equipment, the correlation score within each feature subset of power equipment is calculated using the correlation coefficient formula, thereby obtaining an array of correlation scores within several feature subsets of power equipment.

[0128] The correlation score arrays within several power equipment feature subsets and the optimal operation control array are standardized to obtain a standardized power equipment feature subset array and a standardized optimal operation control array.

[0129] The standardized operating parameters of the standardized optimal operating control array are used as the optimal operating threshold.

[0130] Based on the standardized power equipment feature subset array, the maximum feature score is used as the segmentation threshold to divide the standardized power equipment feature subset array into two leaf sets, resulting in a left leaf tree and a right leaf tree;

[0131] Calculate the path length of each cotyledon in the left and right cotyledon trees respectively, and calculate the average path length of each cotyledon based on the path length;

[0132] Based on the path length of each sub-leaf and the average path length of each sub-leaf, the anomaly score is calculated using an exponential function formula.

[0133] The abnormal score is compared with the optimal operating threshold to determine whether the abnormal score exceeds the optimal operating threshold. If it does, it is determined to be an abnormal operation; otherwise, no output is made.

[0134] The formula for the correlation coefficient is:

[0135] C=|ρ(g ij ,M)|

[0136] In the formula, C is the relevance score within each subset of power equipment features, and ρ is the density symbol;

[0137] Specifically, the path length of each coleoptile in the left and right coleoptile trees is calculated separately, and the average path length of each coleoptile is calculated based on the path lengths.

[0138]

[0139] In the formula, Let be the average length of the r-th leaf path, k be the leaf, h be the processing function, r be the leaf path, and Y be the total number of leaves in the leaf tree.

[0140] The formula for the exponential function is:

[0141]

[0142] In the formula, C 异 This is an abnormal score.

[0143] This scheme is an anomaly detection method designed for power equipment systems. First, based on the power equipment feature matrix, a feature subset is selected, and their correlation scores are calculated. Then, the correlation score array is standardized against an optimal operating control array, and the standardized parameters of the optimal operating control array are used as a threshold. Next, the standardized feature subset array is divided into left and right leaf trees based on the maximum feature score, and the path length and average path length of each leaf are calculated. Anomaly scores are calculated using an exponential function formula and compared with the optimal operating threshold to determine whether abnormal operation exists. This scheme combines multiple indicators such as feature correlation, path length, and anomaly scores, effectively monitoring anomalies in power equipment systems.

[0144] Furthermore, based on the same inventive concept as the aforementioned partial discharge monitoring and location method based on an Internet of Things (IoT) architecture, a partial discharge monitoring and location system based on an IoT architecture is proposed, comprising:

[0145] The standard parameter acquisition module is used to collect ideal operating parameters of power equipment and establish an optimal operating comparison array.

[0146] The acoustic emission data acquisition module is used to select several ultrasonic emission points and ultrasonic collection points based on the electrical equipment under test, and acquire several arrays of acoustic emission signals from the electrical equipment.

[0147] The acoustic emission data analysis module is electrically connected to the acoustic emission data acquisition module. The acoustic emission data analysis module is used to analyze the acoustic emission signals in several sets of acoustic emission signal arrays of power equipment based on several sets of acoustic emission signal arrays of power equipment, determine the characteristic data in the acoustic emission signals, and establish the characteristic matrix of power equipment.

[0148] The anomaly assessment module is electrically connected to the acoustic emission data analysis module. The anomaly assessment module is used to assess the abnormal features in the power equipment feature matrix based on the power equipment feature matrix and the optimal operation comparison array, and to determine the abnormal state of the power equipment.

[0149] The marking module is electrically connected to the anomaly assessment module. The marking module is used to mark the abnormal locations of power equipment based on the abnormal state of the power equipment.

[0150] The acoustic data acquisition module includes:

[0151] The acquisition unit marks the power equipment to be tested, selects several ultrasonic emission points and ultrasonic collection points for the power equipment in the order of the marking, and acquires ultrasonic signal data for each power equipment.

[0152] The amplification processing unit amplifies the ultrasonic signal data of each power device.

[0153] The signal conversion unit discretizes the continuous ultrasonic signal data into a series of pulses in the form of analog signals based on the amplified ultrasonic signal data, and encodes the value of the analog signal by the amplitude and time of the pulses, thereby realizing the conversion of ultrasonic signals into digital signals.

[0154] The acoustic signal array unit converts the ultrasonic signal of each power device into a digital signal, maps the digital signals according to the order of the power devices, and establishes several acoustic signal arrays for power devices.

