A method for accurately locating distribution network faults based on big data analysis

Abstract

The present invention discloses a method for accurately locating distribution network recording faults based on big data analysis, which relates to the technical field of distribution network fault locating. The method comprises the following steps: deploying a fault indicator; setting a trigger threshold, and triggering high-precision transient recording to generate recording data when a current mutation or an electric field mutation occurs; time-aligning the recording data and synchronously collected current intensity, electric field intensity, cable temperature, and environmental data, extracting transient features and environmental features after de-noising through wavelet transform, and combining the transient features with the environmental features to generate a comprehensive feature matrix; binding the comprehensive feature matrix to line topology information, and uploading the result to a monitoring center through encrypted communication; formulating dynamic threshold association rules to determine the fault type; the monitoring center matches the current transient waveform with a historical database to calculate the fault signal propagation time difference; calculating the fault distance in combination with the line topology information, and fusing the fault signal propagation time difference and the fault distance to obtain the fault coordinates.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault location, and in particular to a distribution network recording fault accurate location method based on big data analysis. Background Art

[0002] Distribution network fault location technology is a core link in ensuring the reliable operation of power systems. Early fault location mainly relies on steady-state electrical quantity analysis and manual inspections. However, due to the complex structure of the distribution network, numerous branches, and severe signal attenuation, the positioning accuracy is generally low. With the development of intelligent sensors and communication technologies, traveling wave positioning technology has gradually become mainstream due to its high precision. It captures the transient traveling wave signal generated by the fault and combines it with the principle of two-terminal ranging to achieve positioning, with the error controlled within 100 meters. In recent years, fault diagnosis methods based on transient characteristics have further improved positioning efficiency. For example, wavelet transform technology effectively filters out noise interference by decomposing the frequency domain characteristics of the signal.

[0003] Although existing traveling wave positioning technology performs well in transmission networks, it faces significant challenges in distribution networks. First, traveling wave signals attenuate rapidly in short distribution network lines and are easily interfered with by reflected waves from branch nodes, resulting in accumulated wave head detection errors. For example, high humidity environments can reduce the sensitivity of electric field sensors, requiring dynamic threshold correction, while traditional methods lack a real-time compensation mechanism. Second, existing fault indicators typically collect electrical quantities independently and do not synchronously record environmental parameters such as temperature and humidity, resulting in an incomplete feature matrix and difficulty in distinguishing similar faults. Finally, data from distribution network terminal devices is mostly transmitted through non-encrypted protocols, which poses a risk of tampering, and topology information updates lag, affecting the real-time performance of the impedance inversion algorithm. In addition, traditional threshold rules (such as the IEEE 519 harmonic standard) are designed based on static operating conditions and cannot adapt to the dynamic characteristics of transient processes. For example, a transient increase in motor starting current can trigger false recording waves. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a distribution network recording fault precise positioning method based on big data analysis to solve the problem that the traveling wave signal attenuates quickly in the short line of the distribution network and is easily interfered by the reflected wave of the branch node, resulting in the accumulation of wave head detection errors.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for accurately locating distribution network faults based on big data analysis, which includes:

[0008] Deploy fault indicators to monitor current and electric field strength in real time, set trigger thresholds, and trigger high-precision transient recording when current or electric field changes suddenly, generating recording data containing fault timestamps, transient waveforms, and trigger parameters.

[0009] Synchronously collect current intensity, electric field intensity, cable temperature, and environmental data. Time-align the recorded data and the synchronously collected current intensity, electric field intensity, cable temperature, and environmental data. After de-noising through wavelet transform, extract transient and environmental features. Combine the transient and environmental features to generate a comprehensive feature matrix.

[0010] Bind the comprehensive feature matrix and line topology information and upload them to the monitoring center via encrypted communication;

[0011] The monitoring center formulates dynamic threshold association rules based on the comprehensive feature matrix to determine the fault type;

[0012] The monitoring center matches the current transient waveform with the historical database, calculates the fault signal propagation time difference, combines the line topology information, calculates the fault distance through the impedance-distance inversion algorithm, and fuses the fault signal propagation time difference and the fault distance to obtain the fault coordinates.

