Electrical fault detection control system

Abstract

The invention relates to the technical field of fault detection, in particular to an electrical fault detection control system which comprises a phase monitoring module, an abnormity positioning module, a current control module and a result verification module. According to the method, the phase difference is calculated by combining collection and dynamic time warping, an abnormal mark is generated by comparing a phase tolerance threshold value, the phase delay detection precision is improved, based on an adjacent node phase difference value sequence, the slope kick is analyzed through linear regression, the candidate section judgment accuracy is enhanced, and a constant current source collects a multi-node voltage difference value; the Kalman filtering smoothes the voltage drop difference and compares the threshold value, improves the anti-interference capability of the complex working condition, compares the overlapping degree of the path coordinates and forms a closed loop verification mechanism with a re-check instruction, guarantees the reliability of the positioning result, shortens the fault response time through the multi-stage collaborative optimization, reduces the misoperation, and improves the discrimination robustness through the dynamic modeling and multi-source signal fusion. The dependence on experience or threshold is reduced; and the self-adaptive capability is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of fault detection, and particularly to an electrical fault detection and control system. Background Art

[0002] The technical field of fault detection includes the technical content of identifying, judging, and processing various abnormal states that occur during the operation of electrical equipment or systems. Its core lies in continuously monitoring and identifying problems such as short circuits, open circuits, overloads, voltage fluctuations, and temperature abnormalities that may occur in circuits, electrical equipment, or the overall system during operation, and judging whether there is a fault by setting parameters, comparing thresholds, analyzing current and voltage fluctuations, etc. This technical field systematically covers the research and implementation of fault signal acquisition, fault type discrimination, fault location, and related response mechanisms, and is often applied in power systems, industrial control, automation equipment, and intelligent terminals. It is an important part to ensure the safe and stable operation of electrical systems.

[0003] Among them, an electrical fault detection and control system refers to a system that, for possible fault conditions during the operation of an electrical system, collects electrical parameter data such as voltage, current, and frequency, uses preset detection rules to compare and identify fault types, and executes relevant control operations based on the fault judgment results. Its specific technical content includes electrical state monitoring based on real-time sampling of current and voltage, using differential calculation to identify short-term overload situations, judging open circuits or poor contacts by setting the change rate of electrical parameters, identifying abnormal power supply states using frequency offset, and constructing a fault detection and control linkage mechanism by means of controlling the corresponding circuit to cut off or starting an alarm device with an abnormal trigger signal. This system mainly constitutes the overall technical process through an analog electrical parameter acquisition unit, digital signal recognition logic, fault type logic discrimination rules, and control response instruction sets.

[0004] Fixed thresholds or preset rules are difficult to adapt to dynamic load fluctuations, resulting in misjudgment or missed detection. Short-term overload detection relies on differential calculation but lacks context feature association, and is prone to false alarms due to instantaneous fluctuations. Single electrical parameter analysis cannot distinguish similar faults such as poor contact and abnormal power supply, reducing maintenance efficiency. The independent monitoring unit does not establish cross-node timing correlation, and has insufficient early identification of progressive faults. The one-way trigger control mechanism lacks closed-loop verification, and detection rule deviations may directly cause mis-cutoff or delayed alarm. Voltage fluctuation analysis does not cooperate with frequency offset and phase delay changes, and is prone to misjudging the root cause of faults. Existing technologies are limited in detection accuracy and reliability in multi-node coupled fault or transient abnormal scenarios, increasing operation and maintenance costs and potential safety hazards. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art and propose an electrical fault detection and control system.

[0006] To achieve the above object, the present invention adopts the following technical solutions: The electrical fault detection control system includes: A phase monitoring module, configured to obtain voltage waveform sampling time points through a synchronous voltage acquisition device, call a dynamic time warping algorithm to calculate the phase difference between the time points from the sampling start moment to the zero-crossing point and the standard node reference value, compare the phase delay parameter item with the phase tolerance threshold, generate a phase anomaly mark, and transmit the phase anomaly mark to the anomaly location module; An anomaly location module, configured to obtain a sequence of phase differences of adjacent nodes based on the phase anomaly mark, determine a sudden jump in the change slope of the phase differences of three consecutive nodes through a linear regression model, generate candidate section coordinates, and transmit the candidate section coordinates to the current control module; A current control module, configured to start a constant current source device according to the candidate section coordinates, collect multi-node voltage difference data after injecting a constant current, use a Kalman filtering algorithm to smooth the voltage drop difference sequence between adjacent nodes, compare the voltage drop difference value with a set threshold parameter, generate a resistance sudden rise section mark, and transmit the resistance sudden rise section mark to the result verification module.

[0007] As a further solution of the present invention, the phase anomaly mark specifically includes a phase delay level, a waveform offset feature, and a time series label, the candidate section coordinates include a node index interval, a change rate identifier, and a time label, and the resistance sudden rise section mark specifically refers to a voltage drop change rate, a target node index, and an abnormal section length.

[0008] As a further solution of the present invention, the dynamic time warping algorithm uses a Gaussian window function to align the start moment to the reference time axis of the standard node reference value, the window width is 1 / 10 to 1 / 5 of the sampling period, and the weight factor is 0.6 to 0.8; The state transition matrix of the Kalman filtering algorithm is [1, Δt; 0, 1], the observation matrix is [1, 0], the process noise covariance is 0.01 to 0.05, and the observation noise covariance is 0.1 to 0.3.

[0009] As a further solution of the present invention, the phase monitoring module includes: A waveform sampling sub-module monitors the voltage waveform through a synchronous voltage acquisition device, records the start moment coordinates within the sampling period, detects the timing position of the first zero-crossing point, extracts the time interval data from the start moment to the zero-crossing point, and generates a time point sequence; A phase difference calculation sub-module calls a dynamic time warping algorithm to align the start moment of the time point sequence to the reference time axis of the standard node reference value, calculates the timing offset amount between the two in the zero-crossing point interval, obtains the average value of the offset amount as the waveform alignment deviation, and generates a phase difference parameter; The phase anomaly determination sub-module compares the phase difference parameter with a preset phase tolerance threshold range, determines whether the average offset exceeds the upper or lower threshold. If it exceeds the limit, the binary flag variable is assigned a value of 1, otherwise 0, to generate a phase anomaly flag. The phase tolerance threshold range is determined by optimizing historical fault data through the gradient descent method. The upper limit value is 5% to 8% of the sampling period, and the lower limit value is 2% to 4% of the sampling period.