[0155] The acoustic data analysis module includes:

[0156] The quantization unit, based on several sets of acoustic signal arrays of power equipment, performs quantization processing on the acoustic data in the arrays to obtain a quantized array of acoustic signals of power equipment.

[0157] The time-domain plotting unit, based on the quantization array of the acoustic signal emitted by the power equipment, plots the time-domain plot, marking the start point, peak, trough, and end point of the waveform in the time-domain plot to determine the time-domain plot of the power equipment;

[0158] The maximum amplitude characteristic unit is based on the time-domain diagram of power equipment and uses the maximum amplitude formula to calculate the maximum amplitude value characteristic in the time-domain diagram of power equipment.

[0159] The root mean square amplitude characteristic element is based on the time-domain diagram of power equipment and uses the root mean square amplitude formula to calculate the root mean square amplitude value characteristics in the time-domain diagram of power equipment.

[0160] The waveform kurtosis feature unit calculates the waveform kurtosis feature in the time domain diagram of power equipment using the waveform kurtosis formula, based on the time domain diagram of power equipment.

[0161] The waveform skewness feature unit calculates the waveform skewness features in the time domain diagram of power equipment using the waveform skewness formula.

[0162] The feature matrix unit constructs the power equipment feature matrix G based on the maximum amplitude value feature, the root mean square amplitude value feature, the waveform kurtosis feature, and the waveform skewness feature in the power equipment time domain graph.

[0163] The anomaly assessment module includes:

[0164] The filtering unit, based on the power equipment feature matrix, filters out a subset of features of several power equipment according to the label order;

[0165] The correlation calculation unit, based on the feature subsets of several power equipment, uses the correlation coefficient formula to calculate the correlation score within each feature subset of power equipment, thereby obtaining an array of correlation scores within several feature subsets of power equipment.

[0166] The standardization unit standardizes the correlation score array within several power equipment feature subsets and the optimal operation reference array to obtain a standardized power equipment feature subset array and a standardized optimal operation reference array.

[0167] The threshold unit uses the standardized operating parameters of the standardized optimal operating control array as the optimal operating threshold.

[0168] The segmentation unit, based on the standardized power equipment feature subset array, uses the maximum feature score as the segmentation threshold to divide the standardized power equipment feature subset array into two leaf sets, obtaining the left leaf tree and the right leaf tree;

[0169] The path length calculation unit calculates the path length of each cotyledon in the left and right cotyledon trees respectively, and calculates the average path length of each cotyledon based on the path length.

[0170] The anomaly score calculation unit calculates the anomaly score based on the path length of each sub-leaf and the average path length of each sub-leaf, using an exponential function formula.

[0171] The anomaly detection unit compares the anomaly score with the optimal operating threshold to determine whether the anomaly score exceeds the optimal operating threshold. If it does, the operation is considered abnormal; otherwise, no output is made.

[0172] The usage process of the above-mentioned partial discharge monitoring and positioning system based on the Internet of Things architecture is as follows:

[0173] Step 1: Collect ideal operating parameters of power equipment and establish an optimal operating control array;

[0174] Step 2: Mark the electrical equipment to be tested, select several ultrasonic emission points and ultrasonic collection points for the electrical equipment in the marked order, and obtain ultrasonic signal data for each electrical equipment.

[0175] Step 3: Amplify the ultrasonic signal data of each power device.

[0176] Step 4: Based on the amplified ultrasonic signal data, the continuous ultrasonic signal data is discretized into a series of pulses in the form of analog signals, and the value of the analog signal is represented by the amplitude and time of the pulses, thereby realizing the conversion of ultrasonic signals into digital signals;

[0177] Step 5: Convert the ultrasonic signal of each power device into a digital signal, map the digital signals according to the order of the power devices, and establish several arrays of acoustic signals of power devices;

[0178] Step 6: Based on several sets of acoustic signal arrays from power equipment, quantize the acoustic data in the arrays to obtain quantized arrays of acoustic signals from power equipment.

[0179] Step 7: Based on the quantization array of the acoustic signal emitted by the power equipment, draw a time-domain diagram, mark the start point, peak, trough and end point of the waveform in the time-domain diagram, and determine the time-domain diagram of the power equipment;

[0180] Step 8: Based on the time-domain graph of the power equipment, use the maximum amplitude formula to calculate the maximum amplitude value characteristic in the time-domain graph of the power equipment;

[0181] Step 9: Based on the time-domain diagram of the power equipment, use the root mean square amplitude formula to calculate the root mean square amplitude characteristics in the time-domain diagram of the power equipment.

