LoRa Communication System for Fault Detection in Power Distribution Network Lines and Its Usage

Abstract

This invention belongs to the field of wireless communication network technology, specifically disclosing a LoRa communication system and its usage method for fault detection in power distribution lines. The system includes a distributed sensor network, which deploys distributed sensors at key nodes of the power distribution line. For each node, it continuously monitors changes in voltage, current, and temperature within the monitoring range using high voltage acquisition accuracy and a specific current signal sampling rate. It also incorporates environmental interference filtering technology to remove noise influence and generates raw monitoring data sets according to a preset data acquisition frequency. A data preprocessing module is used to standardize the acquired signals based on the raw monitoring data sets. The purpose of this invention is to solve the logical problems arising from the combination of low monitoring accuracy, high false alarm rate, unstable data transmission, and slow fault isolation response in traditional power distribution systems.

Description

Technical Field

[0001] This invention relates to the field of wireless communication network technology, specifically to a LoRa communication system for detecting faults in power distribution network lines and its usage method. Background Technology

[0002] In modern power systems, the distribution network, as the core link in power transmission and distribution, plays a crucial role in ensuring the stable operation of electricity for social production and daily life. Especially in complex urban and rural power grid environments, the timely detection and location of line faults directly affects the reliability and security of power supply. However, given the wide coverage, numerous nodes, and complex environment of distribution networks, achieving rapid and accurate fault detection has become a pressing challenge for the power industry.

[0003] Currently, although some traditional fault detection methods can identify line problems to a certain extent, these methods often rely on fixed monitoring points and limited coverage, making them difficult to adapt to dynamically changing power grid environments. This is especially true when facing sudden or hidden faults, where detection blind spots and response delays are common. These limitations make it difficult to guarantee the comprehensiveness and real-time nature of fault detection, particularly in remote areas or complex terrain conditions.

[0004] Against this backdrop, the technical challenges of distribution network fault detection are becoming increasingly apparent. A core issue lies in capturing subtle changes in the operating status of power lines, as certain key parameters fluctuate abnormally when a fault occurs. Without real-time detection of these changes, it's difficult to take effective measures in the early stages of a fault. A deeper problem is that even if these changes are captured, achieving low-cost, long-distance data transmission across a widely distributed power grid presents a significant technical challenge. These two issues are closely related: sensing parameter changes requires high-precision monitoring methods, while data transmission must overcome geographical and environmental limitations. For example, on a distribution line traversing mountainous areas, an aging node may cause abnormal parameters, but due to the vast distance and insufficient signal coverage, monitoring data cannot be delivered to the control center in a timely manner, ultimately leading to the escalation of the fault.

[0005] Therefore, achieving real-time and accurate monitoring of key parameters in the power distribution network and transmitting the data to the management terminal through efficient communication methods has become a crucial issue in improving fault detection capabilities. Solving this problem not only involves technological breakthroughs but also directly impacts the overall stability of the power system and the user experience. Summary of the Invention

[0006] This invention provides a LoRa communication system and its usage method for fault detection in power distribution networks. The purpose is to solve the logical correlation problems caused by the combination of low monitoring accuracy, high false alarm rate, unstable data transmission, and slow fault isolation response in traditional power distribution systems.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0008] A LoRa communication system for power distribution line fault detection includes: a distributed sensor network, which deploys distributed sensors at key nodes of the power distribution line to continuously monitor changes in voltage, current, and temperature within the monitoring range of each node with high voltage acquisition accuracy and a specific current signal sampling rate; it combines environmental interference filtering technology to remove noise influence and generates raw monitoring data sets according to a preset data acquisition frequency; a data preprocessing module, which standardizes the acquired signals based on the raw monitoring data sets, extracts key indicators after feature dimension selection for voltage and current fluctuations, and determines preliminary abnormal signals when the processed indicators deviate from the abnormal threshold set; an abnormal signal optimization module, which refines and stratifies the abnormal data using classification boundary optimization technology for preliminary abnormal signals, and filters out high-confidence abnormal events with a false alarm rate control mechanism to obtain a precisely located abnormal signal set; and a report generation module, which generates reports from the precise data. In the set of abnormal signals located, the abnormal events are associated with specific line nodes according to the node communication protocol, generating an abnormal report data packet containing location information. The transmission management module is used to ensure the stability of sensor nodes in long-distance transmission through power management technology, converting the abnormal report data packet into a transmission sequence through a specific encoding method, and adjusting the transmission route through a backup communication path when signal interruption or delay is detected during transmission to determine the optimized transmission channel. The data verification module is used to obtain complete abnormal report data arriving at the control center according to the optimized transmission channel, perform integrity verification on the data packet, and trigger a retransmission mechanism when data is missing during verification to obtain the final confirmed fault location information. The isolation command execution module is used to generate a targeted isolation command sequence for the final confirmed fault location information, and ensures that the command is quickly sent to the field equipment through real-time classification speed to execute the automatic isolation operation of the abnormal node on the line and obtain the final fault isolation result.