[0010] As a further aspect of the present invention, the anomaly location module includes: The phase sequence sampling sub-module filters the set of phase difference values of adjacent nodes based on the phase anomaly flag, extracts the difference data of three consecutive nodes in chronological order, and generates an adjacent phase difference sequence. The slope sudden jump determination sub-module calls a linear regression model, performs least squares fitting on the phase difference value of each node in the adjacent phase difference sequence, calculates the absolute value of the slope difference between the fitting lines of adjacent sections, and compares it with the upper limit value of the preset sudden jump determination threshold. If the absolute value of the slope difference exceeds the threshold, it is marked as a mutation section to generate a slope sudden jump coefficient. The absolute value of the slope difference is dimensionless and processed into a percentage form. The sudden jump determination threshold is 15% to 25%. The candidate section calibration sub-module extracts the starting and ending node numbers corresponding to the mutation section according to the slope sudden jump coefficient, calculates the node spacing between adjacent mutation points. If the spacing is less than the preset section merging length, multiple mutation points are merged into a single interval to generate candidate section coordinates. The preset section merging length is 1.5 to 2.5 times the electrical node spacing, and the node spacing is pre-calibrated based on the cable model to be 0.5 meters to 1.2 meters.

[0011] As a further aspect of the present invention, the current control module includes: The constant current injection control sub-module collects the candidate section coordinates, analyzes the physical node distribution corresponding to the coordinates, sets the output parameters of the constant current source according to the node spacing, activates the constant current source device to inject a preset current value into the target section, synchronously measures the multi-node ground voltage, and calculates the voltage difference between adjacent nodes to generate multi-node voltage difference data. The voltage difference anomaly analysis sub-module calls the multi-node voltage difference data, extracts the adjacent node voltage drop difference values in sequence by node order to form a sequence, and uses the formula: ; Performs iterative calculations on the sequence, eliminates noise interference by gradually correcting the difference values, and generates a smoothed voltage drop difference sequence. Among them, represents the corrected voltage drop difference value in the th iteration, represents the original value of the pressure drop difference collected at the current node in the th iteration, represents the pressure drop difference value between adjacent nodes in the th iteration, represents the suppression factor for the environmental noise amplitude, which is used to reduce the influence of noise on the difference value, represents the environmental noise amplitude collected by the sensor in the th iteration, represents the dynamic sensitivity coefficient, which is used to adjust the response speed of the correction value, represents the total number of data nodes participating in the calculation within the current section, represents the baseline pressure drop offset obtained through statistical analysis of long-term monitoring data, which is used to eliminate the influence of environmental steady-state interference; The threshold discrimination sub-module calls the smoothed pressure drop difference sequence, traverses the corrected difference values of each node in the sequence, compares them numerically with the preset resistance sudden increase threshold parameter, marks the node intervals where the difference values exceed the threshold, and generates a resistance sudden increase section mark according to the node interval coordinates; The preset resistance sudden increase threshold parameter is 3 to 5 times the pressure drop difference value under normal operating conditions, and is determined by statistical analysis of historical fault data.

[0012] As a further solution of the present invention, the system further includes: A result verification module, which is used to receive the candidate section coordinates and the resistance sudden increase section mark, perform a path coordinate comparison operation, output a positioning result when the overlap degree exceeds the proportional value threshold, trigger a review instruction and send it back to the phase monitoring module and the current control module when it is not satisfied, and output the positioning result to an external alarm unit.

[0013] As a further solution of the present invention, the positioning result includes a fault coordinate value, a matching overlap degree, and a positioning confidence level.

[0014] As a further solution of the present invention, the result verification module includes: The section receiving sub-module calls the start node number and end node number of the candidate section coordinates, extracts the coordinate range of the resistance sudden increase section mark, combines the node number intervals and mark states of the two into a unified data format, and generates a section set; The overlap degree calculation sub-module compares the coordinate ranges of the start point of the candidate section and the end point of the resistance sudden increase section in the section set, calculates the length of the intersection interval on the node number axis, and counts the percentage of the intersection node number in the total number of nodes of the candidate section to generate an overlap ratio value; The verification decision sub-module compares the overlapping ratio value with a preset ratio value threshold. If the ratio value exceeds the threshold, it outputs the positioning result to the external alarm unit. If it does not exceed the threshold, it generates a review instruction and sends it back to the phase monitoring module and the current control module; The ratio value threshold is 75% to 90%, which is optimized and determined by the cross-validation method and set to 80% when the positioning accuracy reaches 95% confidence level.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, the phase difference is calculated by collecting and combining with the dynamic time warping algorithm, and the abnormal mark is generated by comparing with the phase tolerance threshold, so as to improve the phase delay detection accuracy. Based on the phase difference sequence of adjacent nodes, the linear regression model is used to analyze the sudden jump of the slope of continuous nodes, so as to enhance the accuracy of candidate section determination. The constant current source device collects the voltage difference data of multiple nodes, combines with the Kalman filter to smooth the voltage drop difference sequence and compares with the threshold, so as to improve the anti-interference ability under complex working conditions. The path coordinate overlap ratio comparison and the review instruction form a closed-loop verification mechanism to ensure the reliability of the positioning result. The multi-stage collaborative optimization shortens the fault response time and reduces the misoperation. The dynamic data modeling and multi-source signal fusion improve the robustness of fault discrimination, avoid the missed detection under transient abnormal scenarios, reduce the dependence on manual experience or preset thresholds, and enhance the adaptive ability. Brief Description of the Drawings