[0013] As a preferred solution of the method for accurately locating distribution network faults based on big data analysis of the present invention, the specific steps of triggering high-precision transient recording when the current or electric field suddenly changes are as follows:

[0014] Install a dual-channel fault indicator at each branch node of the distribution network line, with built-in high-precision current sensor and electric field sensor;

[0015] Set the current mutation and electric field mutation thresholds and recording parameters;

[0016] When the change in current intensity exceeds the current mutation threshold, the recording is immediately started. When the change in current intensity does not exceed the current mutation threshold, and when the change in electric field intensity exceeds the electric field mutation threshold, the recording is also started. The current mutation is triggered first, and the electric field mutation is triggered second.

[0017] After triggering the recording, the fault timestamp is recorded, dual-channel synchronous recording is started, the current waveform and electric field waveform are saved, and the triggered current mutation threshold and electric field mutation threshold are stored at the same time.

[0018] As a preferred solution of the distribution network recording fault accurate positioning method based on big data analysis described in the present invention, wherein: the synchronous collection of current intensity, electric field intensity, cable temperature and environmental data, and the time alignment of the recording data and the synchronously collected current intensity, electric field intensity, cable temperature and environmental data are specifically as follows:

[0019] The fault indicator integrates current sensors, electric field sensors, cable temperature sensors, and ambient temperature and humidity sensors to collect information on current intensity, electric field intensity, cable temperature, ambient temperature, and ambient humidity.

[0020] When triggering the recording, the interpolation alignment method is used for time alignment.

[0021] As a preferred solution of the distribution network recording fault accurate positioning method based on big data analysis described in the present invention, wherein: after the denoising by wavelet transform, transient characteristics and environmental characteristics are extracted, and the transient characteristics and environmental characteristics are combined to generate a comprehensive feature matrix, the specific steps are as follows:

[0022] The Daubechies 4 wavelet basis is selected to perform wavelet decomposition on the current waveform and the electric field waveform, and the high-frequency noise is eliminated by the threshold filtering method;

[0023] Calculate the current change rate within a time window before and after the fault starting point to obtain the current mutation slope;

[0024] Perform Fourier transform on the denoised current waveform and calculate the proportion of high-frequency harmonic energy in the frequency band;

[0025] Calculate the cable temperature gradient before and after the fault;

[0026] Directly obtain the ambient temperature and humidity at the moment the fault is triggered;

[0027] The current mutation slope, high-frequency harmonic energy ratio, cable temperature gradient, ambient temperature and humidity at the moment of fault triggering are aligned according to the timestamps and merged into a matrix to obtain a comprehensive feature matrix.

[0028] As a preferred solution of the distribution network recording fault accurate positioning method based on big data analysis of the present invention, wherein: the comprehensive feature matrix and line topology information are bound and uploaded to the monitoring center through encrypted communication, specifically:

[0029] The electrical connection points in the distribution network are used as nodes to obtain line topology information from the distribution network GIS in real time;

[0030] The branch resistance and branch reactance are gradually added in branch order to generate a node impedance matrix;

[0031] The impedance matrix, node coordinates and hierarchical relationships are encoded into a topological matrix using topological information encoding.

[0032] Assign a unique device ID to each fault indicator, map the device ID in the comprehensive feature matrix to the nodes in the topology matrix, and merge the feature matrix data and topology information within the same time window to form a spatiotemporal correlation data packet.

[0033] The spatiotemporal correlation data packets are uploaded to the monitoring center via the MQTT protocol.

[0034] As a preferred solution of the distribution network recording fault accurate positioning method based on big data analysis of the present invention, the dynamic threshold association rule is used to determine the fault type, specifically:

[0035] Identify the fault type based on the current mutation slope, high-frequency harmonic energy ratio, cable temperature gradient, and ambient humidity;

[0036] Set the short-circuit fault threshold based on the short-circuit current instantaneous change rate requirements in the IEEE C37.118 standard;

[0037] Based on the total harmonic voltage distortion requirements in the IEEE 519-2022 standard, set normal operating thresholds, general fault thresholds, and lightning fault thresholds;

[0038] Based on the Joule heat formula in the IEC 60287-2-1 thermodynamic model, set the overload fault threshold and normal temperature rise threshold;

[0039] According to the dynamic combination of short circuit fault threshold, normal operating condition threshold, general fault threshold, lightning fault threshold, overload fault threshold and normal temperature rise threshold, the fault type of short circuit fault, small current grounding fault, lightning overvoltage fault and other faults can be determined.