[0009] In another aspect, this disclosure also relates to a method for using a LoRa communication system for fault detection in power distribution lines, comprising: deploying a distributed sensor layout at key nodes of the power distribution line; continuously monitoring changes in voltage, current, and temperature within the monitoring range of each node using high voltage acquisition accuracy and a specific current signal sampling rate; removing noise influence by combining environmental interference filtering technology; generating a raw monitoring data set according to a preset data acquisition frequency; standardizing the acquired signals using a data preprocessing process based on the raw monitoring data set; extracting key indicators after feature dimension selection for voltage and current fluctuations; if the processed indicators deviate from the abnormal threshold set range, they are determined to be preliminary abnormal signals; for the preliminary abnormal signals, refining and stratifying the abnormal data using classification boundary optimization technology; and filtering out high-confidence abnormal events by combining a false alarm rate control mechanism to obtain accurately located abnormal events. An abnormal signal set is generated. From the precisely located abnormal signal set, the abnormal events are associated with specific line nodes according to the node communication protocol, generating an abnormal report data packet containing location information. Power management technology ensures the stability of sensor nodes during long-distance transmission. The abnormal report data packet is converted into a transmission sequence through a specific encoding method. If signal interruption or delay is detected during transmission, the transmission route is adjusted through a backup communication path to determine the optimized transmission channel. Complete abnormal report data arriving at the control center is obtained based on the optimized transmission channel. The integrity of the data packet is verified. If data is found to be missing, a retransmission mechanism is triggered to obtain the final confirmed fault location information. A targeted isolation command sequence is generated based on the final confirmed fault location information. Real-time classification speed ensures that the commands are quickly sent to the field equipment to execute the automatic isolation operation of the abnormal line node and obtain the final fault isolation result.

[0010] In one aspect of this disclosure, by deploying a distributed sensor layout at key nodes of the power distribution line, continuously monitoring changes in voltage, current, and temperature within the monitoring range of each node with high voltage acquisition accuracy and a specific current signal sampling rate, and combining environmental interference filtering technology to remove noise influence, the original monitoring data set is generated according to a preset data acquisition frequency, including:

[0011] By acquiring voltage, current, and temperature change data from distributed sensors at key nodes of power distribution lines, and using a pre-set processing module to preliminarily organize the acquired signals, a structured monitoring data set is obtained.

[0012] Based on the structured monitoring data set, environmental interference filtering technology is used to remove noise from the data, generating a denoised monitoring data set.

[0013] In one aspect of this disclosure, the acquired signals are standardized using a data preprocessing procedure based on the original monitoring data set. Key indicators selected based on feature dimensions are extracted to address voltage and current fluctuations. If the processed indicators deviate from the abnormal threshold range, they are determined to be preliminary abnormal signals, including:

[0014] Raw data is obtained from monitoring equipment. The collected signal data is then preliminarily cleaned to remove invalid values ​​and noise interference, resulting in a cleaned signal dataset.

[0015] For the cleaned signal dataset, a standardization operation is used to convert voltage and current data of different dimensions into a unified scale, resulting in a standardized signal matrix;

[0016] From the standardized signal matrix, characteristic indicators of voltage and current fluctuations are extracted, and key dimension values ​​related to fluctuations are determined by calculating the statistical parameters of the time series.

[0017] Based on the key dimension values, anomaly detection rules are constructed. If the calculated dimension values ​​exceed the preset anomaly threshold range, it is determined as a preliminary anomaly signal and an anomaly marker is output.

[0018] In one aspect of this disclosure, for the initial anomalous signals, the anomalous data is refined and stratified using classification boundary optimization technology, and a false alarm rate control mechanism is combined to filter out high-confidence anomalous events, thereby obtaining a precisely located set of anomalous signals, including:

[0019] Preliminary abnormal signals are obtained from the raw data, and a pre-established classification model is used to perform preliminary segmentation of the signals. Abnormal data parts are extracted from the segmentation results to obtain a preliminary abnormal data set.