[0182] Step 10: Based on the time-domain diagram of the power equipment, calculate the waveform kurtosis characteristics in the time-domain diagram of the power equipment using the waveform kurtosis formula;

[0183] Step 11: Based on the time-domain diagram of the power equipment, calculate the waveform skewness characteristics in the time-domain diagram of the power equipment using the waveform skewness formula;

[0184] Step 12: Based on the maximum amplitude value feature, root mean square amplitude value feature, waveform kurtosis feature, and waveform skewness feature in the time domain graph of the power equipment, construct the power equipment feature matrix G;

[0185] Step 13: Based on the power equipment feature matrix, select a subset of features for several power equipment according to the label order;

[0186] Step 14: Based on the feature subsets of several power equipment, use the correlation coefficient formula to calculate the correlation score within each feature subset of power equipment, and obtain the correlation score array within several feature subsets of power equipment.

[0187] Step 15: Standardize the correlation score arrays within several power equipment feature subsets and the optimal operation control array to obtain a standardized power equipment feature subset array and a standardized optimal operation control array;

[0188] Step 16: Use the standardized running parameters of the standardized optimal running control array as the optimal running threshold;

[0189] Step 17: Based on the standardized power equipment feature subset array, the feature score maximum value is used as the segmentation threshold to divide the standardized power equipment feature subset array into two leaf sets, obtaining the left leaf tree and the right leaf tree;

[0190] Step 18: Calculate the path length of each cotyledon in the left and right cotyledon trees respectively, and calculate the average path length of each cotyledon based on the path length;

[0191] Step 19: Based on the path length of each sub-leaf and the average path length of each sub-leaf, calculate the anomaly score using an exponential function formula;

[0192] Step 20: Compare the abnormal score with the optimal operating threshold to determine whether the abnormal score exceeds the optimal operating threshold. If it does, it is determined to be an abnormal operation; otherwise, no output is made.

[0193] Step 21: Based on the abnormal state of the power equipment, mark the abnormal location of the abnormal power equipment.

[0194] In summary, the advantages of this invention are: by collecting and analyzing ultrasonic signals, the operating status of power equipment can be evaluated, abnormal partial discharges can be accurately identified, maintenance costs can be reduced, and the stability of power equipment can be improved.

[0195] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A partial discharge monitoring and locating method based on an Internet of Things architecture, characterized in that, The application relates to an electric power equipment anomaly detection method based on an Internet of Things architecture. Collecting ideal operation parameters of all power equipment accessing the Internet of Things architecture to establish an optimal operation contrast array; Based on the to-be-detected electric power equipment, a plurality of ultrasonic wave emission points and a plurality of ultrasonic wave collection points are selected to obtain a plurality of electric power equipment sound signal arrays; Based on a plurality of power equipment acoustic signal arrays, the acoustic signals in a plurality of power equipment acoustic signal arrays are analyzed to determine characteristic data in the acoustic signals, a power equipment characteristic matrix G is established, G=[g 11 …g ij …g nm ], wherein G is a power equipment characteristic matrix, g ij is the jth characteristic data of the ith sampling point, n is the total number of sampling points, and m is the total number of characteristics. Based on the electric power equipment feature matrix and the optimal operation contrast array, the abnormal features in the electric power equipment feature matrix are evaluated to determine the abnormal state of the electric power equipment; Based on the abnormal state of the electric power equipment, the abnormal discharge position of the abnormal state electric power equipment is marked; The evaluation of the abnormal features in the electric power equipment feature matrix based on the electric power equipment feature matrix and the optimal operation contrast array to determine the abnormal state of the electric power equipment specifically comprises the following steps: Based on the electric power equipment feature matrix, a plurality of feature subsets of the electric power equipment are screened out according to a marking sequence; Based on the plurality of feature subsets of the electric power equipment, a correlation coefficient formula is used to calculate the correlation scores in each feature subset of the electric power equipment to obtain a correlation score array in the plurality of feature subsets of the electric power equipment; The correlation score array in the plurality of feature subsets of the electric power equipment is standardized with the optimal operation contrast array to obtain a standardized electric power equipment feature subset array and a standardized optimal operation contrast array; The standardized operation parameters of the standardized optimal operation contrast array are used as optimal operation thresholds; Based on the standardized electric power equipment feature subset array, a feature score maximum value is used as a segmentation threshold to segment the standardized electric power equipment feature subset array into two sub-leaf sets to obtain a left sub-leaf tree and a right sub-leaf tree; The path lengths of each sub-leaf in the left sub-leaf tree and the right sub-leaf tree are calculated, and the average path lengths of each sub-leaf are calculated according to the path lengths; Based on the path lengths of each sub-leaf and the average path lengths of each sub-leaf, an exponential function formula is used to calculate abnormal scores; The abnormal scores are compared with the optimal operation thresholds to determine whether the abnormal scores exceed the optimal operation thresholds, if yes, the operation is determined to be abnormal, and if not, no output is determined; The correlation coefficient formula is as follows: C = | p(g ij , M) | In the formula, C is the correlation score in each feature subset of the electric power equipment, and rho is a density symbol; The path lengths of each sub-leaf in the left sub-leaf tree and the right sub-leaf tree are calculated, and the average path lengths of each sub-leaf are calculated according to the path lengths; wherein is the average length of the rth sub-leaf path, k is the sub-leaf, h is the processing function, r is the sub-leaf path, and Y is the total number of sub-leaves of the sub-leaf tree; The exponential function formula is as follows: In the formula, C 异 is the abnormal score.