[0040] As a preferred solution of the distribution network recording fault accurate positioning method based on big data analysis described in the present invention, the current transient waveform is matched with the historical database and the fault signal propagation time difference is calculated, specifically:

[0041] Classify historical fault waveforms by fault type, and integrate current waveforms, high-frequency harmonic energy distribution, and corresponding fault coordinates as transient waveform templates for storage;

[0042] Calculate the overlap of the current transient waveform template, select the template with high overlap as the candidate template, perform dynamic time warping on the candidate template, calculate the cumulative distance, set the normalized threshold for matching, and output the historical fault type and historical fault coordinates if the match is successful;

[0043] The fault signal propagation time difference is calculated based on the historical fault coordinates, the traveling wave propagation velocity and the distance difference between the fault point and the fault indicators at both ends.

[0044] As a preferred solution of the distribution network recording fault accurate positioning method based on big data analysis described in the present invention, the fault distance is calculated by the impedance-distance inversion algorithm, and the fault signal propagation time difference and the fault distance are integrated to obtain the fault coordinates, specifically:

[0045] Obtain real-time branch impedance from the distribution network GIS, generate a node impedance matrix using the branch addition method, set an abnormal area threshold, and mark a branch section as a candidate fault section when the branch impedance suddenly changes beyond the abnormal area threshold.

[0046] For the candidate fault section, the fault distance is calculated using the impedance-distance inversion formula;

[0047] The traveling wave propagation velocity and the fault signal propagation time difference are fused to obtain the final fault coordinates.

[0048] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for accurately locating distribution network recording faults based on big data analysis as described in the first aspect of the present invention is implemented.

[0049] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for accurately locating distribution network recording faults based on big data analysis as described in the first aspect of the present invention.

[0050] The beneficial effects of the present invention are as follows: the present invention deploys a dual-channel fault indicator to monitor the current intensity and electric field intensity in real time, sets an intelligent trigger threshold, and preferentially starts high-precision transient recording, thereby solving the defect that traditional detection cannot capture high-frequency transient signals and reducing the misjudgment rate. The low-frequency environmental data and the high-frequency recording time series are synchronized through the interpolation alignment method, and noise interference is eliminated by combining wavelet decomposition and threshold filtering. The current mutation slope and the proportion of high-frequency harmonic energy are extracted, a multi-dimensional feature matrix is ​​constructed, and the feature extraction accuracy is improved. The encrypted communication protocol is used to bind the GIS topology node impedance matrix, and the international standard threshold is dynamically associated to achieve accurate identification of the fault type. Through the rapid matching of historical waveform templates and the weighted fusion of traveling wave propagation and impedance inversion data, high-precision fault coordinates are output, which significantly improves the fault tracing efficiency of complex distribution networks and provides reliable protection for the stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 This is a flow chart of the method for accurately locating distribution network recording faults based on big data analysis in Example 1.

[0053] Figure 2Flowchart of feature extraction and data processing in Example 1.

[0054] Figure 3 This is a flowchart of fault type identification and location in Example 1. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0056] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0057] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0058] Example 1, reference Figures 1 to 3 , which is the first embodiment of the present invention, provides a method for accurately locating distribution network faults based on big data analysis, comprising the following steps:

[0059] S1. Deploy fault indicators to monitor current and electric field strength in real time. Set trigger thresholds. When there is a sudden change in current or electric field, high-precision transient recording is triggered to generate recording data containing fault timestamps, transient waveforms, and trigger parameters.

[0060] A dual-channel fault indicator is installed at each branch node of the distribution network line, with a built-in high-precision current sensor (accuracy ±0.5% FS) and electric field sensor (frequency response range 0.1Hz-10kHz);

[0061] Set the current and electric field sudden change thresholds and recording parameters. Specifically, the current sudden change threshold is set to ±20% of the line rated current (for example, when the rated current is 100A, the trigger threshold is 80A or 120A). According to the IEEE 1159 standard, short-circuit fault current usually exceeds 50% of the rated value. However, to avoid misjudgment during normal heavy-load startup (such as a 30% instantaneous increase in motor starting current), ±20% is selected as the balance point.