[0020] For the initial abnormal data set, the data is stratified by the classification boundary optimization method to obtain abnormal data groups at different levels and determine the abnormal data structure after stratification.

[0021] For the hierarchical abnormal data structure, a false alarm control strategy is used to filter the data. If the abnormal data of a certain level is lower than the preset confidence threshold, the data of that level is removed to obtain a subset of abnormal data with high confidence.

[0022] For a subset of anomalous data with high confidence, an event correlation analysis is performed on each set of data through a screening strategy. If the correlation of events in a certain data set is higher than the preset standard, the data set is retained, and the set of anomalous events that meet the conditions is identified.

[0023] For a set of anomalous events that meet the criteria, location mapping is performed based on the results of classification boundary optimization to obtain the specific location information of each anomalous event and determine a list of anomalous events with precise location.

[0024] For the precisely located list of abnormal events, all relevant data are integrated to form a unified signal set. Data verification tools are used to perform consistency checks on the signal set to obtain the final abnormal signal set.

[0025] In one aspect of this disclosure, the step of associating abnormal events with specific line nodes from a precisely located set of abnormal signals, according to a node communication protocol, to generate an abnormal report data packet containing location information, includes:

[0026] Abnormal signal data is obtained from the stored signal set. Each group of signal data is initially screened to determine whether there are abnormal fluctuations. If abnormal fluctuations are detected, it is identified as an abnormal signal group to be processed.

[0027] For a given abnormal signal group, a preset node communication rule is used to parse the signal source through the communication protocol and obtain the corresponding line node identifier.

[0028] Based on the line node identifier, obtain the associated geographical location information, and combine it with the characteristics of the abnormal signal group to generate an abnormal event record containing location information;

[0029] By classifying and processing abnormal event records, and binding abnormal events to line nodes according to the format requirements of communication protocols, structured abnormal data units are obtained.

[0030] For structured abnormal data units, a unified data encapsulation method is adopted to generate abnormal data packets that conform to the protocol specifications.

[0031] The generated abnormal data packets are forwarded to the target system through a preset transmission channel, thus completing the transmission and storage of abnormal information.

[0032] In one aspect of this disclosure, the method of ensuring the stability of sensor nodes during long-distance transmission through power management technology involves converting abnormal report data packets into a transmission sequence using a specific encoding method. If signal interruption or delay is detected during transmission, the transmission route is adjusted through a backup communication path to determine an optimized transmission channel, including:

[0033] The power management module is pre-established to monitor the energy consumption of the sensor nodes in real time, and the power supply mode is adjusted by a dynamic allocation strategy. The power supply status data of the current node is obtained to determine whether it meets the energy requirements for long-distance transmission.

[0034] Based on the remaining power status data, if the energy reserve is detected to be lower than the preset threshold, the energy-saving mode is triggered, non-essential data processing tasks are restricted, an adjusted energy allocation scheme is obtained, and it is determined whether the node can maintain stable operation.

[0035] For stable sensor nodes, acquire anomaly report data packets, convert them into transmission sequences using specific encoding rules, complete the structured processing of data packets, and determine whether the encoded sequence conforms to the transmission protocol requirements;

[0036] The communication channel status is monitored in real time during long-distance transmission using the encoded transmission sequence. If signal interruption or transmission delay exceeds a preset range, the current channel status information is recorded to obtain the abnormal transmission judgment result.

[0037] Based on the abnormal transmission determination result, switch to the pre-configured backup path, recalculate the routing scheme, obtain new transmission channel parameters, and determine whether the normal transmission rate can be restored.

[0038] For the new transmission channel parameters, a routing adjustment operation is performed to optimize the bandwidth allocation of the communication path, obtain the final optimized channel configuration, and determine whether the stability of long-distance transmission is maintained.

[0039] Through the optimized channel configuration, the operating status of sensor nodes and the integrity of transmission sequences are continuously monitored. If a new abnormal report data packet is detected, the above encoding and path adjustment process is executed repeatedly to obtain the latest transmission status information and determine whether the overall system operation meets expectations.

[0040] In one aspect of this disclosure, the step of obtaining complete anomaly report data arriving at the control center through the optimized transmission channel, performing integrity verification on the data packets, and triggering a retransmission mechanism if data is found to be missing, to obtain finally confirmed fault location information, includes:

[0041] The abnormal report data is obtained through the transmission channel, and the received data packets are initially parsed to obtain the initial data structure content;

[0042] Based on the initial data structure content, a verification mechanism is used to check the integrity of the data packet. If data is missing, the identification information of the missing part is recorded to determine the range of missing data.