[0016] Figure 1 is the system flow chart of the present invention; Figure 2 is the flow chart of the phase monitoring module of the present invention; Figure 3 is the flow chart of the abnormal positioning module of the present invention; Figure 4 is the flow chart of the current control module of the present invention; Figure 5 is the flow chart of the result verification module of the present invention. Detailed Embodiment

[0017] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0018] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, in the description of the present invention, the meaning of "a plurality" is two or more, unless otherwise specifically defined. Embodiment 1

[0019] Please refer to Figure 1 , the electrical fault detection and control system includes: A phase monitoring module, which is used to obtain the voltage waveform sampling time points through a synchronous voltage acquisition device, call the dynamic time warping algorithm to calculate the phase difference between the time points from the sampling start moment to the zero crossing point and the standard node reference value, compare the phase delay parameter item with the phase tolerance threshold, generate a phase anomaly mark, and transfer the phase anomaly mark to the anomaly location module; An anomaly location module, which is used to obtain the phase difference sequence of adjacent nodes based on the phase anomaly mark, determine the sudden jump improvement of the phase difference change slope of three consecutive nodes through a linear regression model, generate candidate section coordinates, and transfer the candidate section coordinates to the current control module; A current control module, which is used to start a constant current source device according to the candidate section coordinates, collect the multi-node voltage difference data after injecting a constant current, use the Kalman filter algorithm to smooth the voltage drop difference sequence between adjacent nodes, compare the voltage drop difference value with the set threshold parameter, generate a resistance sudden rise section mark, and transfer the resistance sudden rise section mark to the result verification module; A result verification module, which is used to receive the candidate section coordinates and the resistance sudden rise section mark, perform path coordinate comparison operations, output the location result when the overlap degree exceeds the proportional value threshold, trigger a review instruction and send it back to the phase monitoring module and the current control module when it is not satisfied, and output the location result to the external alarm unit.

[0020] The phase anomaly mark specifically includes a phase delay level, a waveform offset feature, and a time series label. The candidate section coordinates include a node index interval, a change rate identifier, and a time label. The resistance sudden rise section mark specifically refers to a voltage drop change rate, a target node index, and an abnormal section length. The location result includes a fault coordinate value, a matching overlap degree, and a location confidence level.

[0021] The dynamic time warping algorithm uses a Gaussian window function to align the start moment to the reference time axis of the standard node. The window width is 1 / 10 to 1 / 5 of the sampling period, and the weight factor is 0.6 to 0.8; The state transition matrix of the Kalman filter algorithm is [1, Δt; 0, 1], the observation matrix is [1, 0], the process noise covariance is from 0.01 to 0.05, and the observation noise covariance is from 0.1 to 0.3.

[0022] Please refer to Figure 2 , the phase monitoring module includes: The waveform sampling sub-module monitors the voltage waveform through a synchronous voltage acquisition device, records the starting time coordinate within the sampling period, detects the timing position of the first zero-crossing point, extracts the time interval data from the starting time to the zero-crossing point, and generates a time point sequence; The waveform sampling sub-module monitors the voltage waveform of the line through a synchronous voltage acquisition device. In a specific implementation scenario, for an AC power system with a power frequency of 50 Hz, the sampling frequency of the voltage acquisition device is set to 10 kHz, which means that the instantaneous voltage value is recorded every milliseconds (ms). A complete power frequency cycle T = 1 / 50 = 0.02 s = 20 ms, and within this cycle, 20 ms ÷ 0.1 ms / point = 200 points of instantaneous voltage values will be collected.

[0023] After the acquisition starts, the system records the starting time coordinate within the sampling period, and the specific value is ms, and the corresponding voltage value at this time . Subsequently, the device continuously collects voltage values and corresponding timestamps to form the original voltage sequence data. By continuously comparing the voltage value signs of adjacent sampling points, the system detects when the voltage waveform crosses zero. When it is monitored that ms, the voltage is V, and at the immediately following sampling point ms, the voltage becomes V. The system determines that a zero crossing occurs between and time points. Use the linear interpolation method to estimate a more accurate zero crossing time point , and the calculation process is: .

[0024] The system then extracts all sampling data points within the time interval from the recorded starting time = 10.000 ms to the calculated first zero crossing point = 15.047 ms, including their timestamps and voltage values.

[0025] Table 1 Voltage data table for sampling time interval

[0026] As shown in Table 1, the table lists some sampling time points and their corresponding voltage values. Based on these extracted data points, the system generates an ordered sequence of time points , which forms the basis for subsequent phase difference calculations.

[0027] The phase difference calculation sub-module calls the dynamic time warping algorithm to align the start time of the time point sequence to the reference time axis of the standard node reference value, calculates the timing offset in the zero-crossing point interval between the two, obtains the average value of the offset as the waveform alignment deviation, and generates the phase difference parameter; The phase difference calculation sub-module receives the time point sequence generated by the previous sub-module and calls the dynamic time warping (DTW) algorithm. First, the system obtains a time point sequence of the voltage waveform generated by a specified standard node (in this embodiment, the system stable reference power supply point is selected) within the same sampling period as the reference sequence . The starting sampling time of this reference sequence is defined as the zero point of the reference time axis.

[0028] Subsequently, the time point sequence of the node to be measured is subjected to DTW alignment processing with the standard reference sequence. The specific execution action of DTW is: construct a cost matrix, and the element in the matrix represents the Euclidean distance between the th point of the sequence to be measured and the th point of the reference sequence . Through dynamic programming, find an optimal path from the starting point to the end point of the matrix, and this path minimizes the sum of the cumulative distances of all elements along the way. This path defines the non-linear time correspondence relationship between the two time series.