[0062] The electric field mutation threshold is set to ±15% of the rated phase electric field strength (for example, when the rated phase electric field strength is 10 kV / m, the trigger threshold is 8.5 kV / m or 11.5 kV / m). The phase electric field strength is directly related to the line voltage. During normal operation, the phase electric field strength fluctuation is typically less than ±5%. During faults (such as single-phase grounding faults), the electric field strength suddenly drops (E approaches 0 in the case of metallic grounding) or suddenly rises (such as lightning overvoltage). The ±15% electric field mutation threshold covers typical fault scenarios while also avoiding environmental interference (such as ±5% fluctuations caused by humidity changes).

[0063] The sampling frequency should be ≥10kHz to ensure that transient waveform details (such as high-frequency harmonics) are captured. The recording duration should be ≥200ms (covering the entire fault process). The sampling frequency of ≥10kHz is set based on the fact that the frequency range of transient fault signals (such as high-frequency harmonics and traveling waves) is typically 0.1kHz to 5kHz. According to the Nyquist sampling theorem, the sampling frequency must be at least twice the highest frequency of the signal (i.e., ≥10kHz) to fully capture waveform details. According to the CIGRE research report, a recording duration of 200ms can cover the complete transient process of more than 90% of faults.

[0064] When the change in current intensity exceeds the current mutation threshold, the recording is immediately started. When the change in current intensity does not exceed the current mutation threshold, and when the change in electric field intensity exceeds the electric field mutation threshold, the recording is also started. The current mutation is triggered first, and the electric field mutation is triggered second.

[0065] After triggering the recording, the fault timestamp is recorded, dual-channel synchronous recording is started, the current waveform and electric field waveform are saved, and the triggered current mutation threshold and electric field mutation threshold are stored at the same time.

[0066] S2. Synchronously collect phase current intensity, electric field intensity, cable temperature, and environmental data, time-align the recorded data and the synchronously collected current intensity, electric field intensity, cable temperature, and environmental data, perform wavelet transform de-noising, extract transient and environmental features, and combine the transient and environmental features to generate a comprehensive feature matrix.

[0067] Transient characteristics include current mutation slope and high-frequency harmonic energy ratio, and environmental characteristics include cable temperature gradient and ambient temperature and humidity;

[0068] Integrate current sensor, electric field sensor, cable temperature sensor and ambient temperature and humidity sensor into the fault indicator;

[0069] When the recording is triggered, the cable temperature, ambient temperature and ambient humidity are recorded synchronously;

[0070] The interpolation alignment method is used to time align non-waveform data (ambient temperature and humidity);

[0071] The ambient temperature and humidity sensors have a sampling frequency of 1 Hz. The cable temperature, ambient temperature, and ambient humidity are aligned to the recording timestamp (10 kHz sampling rate) through linear interpolation. All sensors use a unified timestamp using the GPS clock to ensure consistent sampling time references across devices. Specifically:

[0072] For each recorded timestamp (e.g., t=0ms, 0.1ms, 0.2ms, …, 200ms), the interpolation value of the current point is calculated based on the values ​​of two adjacent low-frequency data points (e.g., t=0s and t=1s) according to the time ratio;

[0073] The 1Hz low-frequency data is converted to a 10kHz sampling rate through interpolation (i.e., one interpolation point every 0.1ms) to generate a temperature and humidity sequence consistent with the time axis of the recorded data. The recorded data (current waveform, electric field waveform) itself has a 10kHz sampling rate, so the original timestamp is directly retained without interpolation processing. The recorded data, cable temperature, ambient temperature, and ambient humidity are then merged according to the timestamp;

[0074] Select Daubechies 4 wavelet basis (suitable for power transient signals) and perform 3-layer wavelet decomposition on the current waveform and electric field waveform;

[0075] Eliminate high-frequency noise through threshold filtering;

[0076] Calculate the current change rate within the 1ms window before and after the fault starting point to obtain the current mutation slope , the expression is:

[0077] ;

[0078] Among them, the current mutation slope The unit is ampere per millisecond (A / ms), which indicates the instantaneous rate of change of current when a fault occurs. The value range is 0~1000 A / ms. A typical short circuit fault ≥500A / ms, small current ground fault ≤200A / ms, Indicates the time point 1ms before the fault is triggered. Indicates the time point 1ms after triggering. Indicates the instantaneous current value 1ms after the fault is triggered, in amperes (A). Indicates the instantaneous current value 1ms before the fault is triggered, in amperes (A). The unit conversion factor is to convert milliseconds (ms) in the denominator to seconds (s) to ensure that the slope unit is A / ms instead of A / s. The current sudden change slope formula is improved based on the differential definition and refers to the IEEE C37.118 standard.