[0043] For the range of missing data, a retransmission mechanism is initiated to retrieve the corresponding data packet content from the transmission channel, resulting in a supplemented set of data packets.

[0044] The supplemented data packet set is used to perform a secondary verification mechanism to determine whether the data integrity meets the preset standard. If there are still missing data, the retransmission mechanism is repeated until the complete data is obtained.

[0045] Based on the complete data, analyze the specific content of the anomaly report, extract information fields related to fault location, and determine the key description of the fault.

[0046] After obtaining the key description of the fault, it is compared and analyzed with the fault database pre-established by the control center. The support vector machine algorithm is used to classify the fault type and obtain the final fault location result.

[0047] Based on the final fault location results, an information confirmation record is generated and transmitted back to the control center through the optimized transmission path to complete the data processing flow.

[0048] In one aspect of this disclosure, the generation of a targeted isolation command sequence based on the finally confirmed fault location information, ensuring rapid delivery of commands to field equipment through real-time classification speed, and executing automatic isolation operations for abnormal line nodes to obtain the final fault isolation result, includes:

[0049] By using fault location data, detailed information about abnormal nodes can be obtained to determine the specific location of the line abnormality.

[0050] Based on the location of the line anomaly, generate corresponding isolation instructions and construct an instruction sequence to cover all relevant anomaly nodes;

[0051] Real-time classification technology is used to prioritize the instruction sequence and determine the urgency of the instruction. If the urgency is higher than a preset threshold, it is immediately transmitted to the field equipment.

[0052] Upon receiving isolation commands from field devices, perform automatic isolation operations and obtain isolation status information for abnormal nodes;

[0053] Analyze the completion status of fault isolation from the isolation status information. If there are any abnormal nodes that have not been isolated, generate a supplementary isolation command and send it to the field equipment.

[0054] The final fault isolation result is determined by the isolation operation data fed back by the field equipment, and the processing status of all abnormal nodes is recorded.

[0055] After obtaining the final fault isolation results, update the monitoring data of the abnormal line to form a complete isolation result archive.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] This invention utilizes distributed sensors deployed at key nodes for high-precision voltage, current, and temperature monitoring. Combined with environmental interference filtering and data preprocessing to extract key features, it identifies initial anomalies and then refines the screening of high-confidence events using classification boundary optimization and false alarm rate control mechanisms. This generates location-associated anomaly report data packets, which are then managed for power supply continuity and backup path adjustment to ensure stable transmission. Upon arrival at the control center, integrity verification is performed, and retransmission is triggered. Finally, an isolation command sequence is issued to achieve rapid isolation of field equipment. This method significantly improves fault location accuracy and response speed, reduces false alarm interference, optimizes transmission reliability, and ultimately achieves efficient fault isolation of power distribution lines, enhancing the resilience and operational efficiency of the power grid. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0059] Figure 1 This is a flowchart illustrating the method of using the LoRa communication system for fault detection in power distribution networks according to the present invention. Detailed Implementation

[0060] The present invention will be further described below with reference to embodiments. These embodiments are merely some, not all, of the embodiments of the present invention. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the protection scope of the present invention.

[0061] Please see Figure 1 As shown in the figure, this embodiment discloses a method for using a LoRa communication system for fault detection in power distribution networks. The specific steps are as follows:

[0062] S1. By deploying distributed sensors at key nodes of power distribution lines, the voltage, current and temperature of the lines are continuously monitored within the monitoring range with high voltage acquisition accuracy and specific current signal sampling rate. Combined with environmental interference filtering technology to remove noise influence, the original monitoring data sets are generated according to the preset data acquisition frequency.

[0063] S1-1. By acquiring voltage, current and temperature change data collected by distributed sensors at key nodes of the power distribution line, the collected signals are preliminarily processed using a preset processing module to obtain a structured monitoring data set.

[0064] S1-2. Based on the structured monitoring data set, environmental interference filtering technology is used to remove noise from the data, generating a noise-reduced monitoring data set.

[0065] In some embodiments, S1 further includes preliminary screening of the raw detection data set, with the following specific steps:

[0066] S1-3. For the denoised monitoring data set, if the voltage or current signal exceeds the preset threshold range, the abnormal data is marked by the pre-established anomaly detection mechanism to determine the location information of the abnormal node.