[0029] During this alignment process, focus on the corresponding situations near the zero-crossing points of the two sequences (sequence to be measured) and (reference sequence). Through the optimal warping path, determine the time point corresponding to the zero-crossing point = 15.047 ms of the sequence to be measured after alignment to the reference time axis . Suppose the zero-crossing point of the standard reference sequence occurs at = 15.000 ms on the reference time axis. After DTW alignment, the corresponding time of the zero-crossing point of the sequence to be measured is calculated as = 15.185 ms. Calculate the timing offset between these two zero-crossing points.

[0030] To obtain a stable phase difference evaluation, the system calculates the arithmetic mean of the timing offsets ms calculated for 10 consecutive power frequency cycles, and calculates the waveform alignment deviation Output the calculated mean value as the final phase difference parameter, that is, generate the phase difference parameter = 0.187ms

[0031] The phase anomaly determination sub-module compares the phase difference parameter with the preset phase tolerance threshold range, determines whether the mean value of the offset exceeds the upper or lower threshold. If it exceeds the limit, assign the binary flag variable to 1, otherwise 0, and generate a phase anomaly flag The phase tolerance threshold range is determined by optimizing historical fault data through the gradient descent method. The upper limit value is 5% to 8% of the sampling period, and the lower limit value is 2% to 4% of the sampling period

[0032] The phase anomaly determination sub-module receives the phase difference parameter generated by the previous sub-module = 0.187ms. The system compares it with the preset phase tolerance threshold range. This threshold range is not fixed, but is dynamically determined by applying the gradient descent optimization algorithm to a large number of historical fault data (including data records during normal operation and different types of phase faults) The goal of optimization is to find a set of upper and lower threshold values that can best distinguish between normal and abnormal phase offset states in the historical data set, that is, to maximize the classification accuracy

[0033] According to the recommendation of the optimization algorithm and combined with the requirements of engineering experience for system stability, set the upper and lower thresholds to float within a specific percentage range: the upper limit value is 5% to 8% of the sampling period, and the lower limit value is 2% to 4% of the sampling period. In this embodiment, the sampling period is 20ms. After performing gradient descent optimization calculations on 500 collected historical events (350 of which are normal and 150 are known phase anomalies), the algorithm converges to determine the best upper limit ratio of 7% and the lower limit ratio of 3%. Calculate the specific thresholds accordingly: the upper limit of the phase tolerance threshold ; the lower limit of the phase tolerance threshold .

[0034] Next, compare the currently calculated phase difference parameter = 0.187ms with this threshold range . The judgment condition is or . In this specific calculation, 0.187ms is between 0.6ms and 1.4ms, satisfying . Therefore, the system determines that the phase state of this node is normal and assigns the corresponding binary flag variable to 0

[0035] Considering that the calculation result of the phase difference parameter of another node is . Compare: 1.55 > 1.4, satisfying conditions to determine that the phase of the node is abnormal. At this time, the system assigns a binary marker variable of this node a value of 1. By executing this determination process for all nodes in the monitoring network, a phase anomaly marker sequence for each node is finally generated .

[0036] Please refer to Figure 3 , the anomaly location module includes: The phase sequence sampling sub-module filters the set of phase differences of adjacent nodes based on the phase anomaly markers, extracts the difference data of three consecutive nodes in chronological order, and generates an adjacent phase difference sequence; The phase sequence sampling sub-module filters data according to the phase anomaly markers of each node generated by the previous module set. The system traverses the marker set, identifies all nodes with a marker value of 1, and these nodes are initially considered to have phase anomalies. Subsequently, the system queries the phase differences of these abnormal nodes and their immediately adjacent front and rear nodes (these values were previously calculated and stored by the phase difference calculation sub-module).

[0037] In a specific monitored line segment, the phase anomaly markers of nodes 5, 6, 7, and 8 are 0, 1, 1, and 0 respectively. The system filters out the abnormal nodes 6 and 7. Then, the phase difference parameter values of these two abnormal nodes and their directly adjacent nodes (node 5 and node 8) are extracted to obtain a relevant set of phase differences: .

[0038] The system extracts the phase difference data of three consecutive nodes from the above set according to the physical or logical chronological order of the nodes on the line, forming a subsequence required for analysis. Centered on node 6, the sequence is extracted. Then, centered on node 7, the sequence is extracted. All such three-node subsequences extracted are combined to generate an adjacent phase difference sequence for subsequent slope analysis, and its structure is , and the units are all ms.

[0039] The slope sudden jump determination sub-module calls a linear regression model, performs least squares fitting on the phase difference of each node in the adjacent phase difference sequence, calculates the absolute value of the slope difference between the fitting lines of adjacent sections, and compares it with the upper limit value of the preset sudden jump determination threshold. If the absolute value of the slope difference exceeds the threshold, it is marked as a mutation section and a slope sudden jump coefficient is generated; The absolute value of the slope difference is dimensionless and processed into a percentage form, and the sudden jump determination threshold is 15% to 25%; The slope sudden jump determination sub-module receives the adjacent phase difference sequence generated by the previous sub-module . The system processes each three-node subsequence in this sequence one by one.

[0040] For the first subsequence , this represents the phase differences of nodes 5, 6, and 7. The system applies the least squares method to perform piecewise linear fitting on these three data points. Regarding the node numbers (5, 6, 7) as the independent variable x and the corresponding phase differences (0.187, 1.55, 1.62) as the dependent variable y. Calculate the slope of the section from node 5 to node 6 node. Calculate the slope of the section from node 6 to node 7 node.

[0041] Calculate the absolute value of the difference in the slopes of the fitting lines between these two adjacent sections (5 - 6 and 6 - 7) node. For standardized comparison, make this absolute value of the slope difference dimensionless and convert it to a percentage form. Usually, take the slope of the section before the change as the benchmark (provided that ). Calculate the percentage of the slope change: .

[0042] The system sets an upper threshold for slope jump determination. This threshold is set based on statistical analysis of the slope change data of the phase difference sequences under a large number of historical normal operation and fault conditions to determine a boundary that can effectively distinguish normal fluctuations from abnormal mutations, usually set between 15% and 25%. In this embodiment, through analyzing historical data, is selected as the jump determination threshold. A percentage of slope change lower than 10% is regarded as "low" or "stable", between 10% and 20% is regarded as "medium" fluctuation, and exceeding 20% is determined as "high" or "jump", indicating a significant change in the phase difference change rate.