2. The partial discharge monitoring and locating method based on the Internet of Things architecture according to claim 1, characterized in that, The selection of the plurality of ultrasonic wave emission points and the plurality of ultrasonic wave collection points based on the to-be-detected electric power equipment to obtain the plurality of electric power equipment sound signal arrays specifically comprises the following steps: The to-be-detected electric power equipment is marked, and the electric power equipment is selected with a plurality of ultrasonic wave emission points and a plurality of ultrasonic wave collection points according to a marking sequence to obtain ultrasonic wave signal data of each electric power equipment; Based on the ultrasonic wave signal data of each electric power equipment, the ultrasonic wave signal data is amplified; According to the amplified ultrasonic wave signal data, continuous ultrasonic wave signal data is discretized into a string of pulses in the form of an analog signal, and the values of the analog signal are represented by encoding the amplitudes and times of the pulses, so that the ultrasonic wave signal is converted into a digital signal; The ultrasonic signals of each power equipment are converted into digital signals, the digital signals are mapped according to the sequence of the power equipment, and a plurality of power equipment sound signal arrays are established.

3. The partial discharge monitoring and locating method based on the Internet of Things architecture according to claim 2, characterized in that, The analysis of the sound signals in the plurality of power equipment sound signal arrays includes: Based on the plurality of power equipment sound signal arrays, the sound signal data in the arrays is quantized to obtain a power equipment sound signal quantization array; Based on the power equipment sound signal quantization array, a time domain graph is drawn, and the starting point, peak, trough and ending point of the waveform in the time domain graph are marked to determine the power equipment time domain graph; Based on the power equipment time domain graph, the maximum amplitude value characteristic in the power equipment time domain graph is calculated using the maximum amplitude formula; Based on the power equipment time domain graph, the root mean square amplitude value characteristic in the power equipment time domain graph is calculated using the root mean square amplitude formula; Based on the power equipment time domain graph, the waveform kurtosis characteristic in the power equipment time domain graph is calculated using the waveform kurtosis formula; Based on the power equipment time domain graph, the waveform skewness characteristic in the power equipment time domain graph is calculated using the waveform skewness formula; Based on the maximum amplitude value characteristic in the power equipment time domain graph, the root mean square amplitude value characteristic in the power equipment time domain graph, the waveform kurtosis characteristic in the power equipment time domain graph and the waveform skewness characteristic in the power equipment time domain graph, a power equipment characteristic matrix is constructed.

4. The partial discharge monitoring and locating method based on the Internet of Things architecture according to claim 3, characterized in that, The maximum amplitude formula is: A max = max(|x(t)|) In the formula, A max is the maximum amplitude value feature in the time domain graph of the power equipment, x is the amplitude value, and t is the time. The root mean square amplitude formula is: In the formula, B rms is the root mean square amplitude value feature in the time domain graph of the power equipment, n is the total number of sampling points in the time domain graph, and i is the i th sampling point. The waveform kurtosis formula is: where K is the kurtosis characteristic of the waveform in the time domain plot of the power device, x i is the amplitude value of the i-th sampling point, is the mean value of the amplitude; The waveform skewness formula is: In the formula, S is the waveform skewness characteristic in the power equipment time domain graph, and V is the standard deviation in the power equipment time domain graph.