[0079] Perform Fourier transform on the denoised current waveform and calculate the proportion of high-frequency harmonic energy in the 2kHz-5kHz frequency band , the expression is:

[0080] ;

[0081] Among them, the high-frequency harmonic energy accounts for The unit is dimensionless percentage (%), which represents the ratio of the total energy between 2kHz and 5kHz, with a range of 0% to 100%. When ≤5%, it is normal working condition. When ≥15%, it is a fault. The Fourier transform result of the current intensity is expressed in amperes per hertz (A / Hz), which represents the energy distribution of the signal in the frequency domain. It represents the integral of the current intensity energy in the frequency band from 2kHz to 5kHz, with the unit being ampere square hertz (A²·Hz), reflecting the total energy of the high-frequency harmonic components. It represents the integral of the current intensity energy in the frequency band from 0Hz to 5kHz, in amperes squared hertz. Represents frequency differential variables. The formula for high-frequency harmonic energy is based on Fourier's energy conservation theorem and is cited from the IEEE Transactions on Power Delivery paper (DOI: 10.1109 / TPWRD.2015.2405933).

[0082] Environmental characteristics include cable temperature gradient and ambient temperature and humidity;

[0083] Calculate the cable temperature gradient within 1 minute before and after the fault , the expression is:

[0084] ;

[0085] in, Indicates the average cable temperature within 1 minute after the fault occurs, in degrees Celsius (℃). Indicates the average cable temperature within 1 minute before the fault occurs. express Second;

[0086] Directly obtain the ambient temperature and humidity at the moment the fault is triggered;

[0087] The current mutation slope, high-frequency harmonic energy ratio, cable temperature gradient, ambient temperature and humidity at the moment of fault triggering are aligned according to the timestamps and merged into a matrix to obtain a comprehensive feature matrix.

[0088] S3. Bind the comprehensive feature matrix and line topology information and upload them to the monitoring center via encrypted communication;

[0089] The electrical connection points in the distribution network are regarded as nodes, and line topology information is obtained from the distribution network GIS in real time, including node coordinates (latitude and longitude), branch impedance and hierarchical relationships. The hierarchical relationship includes root nodes, branch nodes and leaf nodes.

[0090] The root node refers to the starting point of the distribution network, the branch node refers to the intermediate node connecting two or more child nodes, and the leaf node refers to the terminal node without child nodes;

[0091] The branch resistance and branch reactance are gradually added in branch order to generate a node impedance matrix;

[0092] The impedance matrix, node coordinates and hierarchical relationships are encoded into a topological matrix using topological information encoding.

[0093] Assign a unique device ID to each fault indicator, map the device IDs in the comprehensive feature matrix to the nodes in the topology matrix, and merge the feature matrix data and topology information within the same time window (e.g., 200ms before and after the fault occurs) to form a spatiotemporal correlation data packet.

[0094] The spatiotemporal correlation data packets are uploaded to the monitoring center via the MQTT protocol.

[0095] S4. The monitoring center formulates dynamic threshold association rules based on the comprehensive feature matrix to determine the fault type;

[0096] After receiving the encrypted spatiotemporal correlation data packet, the monitoring center decrypts it using the TLS 1.3 protocol and verifies the integrity of the device ID data and the data signature. If the spatiotemporal correlation data packet is damaged or the signature is invalid, an alarm is triggered and the data is discarded.

[0097] Extract the current mutation slope, high-frequency harmonic energy ratio, cable temperature gradient and ambient humidity of the comprehensive feature matrix from the spatiotemporal correlation data packet;

[0098] Fault types are identified based on the current mutation slope, high-frequency harmonic energy ratio, cable temperature gradient, and ambient humidity. Fault types include short circuit faults, low-current ground faults, lightning overvoltage faults, and other faults.

[0099] Based on the requirements of the short-circuit current instantaneous change rate in the IEEE C37.118 standard, the short-circuit current is 3 to 5 times the rated value, and the short-circuit fault threshold is set;

[0100] Based on the IEEE 519-2022 standard requirement of total harmonic voltage distortion (THD) ≤ 5%, the normal operating threshold, general fault threshold, and lightning fault threshold are set.