[0043] Compare the calculated percentage of slope change 94.86% with the threshold 20%. Since 94.86% > 20%, the system determines that a slope jump has occurred near node 6 (i.e., section 6 - 7 relative to section 5 - 6), and marks this position as a potential abnormal point.

[0044] Repeat this process for the second subsequence . It involves nodes 6, 7, and 8. Given that = 0.07 ms / node. Calculate the slope of the section from node 7 to node 8 node. Calculate the absolute value of the slope difference node. Take as the benchmark to calculate the percentage change (Note: When the benchmark slope is close to zero, the percentage change may be very large or meaningless. In this case, an absolute difference threshold or the average slope (taking [baseline] as the reference). If the average slope is used as the reference, the percentage change is . Since , the system determines that a slope jump also occurs near Node 7 (section 7 - 8 relative to section 6 - 7).

[0045] Finally, the system generates a slope jump coefficient for each node (in the subsequent section), which is the calculated percentage of slope change, such as , . These coefficients form a sequence of slope jump coefficients.

[0046] The candidate section calibration sub - module extracts the starting and ending node numbers corresponding to the mutation section according to the slope jump coefficient, calculates the node spacing between adjacent mutation points, and if the spacing is less than the preset section merging length, multiple mutation points are merged into a single interval to generate candidate section coordinates; The preset section merging length is 1.5 to 2.5 times the electrical node spacing, and the node spacing is pre - calibrated based on the cable type to be 0.5 meters to 1.2 meters.

[0047] The candidate section calibration sub - module is based on the slope jump coefficient sequence generated by the previous sub - module and the corresponding mutation section markers. According to , it is determined that significant slope changes occur after Node 6 and Node 7, and the sections related to these jumps are marked as mutation sections. Specifically, the first jump occurs after Node 6 (involving sections 5 - 6 and 6 - 7), and the second jump occurs after Node 7 (involving sections 6 - 7 and 7 - 8). The system extracts the node numbers related to the mutation and identifies that the mutation points occur near Node 6 and Node 7.

[0048] The system merges these consecutive mutation points or adjacent mutation sections to form one or more candidate fault sections. First, calculate the node spacing between adjacent mutation points (or the core nodes affected by the mutation). In this case, the mutation points are identified at Nodes 6 and 7, which are adjacent, and the spacing is 1 node.

[0049] The system sets a preset section merging length , which is used to determine whether multiple independent mutation points with close distances need to be merged into a continuous candidate section. The value of is determined based on the average physical spacing of electrical nodes in the line The value of is pre - calibrated according to factors such as the cable type used and the laying method, and its value is usually between 0.5 meters and 1.2 meters. In this embodiment, the calibrated node spacing corresponding to the cable type used is meters. The preset section merging length Set to times the electrical node spacing, where ranges from 1.5 to 2.5. Select , then meters. This means that if the physical distance between two mutation points is less than 2.0 meters (or the difference in node numbers is less than ), they are grouped into the same candidate section.

[0050] Check the spacing between the identified mutation points, node 6 and node 7. They are adjacent nodes with a spacing of 1 node, corresponding to a physical distance of meters. Since (or the node spacing ), the merging condition is satisfied. The system merges these two mutation points and defines the range they affect (usually including these mutation points and the immediately adjacent sections before and after) as a continuous candidate section. Considering that the mutations occur after node 6 and after node 7, the merged candidate section covers from the starting point of the previous section that caused the first mutation (node 5) to the ending point of the subsequent section affected by the last mutation (node 8), so the candidate section is determined to be from node 5 to node 8. Finally, generate the candidate section coordinates .

[0051] Please refer to Figure 4 , the current control module includes: The constant current injection control sub-module collects the candidate section coordinates, analyzes the physical node distribution corresponding to the coordinates, sets the output parameters of the constant current source according to the node spacing, activates the constant current source device to inject a preset current value into the target section, synchronously measures the voltages of multiple nodes to the ground, calculates the voltage differences between adjacent nodes, and generates multi-node voltage difference data; The constant current injection control sub-module receives the candidate section coordinates determined by the previous sub-module . The system analyzes this coordinate and determines that the physical node range that needs to be focused on for detection is node 5, node 6, node 7, and node 8. According to the physical distribution of these nodes and the known node spacing (all 1.0 meters), the system configures the output parameters of the constant current source. The constant current source device is connected between the two ends of the candidate section, that is, between node 5 and node 8. Set the constant current source to output a stable, preset current value, and select A DC current.

[0052] The system activates the constant current source connected between node 5 and node 8 to inject 5 A of current into this line segment. During the stable injection of the current, the voltage measurement units located at the candidate section (nodes 5 to 8) and possibly included reference nodes are triggered synchronously to collect the voltage values of each node to the ground.

[0053]

[0054] As shown in Table 2, this table records the measured values of the ground voltages of each node in the candidate section and calculates the voltage differences between adjacent nodes. The system collects these voltage differences V, V, V. Organize these calculated voltage differences in node order to generate a multi-node voltage difference data sequence V, and this sequence will be passed to the next sub-module for analysis.

[0055] The voltage difference abnormal analysis sub-module calls the multi-node voltage difference data, extracts the voltage drop difference values between adjacent nodes in node order to form a sequence, and uses the formula: ; Perform iterative calculations on the sequence, eliminate noise interference by correcting the difference values item by item, and generate a smooth voltage drop difference sequence; Among them, represents the corrected voltage drop difference value in the th iteration, represents the original value of the voltage drop difference collected at the current node in the th iteration, represents the voltage drop difference value between adjacent nodes in the th iteration, represents the suppression factor for the environmental noise amplitude, which is used to reduce the influence of noise on the difference value, represents the environmental noise amplitude collected by the sensor in the th iteration, represents the dynamic sensitivity coefficient, which is used to adjust the response speed of the correction value, represents the total number of data nodes participating in the calculation in the current section, represents the baseline voltage drop offset obtained by statistical analysis of long-term monitoring data, which is used to eliminate the influence of environmental steady-state interference; The voltage difference abnormal analysis sub-module receives the multi-node voltage difference data sequence V generated by the previous sub-module, where the subscript corresponds to line sections 5-6, 6-7, and 7-8 respectively. The system processes this sequence using a specific iterative formula, aiming to correct the measured values, smooth the data, and highlight potential abnormal voltage drops.