5. A partial discharge monitoring and locating system based on an Internet of Things architecture, characterized by, A partial discharge monitoring and positioning method based on an Internet of Things architecture is implemented, as claimed in any one of claims 1-4, comprising: A standard parameter acquisition module, the standard parameter acquisition module is used for collecting power equipment ideal operation parameters, and establishing an optimal operation control array; A sound data acquisition module, the sound data acquisition module is used for selecting a plurality of ultrasonic emission points and ultrasonic collection points based on the power equipment to be detected, and obtaining a plurality of power equipment sound signal arrays; A sound data analysis module, the sound data analysis module is electrically connected with the sound data acquisition module, and the sound data analysis module is used for analyzing the sound signals in the plurality of power equipment sound signal arrays based on the plurality of power equipment sound signal arrays, determining the characteristic data in the sound signals, and establishing a power equipment characteristic matrix; An abnormality evaluation module, the abnormality evaluation module is electrically connected with the sound data analysis module, and the abnormality evaluation module is used for evaluating the abnormal characteristics in the power equipment characteristic matrix based on the power equipment characteristic matrix and the optimal operation control array, and determining the abnormal state of the power equipment; A marking module, the marking module is electrically connected with the abnormality evaluation module, and the marking module is used for marking the abnormal discharge position of the abnormal state power equipment based on the abnormal state of the power equipment.

6. The partial discharge monitoring and locating system based on the Internet of Things architecture according to claim 5, characterized in that, The sound data acquisition module includes: The acquisition unit marks the power equipment to be detected, selects a plurality of ultrasonic wave emitting points and ultrasonic wave collecting points for the power equipment in the marked order, and acquires ultrasonic wave signal data of each power equipment. The amplification processing unit amplifies the ultrasonic wave signal data of each power equipment based on the ultrasonic wave signal data of each power equipment. The signal conversion unit converts the continuous ultrasonic wave signal data into a series of pulses in the form of an analog signal after amplification processing, and encodes the amplitude and time of the pulses to represent the value of the analog signal, so as to realize the conversion of the ultrasonic wave signal into a digital signal. The acoustic signal array unit converts the ultrasonic wave signal of each power equipment into a digital signal, maps the digital signal in the order of the power equipment, and establishes a plurality of acoustic signal arrays of the power equipment.

7. The partial discharge monitoring and locating system based on the Internet of Things architecture according to claim 5, characterized in that, The acoustic data analysis module includes: The quantization unit quantizes the acoustic data in the array based on a plurality of sets of acoustic signal arrays of the power equipment, and obtains a quantized acoustic signal array of the power equipment. The time domain graph unit draws a time domain graph based on the quantized acoustic signal array of the power equipment, marks the waveform starting point, wave crest, wave trough and waveform ending point in the time domain graph, and determines the time domain graph of the power equipment. The maximum amplitude feature unit calculates the maximum amplitude value feature in the time domain graph of the power equipment based on the time domain graph of the power equipment. The root mean square amplitude feature unit calculates the root mean square amplitude value feature in the time domain graph of the power equipment based on the time domain graph of the power equipment. The waveform kurtosis feature unit calculates the waveform kurtosis feature in the time domain graph of the power equipment based on the time domain graph of the power equipment. The waveform skewness feature unit calculates the waveform skewness feature in the time domain graph of the power equipment based on the time domain graph of the power equipment. The feature matrix unit constructs a feature matrix G of the power equipment based on the maximum amplitude value feature in the time domain graph of the power equipment, the root mean square amplitude value feature in the time domain graph of the power equipment, the waveform kurtosis feature in the time domain graph of the power equipment and the waveform skewness feature in the time domain graph of the power equipment.

8. The partial discharge monitoring and locating system based on the Internet of Things architecture according to claim 5, characterized in that, The abnormality evaluation module includes: The screening unit screens a feature subset of a plurality of power equipments based on the feature matrix of the power equipment according to the marked order. The correlation calculation unit calculates the correlation score in each power equipment feature subset based on the feature subset of the plurality of power equipments by using a correlation coefficient formula, and obtains a correlation score array in the feature subset of the plurality of power equipments. The standardization unit standardizes the correlation score array in the feature subset of the plurality of power equipments and the optimal operation control array to obtain a standardized feature subset array of the power equipment and a standardized optimal operation control array. The threshold unit takes the standardized operation parameter of the standardized optimal operation control array as an optimal operation threshold. The segmentation unit segments the standardized feature subset array of the power equipment into two subsets based on the maximum feature score in the standardized feature subset array of the power equipment, and obtains a left subset tree and a right subset tree. The path length calculation unit calculates the path length of each sub-leaf in the left sub-leaf tree and the right sub-leaf tree respectively, and calculates the average path length of each sub-leaf according to the path length; The abnormal score calculation unit calculates the abnormal score based on the path length of each sub-leaf and the average path length of each sub-leaf through an exponential function formula; The abnormal judgment unit compares the abnormal score with the optimal operation threshold, judges whether the abnormal score exceeds the optimal operation threshold, if yes, determines that the operation is abnormal, if not, determines not to output.

Images