[0067] S1-4. After obtaining the location information of the abnormal node, cross-validate it with the temperature change data. If the temperature data also shows abnormality, the data comparison module will determine that the node has a potential fault risk.

[0068] S1-5. Extract key features from the node data of potential fault risks, and use the support vector machine algorithm to classify the fault types to obtain specific fault category information.

[0069] S1-6. Based on the fault category information, generate a targeted fault warning signal and send the warning signal to the relevant processing unit through a preset transmission channel to complete the automated response process.

[0070] For example, when deploying distributed sensors at key nodes of power distribution lines, smart sensors can be installed at the substation output, branch line connection points, and areas with concentrated loads. A monitoring point can be set up every 500 meters to ensure coverage of more than 95%. The sensors use low-power devices that support 5G communication to transmit data to the cloud platform in real time.

[0071] Regarding the monitoring parameter settings for each node, the voltage acquisition accuracy can reach ±0.1V, the current signal sampling rate is set to 1000 times per second, and the temperature monitoring range covers -40℃ to 85℃. The built-in analog-to-digital converter converts analog signals into digital signals and uploads data at a frequency of once per minute.

[0072] To continuously monitor the line status, the system uses time-series analysis algorithms to predict the trends of voltage, current, and temperature data. For example, it calculates the average voltage fluctuation over the past 10 minutes using the sliding window method. If the fluctuation exceeds ±5V, an abnormal alarm is triggered. At the same time, it combines Fourier transform to extract harmonic components from the current signal and analyzes whether the harmonic distortion rate exceeds 3% to determine the health status of the line load.

[0073] In terms of environmental interference filtering, wavelet transform technology is used to decompose the original signal into multiple resolutions to remove low-frequency noise below 0.5Hz and high-frequency interference above 10kHz, ensuring that the data signal-to-noise ratio is improved to more than 20dB.

[0074] The data acquisition frequency is preset to generate a set of raw monitoring data every 5 seconds. Each set of data contains three-dimensional parameters of voltage, current and temperature. The data is stored in association with the timestamp and node location. Principal component analysis is used to reduce the dimensionality of the data and extract the main feature values ​​to reduce storage pressure while retaining more than 90% of the information.

[0075] These data can then be compared and analyzed with historical fault records. If a voltage drop of more than 10% is found and lasts for more than 2 seconds, the system will automatically associate it with a possible short-circuit fault model and generate a risk assessment report to provide a basis for subsequent operation and maintenance decisions.

[0076] All of the above steps are seamlessly integrated through automated algorithms and cloud computing platforms to ensure the efficiency and accuracy of the monitoring and analysis process.

[0077] S2. Based on the original monitoring data set, the collected signals are standardized using a data preprocessing process. Key indicators after feature dimension selection are extracted for voltage and current fluctuations. If the processed indicators deviate from the abnormal threshold set range, they are judged as preliminary abnormal signals.

[0078] S2-1. Obtain raw data from the monitoring equipment, perform preliminary cleaning on the collected signal data to remove invalid values ​​and noise interference, and obtain the cleaned signal dataset.

[0079] S2-2. For the cleaned signal dataset, a standardization operation is used to convert voltage and current data of different dimensions into a unified scale to obtain a standardized signal matrix.

[0080] S2-3. Extract the characteristic indicators of voltage fluctuation and current fluctuation from the standardized signal matrix, and determine the key dimension values ​​related to fluctuation by calculating the statistical parameters of the time series.

[0081] S2-4. Based on the key dimension values, construct anomaly detection rules. If the calculated dimension values ​​exceed the preset anomaly threshold range, they are determined to be preliminary anomaly signals, and anomaly markers are output.

[0082] S2-5. For the signal with the output anomaly marker, obtain its corresponding time window data, use the support vector machine model to classify the anomaly signal, and determine the category of the anomaly signal.

[0083] S2-6. By classifying the abnormal signal categories and combining them with the fluctuation records of historical monitoring signals, determine the persistence and repetitiveness of the abnormal signals, and output the final anomaly judgment result.

[0084] For example, in the process of standardizing and detecting anomalies in the acquired signals, the original monitoring data set is first preprocessed using a standardization method. For example, the voltage signal data is [220.5, 218.3, 223.1, 219.7, 225.0] volts, and the current signal data is [5.2, 5.0, 5.4, 5.1, 5.3] amperes. Specifically, the Z-score standardization algorithm is used, which calculates the deviation of each data point from the mean and divides it by the standard deviation. The formula is Z=(X-μ) / σ, where μ is the mean and σ is the standard deviation.