[0101] Based on the Joule heat formula in the IEC 60287-2-1 thermodynamic model, the overload fault threshold and normal temperature rise threshold are set. The Joule heat formula is expressed as follows:

[0102] ;

[0103] in, Indicates the rate of change of temperature over time, in degrees Celsius per second ℃ / s, Indicates the current in amperes A. Indicates resistance in ohms. Indicates the specific heat capacity of the conductor, in joules per kilogram per degree Celsius (J / (kg·℃)). Indicates the mass of the conductor in kilograms (kg);

[0104] According to the sensor error compensation requirements in high humidity environments in IEC 62351-6, when H ≥ 85% RH, where RH represents relative humidity, the electric field mutation threshold needs to be dynamically adjusted by ±5%. The humidity correction trigger threshold and the electric field threshold relaxation amount are set. When the ambient humidity is ≥ 85% RH, the electric field threshold is relaxed by ±5% to avoid false positives as ground faults.

[0105] When the current mutation slope is ≥500A / ms, the high-frequency harmonic energy ratio is ≥15%, and the cable temperature gradient is ≥0.5℃ / s, the fault type is judged to be a short circuit fault;

[0106] When the current mutation slope is ≤200A / ms, the high-frequency harmonic energy ratio is ≥20%, and the ambient humidity is ≥70%RH, the fault type is determined to be a low-current grounding fault;

[0107] When the high-frequency harmonic energy ratio is ≥30%, the current mutation slope is ≥300A / ms, and the ambient humidity is ≥85%RH, the fault type is determined to be a lightning overvoltage fault;

[0108] When the fault type is not any of the short circuit fault, low current grounding fault and lightning overvoltage fault, it is judged as other faults, and the backup line or contact switch is immediately activated to restore power supply to the affected area.

[0109] S5. The monitoring center matches the current transient waveform with the historical database, calculates the fault signal propagation time difference, and combines it with the line topology information to calculate the fault distance using the impedance-distance inversion algorithm. The fault signal propagation time difference and the fault distance are then combined to obtain the fault coordinates.

[0110] Classify historical fault waveforms (current waveform and electric field waveform) by fault type, and integrate the current waveform, high-frequency harmonic energy distribution, and corresponding fault coordinates as a transient waveform template for storage;

[0111] A minimum bounding rectangle (MBR) is generated for each transient waveform template. Specifically, the rotating caliper algorithm is used to dynamically adjust the rotation angle of the rectangle boundary. The minimum area rectangle covering all convex hull vertices is calculated. The area of ​​the corresponding rectangle is recorded for each rotation angle. The rectangle with the smallest area is ultimately selected as the MBR. The waveform extreme points (maximum and minimum values) and energy envelope (integrated value in the 2kHz-5kHz frequency band) are recorded. The current current waveform and electric field waveform are compared with the historical templates one by one. The minimum bounding rectangle MBR is stored using a KD tree structure to reduce matching complexity.

[0112] The high-frequency harmonic energy ratio and current mutation slope are extracted from the current current waveform and electric field waveform as transient electrical feature vectors. Candidate templates are screened using the KD tree index. Starting from the root node of the KD tree, the overlap between the current waveform feature point and the MBR of each node is calculated.

[0113] Overlap degree = intersection area of ​​current waveform MBR area Overlap degree = intersection area of ​​current waveform MBR area;

[0114] If the overlap is ≤90%, the subtree is directly pruned and removed. Subtrees with an overlap greater than 90% are selected as candidate templates. Dynamic time warping is performed on the candidate templates, and the cumulative distance is calculated. If the minimum distance is ≤0.2 (normalization threshold), it is determined to be a successful match, and the historical fault type and historical fault coordinates are output.

[0115] Calculate the fault signal propagation time difference , the traveling wave time difference formula is:

[0116] ;

[0117] in, Indicates time, Indicates the distance difference between the fault point and the fault indicators at both ends. represents the speed of traveling wave propagation;

[0118] Obtain real-time branch impedance from the distribution network GIS, generate a node impedance matrix using the branch addition method, set an abnormal area threshold, and mark a candidate fault section when the branch impedance suddenly changes beyond the abnormal area threshold. Specifically:

[0119] Obtaining the baseline impedance of the line under normal operating conditions from the distribution network GIS The average value is obtained through statistics of historical operating data. According to IEEE 141-1993 and IEC 60909 standards, when the line impedance changes by more than ±15% to ±20% of the reference value, it is considered abnormal, thus obtaining the abnormal area threshold;