[0056] The iterative formula is :

[0057] Detailed explanations and assignments of parameters: is the corrected voltage drop difference value (V) of the th node interval.

[0058] is the original measured pressure drop difference value (V) of the th interval, from the sequence .

[0059] is the pressure drop difference value (V) of the previous adjacent interval. In the iterative calculation, it represents the spatial proximity effect.

[0060] is the environmental noise suppression factor. By performing a correlation analysis on historical noise data and pressure drop measurement errors and conducting experimental verification, it is determined that can effectively suppress noise without overly weakening the real signal changes. This is a dimensionless value.

[0061] is the equivalent environmental noise voltage (V) affecting the measurement of the th interval. Environmental noise can be generated by various sources and is obtained through direct measurement or indirect evaluation by sensors. Here, it is assumed that the measured noise amplitude by the sensor is , with the unit of millivolt (mV). In order to perform operations with other voltage values (V) in the formula, unit conversion is required. Set the conversion rule: divide the measured millivolt value by 1000 to obtain the volt value.

[0062] . This rule is based on the physical model where the noise signal is directly superimposed on the measured voltage signal. The measured noise amplitudes corresponding to intervals 1, 2, and 3 are mV, mV, mV respectively. The values used for calculation after conversion are V, V, V.

[0063] is the dynamic sensitivity coefficient, dimensionless. It adjusts the response speed and smoothness of the correction result to the changes in the original data. Through simulation tests, by comparing the detection effects of different values (ranging from 0.1 to 1.0) on typical fault signals, it is found that when , the system responds most quickly and stably to abnormal pressure drops. Therefore, is set.

[0064] is the total number of pressure drop intervals participating in the calculation within the current candidate section. For the section from node 5 to 8, there are 3 intervals (5 - 6, 6 - 7, 7 - 8), so .

[0065] is the baseline pressure drop offset (V), which is obtained by statistically averaging the voltage difference data of this line segment or line segments of the same type during long-term (e.g., 100 hours) operation under the confirmed fault-free state. The calculation result shows that there is an average systematic offset of V, and V is set.

[0066] Iterative calculation process (perform one iteration): First, calculate the summation term in the denominator of the formula: ; Next, calculate the corrected pressure drop value for each interval . For (interval 5 - 6), it is necessary to That is the value of. Set the boundary condition and use the value of as the approximate value of, that is V.

[0067] ; For (interval 6 - 7), use V.

[0068] ; For (interval 7 - 8), use V.

[0069] ; After a single iteration calculation, the system obtains a smoothed pressure drop difference sequence V. The benefit of this formula is that it combines the current measured value, the influence of adjacent intervals, and the equivalent voltage influence of real-time noise, amplifies the abnormal signal through non-linear operations (absolute value, square root), and at the same time performs adaptive smoothing using the normalized denominator and sensitivity coefficient containing all interval information, effectively suppressing the interference of noise and baseline drift, thereby revealing potential resistance sudden rise points more clearly.

[0070] The threshold discrimination sub-module calls the smoothed pressure drop difference sequence, traverses the corrected difference values of each node in the sequence, makes a numerical comparison with the preset resistance sudden rise threshold parameter, marks the node intervals where the difference values exceed the threshold, and generates a resistance sudden rise section mark according to the node interval coordinates; The preset resistance sudden rise threshold parameter is 3 to 5 times the pressure drop difference value under normal operating conditions, which is determined by statistically analyzing historical fault data.

[0071] The threshold discrimination sub-module receives the smoothed pressure drop difference sequence generated by the previous sub-module V. The system will iterate through each of the modified pressure drop difference values in this sequence and compare them one by one with the preset resistance sudden increase threshold .

[0072] This threshold is set based on the analysis of historical normal operating condition data. First, collect a large amount of data on the voltage differences between adjacent nodes when there is no confirmed fault, subtract the known baseline offset V, and calculate the average value of these normal pressure drop difference values . In this embodiment, the calculated value is V. Then, according to the sensitivity requirements of fault detection, set the threshold to times the normal average value, where is a multiplier factor, and its value range is usually set between 3 and 5. This range is selected based on experience and the analysis of historical fault cases, aiming to ensure that the threshold can effectively distinguish the significant increase in pressure drop caused by faults from normal measurement fluctuations. In this embodiment, is selected. Based on this, calculate the resistance sudden increase threshold: V.

[0073] Next, compare each value in the sequence with the threshold V: * Compare V: Since , this value does not exceed the threshold. * Compare V: Since , this value exceeds the threshold. The system marks the corresponding node interval 6 - 7 as having a resistance sudden increase anomaly. * Compare V: Since , this value also exceeds the threshold. The system marks the corresponding node interval 7 - 8 as having a resistance sudden increase anomaly.

[0074] According to the marking results, the system determines that the resistance sudden increase section is composed of the intervals 6 - 7 and 7 - 8. Extract the starting node (6) and the ending node (8) of this section to generate a resistance sudden increase section mark . This mark indicates that a voltage drop pattern indicating an abnormal increase in resistance has been detected in the line segment between node 6 and node 8.