[0085] Taking voltage data as an example, the mean is 221.32, the standard deviation is 2.47, and the standardized result after calculation is [-0.33, -1.23, 0.72, -0.65, 1.49]. The mean of current data is 5.2, the standard deviation is 0.16, and the standardized result is [0.0, -1.25, 1.25, -0.63, 0.63].

[0086] Next, feature dimensions were extracted for voltage and current fluctuations, and key indicators such as volatility and peak deviation were selected. The volatility formula is the ratio of standard deviation to mean. The voltage volatility is 2.47 / 221.32=0.011, and the current volatility is 0.16 / 5.2=0.031. At the same time, the peak deviation was calculated, which is the difference between the maximum value and the mean divided by the mean. The voltage peak deviation is (225.0-221.32) / 221.32=0.017, and the current peak deviation is (5.4-5.2) / 5.2=0.038.

[0087] Next, the processed indicators are compared with the abnormal threshold range. The voltage fluctuation rate threshold is set to 0.015, the current fluctuation rate threshold is set to 0.035, the voltage peak deviation threshold is set to 0.02, and the current peak deviation threshold is set to 0.04. If the values ​​exceed the range, they are judged as preliminary abnormal signals.

[0088] The analysis results show that the voltage fluctuation rate (0.011) is less than 0.015 and the peak deviation (0.017) is less than 0.02, both of which are normal.

[0089] The current fluctuation rate of 0.031 is less than 0.035, which is normal. However, the peak deviation of 0.038 is close to 0.04, which poses a potential risk. The system automatically marks it as a signal that needs attention and compares the result with historical data. If the peak deviation is close to the threshold for three consecutive cycles, it is upgraded to a preliminary abnormal signal, triggering the subsequent in-depth analysis module to ensure that no abnormal signal is missed, forming a complete logical chain from data processing to anomaly judgment.

[0090] S3. For the initial abnormal signals, the abnormal data is refined and layered using the classification boundary optimization technique. Combined with the false alarm rate control mechanism, high-confidence abnormal events are selected to obtain a set of precisely located abnormal signals.

[0091] S3-1. Obtain preliminary abnormal signals from the raw data, use a pre-established classification model to perform preliminary division of the signals, extract the abnormal data part from the division results, and obtain a preliminary abnormal data set.

[0092] S3-2. For the initial abnormal data set, the data is stratified by the classification boundary optimization method to obtain abnormal data groups at different levels and determine the abnormal data structure after stratification.

[0093] S3-3. For the hierarchical abnormal data structure, a false alarm control strategy is adopted to filter the data. If the abnormal data of a certain level is lower than the preset confidence threshold, the data of that level is removed to obtain a subset of abnormal data with high confidence.

[0094] S3-4. For the subset of anomalous data with high confidence, an event correlation analysis is performed on each data set through a screening strategy. If the correlation of events in a certain data set is higher than the preset standard, the data set is retained, and the set of anomalous events that meet the conditions is identified.

[0095] S3-5. For the set of abnormal events that meet the conditions, perform location mapping based on the results of the classification boundary optimization, obtain the specific location information of each abnormal event, and determine the list of abnormal events with precise location.

[0096] S3-6. For the precisely located list of abnormal events, integrate all relevant data to form a unified signal set, and use a data verification tool to perform consistency checks on the signal set to obtain the final abnormal signal set.

[0097] For example, in processing the initial abnormal signals, the abnormal data is first refined and layered using the classification boundary optimization technique. It is assumed that the initial abnormal signal dataset contains 1,000 data points, of which 200 are initially marked as abnormal.

[0098] Using the Support Vector Machine (SVM) algorithm, by setting the kernel function to the radial basis function (RBF) and adjusting the penalty parameter C to 10.0 and the kernel parameter γ to 0.1, the classification boundary was optimized, further dividing the 200 outliers into high-risk outliers (60%, i.e., 120) and low-risk outliers (40%, i.e., 80). The analysis showed that the average deviation of the data characteristics of high-risk outliers from the normal range was 3.5 standard deviations, while that of low-risk outliers was only 1.8 standard deviations, indicating that the stratification was reasonable.

[0099] Next, a false alarm rate control mechanism was used to screen high-confidence anomalies. The false alarm rate threshold was set at 0.05. Statistical tests (such as the Z-score test) were used to calculate the confidence level of 120 high-risk anomalies. Points with Z-scores below 2.0 were removed, and finally 90 anomalies with a confidence level above 95% were selected. The analysis showed that the false alarm rate was controlled at 0.04, which was below the threshold, and the effect was ideal.