[0120] For the candidate fault section, the fault distance is calculated using the impedance-distance inversion formula. The impedance inversion formula is:

[0121] ;

[0122] in, Indicates the fault distance in kilometers (km), specifically the distance between the fault point and the nearest fault indicator. represents the measured impedance at the fault point, Indicates the baseline impedance, which corresponds to the impedance value of the line under normal working conditions and is obtained in real time by the distribution network GIS. Indicates length impedance, directly obtained through a loop resistance tester;

[0123] The final fault coordinates are obtained by weighting the traveling wave time difference formula (weight 70%) and the impedance inversion formula (weight 30%), and the expression is:

[0124] ;

[0125] in, Indicates the final fault distance in kilometers (km).

[0126] This embodiment also provides a computer device, which is suitable for the distribution network recording fault precise positioning method based on big data analysis, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the distribution network recording fault precise positioning method based on big data analysis proposed in the above embodiment.

[0127] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0128] This embodiment also provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the method for accurately locating distribution network faults based on big data analysis proposed in the above embodiment is implemented. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0129] In summary, the present invention solves the defect that traditional detection cannot capture high-frequency transient signals by deploying a dual-channel fault indicator to monitor current intensity and electric field intensity in real time, setting an intelligent trigger threshold, and prioritizing the start of high-precision transient recording, thereby reducing the misjudgment rate. The low-frequency environmental data and high-frequency recording time series are synchronized through the interpolation alignment method, and noise interference is eliminated by combining wavelet decomposition and threshold filtering. The current mutation slope and the proportion of high-frequency harmonic energy are extracted, a multi-dimensional feature matrix is ​​constructed, and the feature extraction accuracy is improved. The encrypted communication protocol is used to bind the GIS topology node impedance matrix, and the international standard threshold is dynamically associated to achieve accurate identification of the fault type. Through the rapid matching of historical waveform templates and the weighted fusion of traveling wave propagation and impedance inversion data, high-precision fault coordinates are output, which significantly improves the fault tracing efficiency of complex distribution networks and provides reliable protection for the stable operation of the power system.

[0130] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for accurately locating distribution network faults based on big data analysis, characterized by: include: Deploy fault indicators to monitor current and electric field strength in real time, set trigger thresholds, and trigger high-precision transient recording when current or electric field changes suddenly, generating recording data containing fault timestamps, transient waveforms, and trigger parameters. Synchronously collect current intensity, electric field intensity, cable temperature and environmental data, time-align the recorded data and the synchronously collected current intensity, electric field intensity, cable temperature and environmental data, perform wavelet transform denoising, extract transient characteristics and environmental characteristics, and combine the transient characteristics with the environmental characteristics to generate a comprehensive feature matrix; the specific steps of extracting transient characteristics and environmental characteristics after wavelet transform denoising, and combining the transient characteristics with the environmental characteristics to generate a comprehensive feature matrix are as follows: The Daubechies 4 wavelet basis is selected to perform wavelet decomposition on the current waveform and the electric field waveform, and the high-frequency noise is eliminated by the threshold filtering method. Calculate the current change rate within a time window before and after the fault starting point to obtain the current mutation slope; perform Fourier transform on the denoised current waveform to calculate the proportion of high-frequency harmonic energy in the frequency band; Calculate the cable temperature gradient before and after the fault; directly obtain the ambient temperature and humidity at the moment of fault triggering; align the current mutation slope, high-frequency harmonic energy ratio, cable temperature gradient, ambient temperature and humidity at the moment of fault triggering according to timestamps and merge them into a matrix to obtain a comprehensive feature matrix; Bind the comprehensive feature matrix and line topology information and upload them to the monitoring center via encrypted communication; The monitoring center formulates dynamic threshold association rules based on the comprehensive feature matrix to determine the fault type; The monitoring center matches the current transient waveform with the historical database, calculates the fault signal propagation time difference, combines the line topology information, calculates the fault distance through the impedance-distance inversion algorithm, and fuses the fault signal propagation time difference and the fault distance to obtain the fault coordinates.