[0075] Please refer to Figure 5 , the result verification module includes: The section receiving sub-module calls the starting node number and the ending node number of the candidate section coordinates, extracts the coordinate range of the resistance sudden increase section mark, combines the node number intervals and the marking status of the two into a unified data format, and generates a section set; The function of the section receiving sub-module is to integrate the results from different analysis phases. It calls the candidate section coordinates generated by the anomaly location module (specifically the candidate section calibrator sub-module). , which coordinates indicate the line range where anomalies are initially suspected, starting from node 5 and ending at node 8. At the same time, it calls the resistance sudden rise section marker generated by the current control module (specifically the threshold discrimination sub-module). , which marker indicates the specific section where abnormal resistance characteristics are confirmed through voltage drop analysis, starting from node 6 and ending at node 8.

[0076] The system extracts the coordinate information of these two sections: the candidate section covers nodes {5, 6, 7, 8}, and the resistance sudden rise section covers nodes {6, 7, 8}. Then, the types of these two sections (candidate, resistance sudden rise) and their corresponding node number ranges are combined into a structured data set for subsequent comparison. The generated data structure is the section set {Type: Candidate, StartNode: 5, EndNode: 8; Type: ResistanceRise, StartNode: 6, EndNode: 8}.

[0077] The overlap degree calculation sub-module compares the coordinate ranges of the starting point of the candidate section and the ending point of the resistance sudden rise section in the section set, calculates the length of the intersection interval on the node number axis, counts the percentage of the intersection node number in the total node number of the candidate section, and generates an overlap ratio value; The overlap degree calculation sub-module receives the section set {Candidate: [5, 8], ResistanceRise: [6, 8]} generated by the previous sub-module. Its core task is to quantify the coincidence degree of these two sections in space (node number axis).

[0078] The system compares the coordinate ranges of the candidate section (including nodes 5, 6, 7, 8) and the resistance sudden rise section (including nodes 6, 7, 8). Calculate the intersection of these two intervals. The intersection contains all nodes that exist in both sections, namely nodes {6, 7, 8}. Calculate the length of the intersection interval, that is, the number of nodes included in the intersection, which is 3 nodes here.

[0079] Then, calculate the percentage of the intersection node number in the total node number of the candidate section. The total node number of the candidate section is nodes. Calculate the overlap ratio value: . Finally, generate the overlap ratio value This value reflects the proportion of the abnormal area verified through the second stage (current injection and voltage drop analysis) in the candidate areas initially located in the first stage (phase analysis).

[0080] The verification decision sub-module compares the overlap ratio value with a preset ratio value threshold. If the ratio value exceeds the threshold, it outputs the positioning result to the external alarm unit. If it does not exceed, it generates a review instruction and sends it back to the phase monitoring module and the current control module; The ratio value threshold is 75% to 90%, which is optimized and determined by the cross-validation method, and is set to 80% when the positioning accuracy reaches a 95% confidence level.

[0081] The verification decision sub-module receives the overlap ratio value calculated and generated by the previous sub-module 。The system compares this value with a preset ratio value threshold to determine whether the confidence level of the current positioning result is high enough to be output as the final result.

[0082] This ratio value threshold is optimized through cross-validation (Cross-Validation) of a large number of historical positioning cases (including successful positioning and positioning deviation cases). The optimization goal is to set a threshold that can best distinguish high-confidence positioning from low-confidence positioning under the condition of requiring a specific confidence level (such as 95%). The reasonable range of the threshold is usually between 75% and 90%. In this embodiment, based on the goal of reaching a 95% confidence level, the threshold determined through cross-validation optimization is 。

[0083] Compare the calculated overlap ratio value with the set threshold 。The judgment condition is 。In this instance, , that is, the overlap ratio value does not exceed the threshold. Based on this, the system determines that the consistency of the analysis results in the current two stages does not reach the preset high-confidence standard, and the positioning result needs to be further confirmed. Therefore, the system does not directly output the fault positioning result, but generates a review instruction. This instruction contains the current candidate section and the section with sudden resistance increase and other diagnostic information, and sends this instruction back to the phase monitoring module and the current control module to initiate possible parameter adjustment, supplementary measurement, or re-analysis processes.

[0084] If in another scenario, the calculated overlap ratio value is 。Since , satisfying Condition. The system determines that the positioning result has a high credibility. At this time, the system will output the determined fault section information (usually the overlapping area , or the final section after fine-tuning in combination with the candidate section information) to an external alarm system, operation and maintenance platform or display interface for the operation personnel to refer to and handle.

[0085] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still belong to the protection scope of the technical solution of the present invention.

Claims

1. An electrical fault detection and control system, characterized in that: The system comprises: The phase monitoring module is used to obtain the voltage waveform sampling time point through the synchronous voltage acquisition device, call the dynamic time warping algorithm to calculate the phase difference between the sampling start time and the zero crossing time point and the standard node reference value, compare the phase delay parameter item with the phase tolerance value threshold, generate a phase anomaly mark, and pass the phase anomaly mark to the anomaly positioning module; An anomaly locating module is used to obtain a phase difference sequence of adjacent nodes based on the phase anomaly mark, perform a sudden jump improvement judgment on the phase difference change slope of three consecutive nodes through a linear regression model, generate candidate segment coordinates, and transmit the candidate segment coordinates to the current control module; The current control module is used to start the constant current source device according to the candidate segment coordinates, collect multi-node voltage difference data after constant current is injected, use the Kalman filter algorithm to smooth the voltage drop difference sequence between adjacent nodes, compare the voltage drop difference value with the set threshold parameter, generate a resistance sudden rise segment mark, and pass the resistance sudden rise segment mark to the result verification module.

2. The electrical fault detection and control system according to claim 1, characterized in that: The phase anomaly mark specifically includes the phase delay level, waveform offset characteristics, and time series label. The candidate segment coordinates include the node index interval, change rate identifier, and time label. The resistance surge segment mark specifically refers to the voltage drop change rate, target node index, and abnormal segment length.