[0100] Finally, for the precisely located set of abnormal signals, a time series analysis algorithm (such as the sliding window method) was used with a window size of 5 minutes and a step size of 1 minute to perform spatiotemporal localization on these 90 high-confidence abnormal events. Combined with business log data, it was found that 80 of the abnormal signals were concentrated during the peak period of system load (load rate exceeding 85%), and the remaining 10 were related to network latency (latency exceeding 200ms), thus forming a precisely located set of abnormal signals. The analysis showed that the localization accuracy reached 92%, providing a reliable basis for subsequent fault diagnosis.

[0101] The above process is automated through algorithms. Data stratification, filtering, and positioning all rely on machine learning and statistical models to ensure that the logic is rigorous and closely related to business scenarios (such as system load and network latency).

[0102] S4. From the set of precisely located abnormal signals, associate the abnormal events with specific line nodes according to the node communication protocol, and generate an abnormal report data packet containing location information.

[0103] S4-1. Obtain abnormal signal data from the stored signal set, perform preliminary screening for each group of signal data to determine whether there are abnormal fluctuations, and if abnormal fluctuations are detected, determine it as an abnormal signal group to be processed.

[0104] S4-2. For a given abnormal signal group, a preset node communication rule is used to parse the signal source through the communication protocol and obtain the corresponding line node identifier.

[0105] S4-3. Based on the line node identifier, obtain the associated geographical location information, and combine it with the characteristics of the abnormal signal group to generate an abnormal event record containing location information.

[0106] S4-4. By classifying and processing the abnormal event records, and binding the abnormal events with line nodes according to the format requirements of the communication protocol, a structured abnormal data unit is obtained.

[0107] S4-5. For structured abnormal data units, a unified data encapsulation method is adopted to generate abnormal data packets that conform to the protocol specifications.

[0108] S4-6. The generated abnormal data packet is forwarded to the target system through the preset transmission channel to complete the transmission and storage of abnormal information.

[0109] For example, when processing anomalous signal data, suppose we extract data from a set containing 1000 anomalous signals. First, we perform preliminary filtering by signal strength and timestamp, filtering out signals with signal strength exceeding the threshold of -75dBm and timestamps within the last 24 hours, resulting in 200 high-priority anomalous signals.

[0110] Next, based on the node communication protocol (such as a custom protocol based on TCP / IP), these signals are associated with specific line nodes. The specific method is to parse the node ID field in the signal data packet and match it with the geographical location information of 5,000 nodes in the pre-stored node database. For example, if the node ID of a certain signal is N1234, the corresponding line node location recorded in the database is 116.39 degrees east longitude and 39.91 degrees north latitude.

[0111] During the association process, a distance calculation algorithm (such as Euclidean distance) is used to verify the matching degree between the signal source and the node location. The distance error threshold is set to 50 meters. If the calculation result is 30.2 meters, the association is confirmed to be successful; otherwise, it is marked as pending verification.

[0112] Subsequently, an anomaly report data packet containing location information is generated. The data packet format follows a JSON structure and includes fields such as "Node ID: N1234", "Anomaly Type: Signal Interruption", "Location: 116.39 degrees East Longitude, 39.91 degrees North Latitude", "Time: 2023-10-15, 14:30:25", and "Severity: High".

[0113] During the generation process, the system automatically calls the severity assessment algorithm to calculate the severity score based on the duration of the signal interruption (e.g., 5 minutes) and the scope of impact (e.g., covering 100 user terminals). The score is 8.5 (out of 10), thus classifying it as "high" severity.

[0114] Finally, the data packets are sent to the monitoring center via an encrypted transmission protocol (such as AES-256) to ensure data security.

[0115] To form a business logic chain, if the monitoring center reports that further analysis is needed, the historical data comparison module can be automatically triggered to extract abnormal records of the same node in the past 7 days (such as 3 similar interruptions). The probability of it happening again in the next 24 hours is predicted by trend analysis algorithms (such as linear regression). The calculated result is 75.3%, and the prediction result is attached to the report to form a closed-loop processing mechanism.

[0116] S5. Power management technology ensures the stability of sensor nodes during long-distance transmission. Abnormal report data packets are converted into transmission sequences through a specific encoding method. If signal interruption or delay is detected during transmission, the transmission route is adjusted through a backup communication path to determine the optimized transmission channel.