2. The method for accurately locating distribution network faults based on big data analysis according to claim 1, characterized in that: The specific steps of triggering high-precision transient recording when the current or electric field suddenly changes are as follows: Install a dual-channel fault indicator at each branch node of the distribution network line, with built-in high-precision current sensor and electric field sensor; Set the current mutation and electric field mutation thresholds and recording parameters; When the change in current intensity exceeds the current mutation threshold, the recording is immediately started. When the change in current intensity does not exceed the current mutation threshold, and when the change in electric field intensity exceeds the electric field mutation threshold, the recording is also started. The current mutation is triggered first, and the electric field mutation is triggered second. After triggering the recording, the fault timestamp is recorded, dual-channel synchronous recording is started, the current waveform and electric field waveform are saved, and the triggered current mutation threshold and electric field mutation threshold are stored at the same time.

3. The method for accurately locating distribution network faults based on big data analysis according to claim 2, characterized in that: The synchronous collection of current intensity, electric field intensity, cable temperature and environmental data, and the time alignment of the recorded data and the synchronously collected current intensity, electric field intensity, cable temperature and environmental data are specifically as follows: The fault indicator integrates current sensors, electric field sensors, cable temperature sensors, and ambient temperature and humidity sensors to collect information on current intensity, electric field intensity, cable temperature, ambient temperature, and ambient humidity. When triggering the recording, the interpolation alignment method is used for time alignment.

4. The method for accurately locating distribution network faults based on big data analysis according to claim 3, characterized in that: The comprehensive feature matrix and line topology information are bound together and uploaded to the monitoring center via encrypted communication, specifically: The electrical connection points in the distribution network are used as nodes to obtain line topology information from the distribution network GIS in real time; The branch resistance and branch reactance are gradually added in branch order to generate a node impedance matrix; The impedance matrix, node coordinates and hierarchical relationships are encoded into a topological matrix using topological information encoding. Assign a unique device ID to each fault indicator, map the device ID in the comprehensive feature matrix to the nodes in the topology matrix one by one, and merge the feature matrix data and topology information in the same time window to form a spatiotemporal correlation data packet; The spatiotemporal correlation data packets are uploaded to the monitoring center via the MQTT protocol.

5. The method for accurately locating distribution network faults based on big data analysis according to claim 4, characterized in that: The dynamic threshold association rule is used to determine the fault type, specifically: Identify the fault type based on the current mutation slope, high-frequency harmonic energy ratio, cable temperature gradient, and ambient humidity; Set the short-circuit fault threshold based on the short-circuit current instantaneous change rate requirements in the IEEE C37.118 standard; Based on the total harmonic voltage distortion requirements in the IEEE 519-2022 standard, set normal operating thresholds, general fault thresholds, and lightning fault thresholds; Based on the Joule heat formula in the IEC 60287-2-1 thermodynamic model, set the overload fault threshold and normal temperature rise threshold; According to the dynamic combination of short circuit fault threshold, normal operating condition threshold, general fault threshold, lightning fault threshold, overload fault threshold and normal temperature rise threshold, the fault type of short circuit fault, small current grounding fault, lightning overvoltage fault and other faults can be determined.

6. The method for accurately locating distribution network faults based on big data analysis according to claim 5, characterized in that: The current transient waveform is matched with the historical database to calculate the fault signal propagation time difference, specifically: Classify historical fault waveforms by fault type, and integrate current waveforms, high-frequency harmonic energy distribution, and corresponding fault coordinates as transient waveform templates for storage; Calculate the overlap of the current transient waveform template, select the template with high overlap as the candidate template, perform dynamic time warping on the candidate template, calculate the cumulative distance, set the normalized threshold for matching, and output the historical fault type and historical fault coordinates if the match is successful; The fault signal propagation time difference is calculated based on the historical fault coordinates, the traveling wave propagation speed and the distance difference between the fault point and the fault indicators at both ends.

7. The method for accurately locating distribution network faults based on big data analysis according to claim 6, characterized in that: The fault distance is calculated by the impedance-distance inversion algorithm, and the fault signal propagation time difference and the fault distance are integrated to obtain the fault coordinates, specifically: Obtain real-time branch impedance from the distribution network GIS, generate a node impedance matrix using the branch addition method, set an abnormal area threshold, and mark a branch section as a candidate fault section when the branch impedance suddenly changes beyond the abnormal area threshold. For the candidate fault section, the fault distance is calculated using the impedance-distance inversion formula; The traveling wave propagation velocity and the fault signal propagation time difference are fused to obtain the final fault coordinates.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for accurately locating distribution network recording faults based on big data analysis according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for accurately locating distribution network recording faults based on big data analysis according to any one of claims 1 to 7 are implemented.

Images