3. The electrical fault detection and control system according to claim 2, characterized in that: The dynamic time warping algorithm uses a Gaussian window function to align the starting time to the reference time axis of the standard node reference value, the window width is 1 / 10 to 1 / 5 of the sampling period, and the weight factor is 0.6 to 0.8; The state transfer matrix of the Kalman filter algorithm is [1, Δt; 0, 1], the observation matrix is ​​[1, 0], the process noise covariance is 0.01 to 0.05, and the observation noise covariance is 0.1 to 0.

3.

4. The electrical fault detection and control system according to claim 3, characterized in that: The phase monitoring module comprises: The waveform sampling submodule monitors the voltage waveform through a synchronous voltage acquisition device, records the starting time coordinates within the sampling period, detects the timing position of the first zero crossing point, extracts the time interval data from the starting time to the zero crossing point, and generates a time point sequence; The phase difference calculation submodule calls the dynamic time warping algorithm to align the starting time of the time point sequence to the reference time axis of the standard node reference value, calculates the timing offset between the two in the zero crossing point interval, obtains the mean value of the offset as the waveform alignment deviation, and generates the phase difference parameter; The phase anomaly determination submodule compares the phase difference parameter with a preset phase tolerance threshold range to determine whether the offset mean exceeds the upper or lower threshold limit. If so, the binary mark variable is assigned a value of 1, otherwise it is assigned a value of 0, and a phase anomaly mark is generated; The phase tolerance threshold range is determined by optimizing historical fault data using a gradient descent method, with an upper limit of 5% to 8% of a sampling period and a lower limit of 2% to 4% of a sampling period.

5. The electrical fault detection and control system according to claim 4, characterized in that: The abnormality positioning module includes: The phase sequence sampling submodule screens the phase difference value set of adjacent nodes based on the phase anomaly mark, extracts the difference data of three consecutive nodes in a time sequence, and generates an adjacent phase difference sequence; The slope jump determination submodule calls the linear regression model, performs least squares fitting on the phase difference value of each node in the adjacent phase difference sequence, calculates the absolute value of the slope difference of the fitting straight line between adjacent segments, and compares it with the upper limit value of the preset jump determination threshold. If the absolute value of the slope difference exceeds the threshold, it is marked as a sudden change segment, and a slope jump coefficient is generated; The absolute value of the slope difference is dimensionlessly processed into a percentage form, and the sudden jump determination threshold is 15% to 25%; The candidate segment calibration submodule extracts the start and end node numbers corresponding to the mutation segment according to the slope jump coefficient, calculates the node spacing between adjacent mutation points, and if the spacing is less than the preset segment merging length, merges multiple mutation points into a single interval to generate candidate segment coordinates; The combined length of the preset sections is 1.5 to 2.5 times the electrical node spacing, and the node spacing is pre-calibrated to 0.5 to 1.2 meters based on the cable model.

6. The electrical fault detection and control system according to claim 5, characterized in that: The current control module comprises: The constant current injection control submodule collects the coordinates of the candidate section, analyzes the physical node distribution corresponding to the coordinates, sets the constant current source output parameters according to the node spacing, activates the constant current source device to inject a preset current value into the target section, synchronously measures the voltage of multiple nodes to the ground, calculates the voltage difference between adjacent nodes, and generates multi-node voltage difference data; The voltage difference abnormality analysis submodule calls the multi-node voltage difference data, extracts the voltage drop difference values ​​of adjacent nodes in node order to form a sequence, and adopts the formula: ； Iteratively calculate the sequence, eliminate noise interference by correcting the difference value item by item, and generate a smoothed pressure drop difference sequence; in, Representative The corrected pressure drop difference in the iteration, Represents the current node in The original value of the pressure drop difference collected in the iteration, Representative The difference in pressure drop between adjacent nodes in the iteration, Represents the suppression factor for the environmental noise amplitude, which is used to reduce the impact of noise on the difference value. Representative The amplitude of the ambient noise collected by the sensor in the iteration, Represents the dynamic sensitivity coefficient, which is used to adjust the response speed of the correction value. Represents the total number of data nodes participating in the calculation in the current segment. Represents the baseline voltage drop offset obtained through long-term monitoring data statistics, which is used to eliminate environmental steady-state interference; The threshold determination submodule calls the smoothed voltage drop difference sequence, traverses the corrected difference value of each node in the sequence, compares the value with the preset resistance sudden rise threshold parameter, marks the node interval whose difference value exceeds the threshold, and generates the resistance sudden rise section mark according to the node interval coordinates; The preset resistance sudden rise threshold parameter is 3 to 5 times of the voltage drop difference value under normal working conditions, and is determined by statistically analyzing historical fault data.

7. The electrical fault detection and control system according to claim 6, characterized in that: The system further comprises: The result verification module is used to receive the candidate segment coordinates and the resistance sudden rise segment mark, perform path coordinate comparison operation, output the positioning result when the overlap exceeds the ratio value threshold, trigger the review instruction and return it to the phase monitoring module and the current control module when it does not meet the requirements, and output the positioning result to the external alarm unit.

8. The electrical fault detection and control system according to claim 7, characterized in that: The positioning result includes the fault coordinate value, matching overlap, and positioning confidence level.

9. The electrical fault detection and control system according to claim 8, characterized in that: The result verification module comprises: The segment receiving submodule calls the starting node number and the ending node number of the candidate segment coordinates, extracts the coordinate range of the resistance sudden rise segment mark, merges the node number intervals and mark states of the two into a unified data format, and generates a segment set; The overlap calculation submodule compares the coordinate range of the starting point of the candidate segment in the segment set with the end point of the resistance sudden rise segment, calculates the length of the intersection interval between the two on the node number axis, counts the percentage of the number of intersection nodes to the total number of nodes in the candidate segment, and generates an overlap ratio value; The verification decision submodule compares the overlap ratio value with a preset ratio value threshold value, and outputs the positioning result to the external alarm unit if the ratio value exceeds the threshold value; if not, a review instruction is generated and sent back to the phase monitoring module and the current control module; The ratio value threshold is 75% to 90%, which is determined by cross-validation optimization and is set to 80% when the positioning accuracy reaches 95% confidence level.

Images