[0117] S5-1. Through a pre-established power management module, the energy consumption of the sensor node is monitored in real time, the power supply mode is adjusted by a dynamic allocation strategy, the current endurance status data of the node is obtained, and it is determined whether it meets the energy requirements for long-distance transmission.

[0118] S5-2. Based on the remaining power status data, if the energy reserve is detected to be lower than the preset threshold, the energy-saving mode is triggered, non-essential data processing tasks are restricted, an adjusted energy allocation scheme is obtained, and it is determined whether the node can maintain stable operation.

[0119] S5-3. For stable sensor nodes, acquire abnormal report data packets, convert them into transmission sequences using specific encoding rules, complete the structured processing of data packets, and determine whether the encoded sequence meets the requirements of the transmission protocol.

[0120] S5-4. Using the encoded transmission sequence, the communication channel status is detected in real time during long-distance transmission. If signal interruption or transmission delay exceeds the preset range, the current channel status information is recorded to obtain the abnormal transmission judgment result.

[0121] S5-5. Based on the result of the abnormal transmission determination, switch to the pre-configured backup path, recalculate the routing scheme, obtain new transmission channel parameters, and determine whether the normal transmission rate can be restored.

[0122] S5-6. For the new transmission channel parameters, perform a routing adjustment operation to optimize the bandwidth allocation of the communication path, obtain the final optimized channel configuration, and determine whether the stability of long-distance transmission is maintained.

[0123] S5-7. Through the optimized channel configuration, continuously monitor the operating status of sensor nodes and the integrity of transmission sequences. If a new abnormal report data packet is detected, the above encoding and path adjustment process is executed repeatedly to obtain the latest transmission status information and determine whether the overall system operation meets expectations.

[0124] For example, in ensuring the stability of long-distance transmission of sensor nodes, the power consumption of the nodes is first optimized through power management technology. Specifically, a dynamic voltage regulation algorithm can be used to adjust the power supply voltage according to the node's workload. For example, the voltage can be reduced from 3.3V to 2.8V under low load, reducing power consumption by about 15%. By monitoring the node's power level in real time, the node can automatically switch to a low-power mode when the power level is below 20%, extending the battery life to more than 48 hours. Analysis shows that this method can reduce ineffective power consumption by 30%.

[0125] Secondly, when converting abnormal report data packets into transmission sequences, Huffman coding can be used to assign shorter coding bits (e.g., 2 bits) to frequently occurring abnormal codes (e.g., error code 001) and longer coding bits (e.g., 5 bits) to low-frequency codes (e.g., error code 999). Through statistical analysis, the data compression rate after encoding reaches 40%, effectively reducing the transmission bandwidth usage.

[0126] Next, if a signal interruption or delay is detected during transmission, such as a delay exceeding 200ms or a packet loss rate higher than 5%, the system automatically triggers the backup communication path selection algorithm. Based on the shortest path priority principle, the system calculates the backup path delay from node A to node B. Assuming the primary path delay is 150ms, the backup path 1 delay is 180ms, and the backup path 2 delay is 220ms, the system selects backup path 1 as the new route. Analysis shows that this adjustment can improve the transmission success rate to over 95%.

[0127] Finally, when determining the optimized transmission channel, the system evaluates the path quality through the signal-to-noise ratio (SNR), setting a threshold of 20dB. If the SNR of a certain path is 22dB, it is selected first. At the same time, combined with the analysis of historical transmission data, the system predicts the path stability in the next hour. If the stability probability is higher than 90%, the channel is locked to ensure the continuity of data transmission.

[0128] Through the aforementioned technical means, each link forms a tight logical chain, from energy consumption optimization to data encoding, path adjustment and channel locking, ensuring the high efficiency and stability of long-distance transmission.

[0129] S6. Obtain complete anomaly report data arriving at the control center based on the optimized transmission channel, perform integrity verification on the data packets, and if the verification finds that the data is missing, trigger the retransmission mechanism to obtain the final confirmed fault location information.

[0130] S6-1. Obtain anomaly report data through the transmission channel, perform preliminary parsing on the received data packets, and obtain the initial data structure content;

[0131] S6-2. Based on the initial data structure content, a verification mechanism is used to check the integrity of the data packet. If data is missing, the identification information of the missing part is recorded to determine the range of missing data.

[0132] S6-3. For the range of missing data, start the retransmission mechanism to retrieve the corresponding data packet content from the transmission channel and obtain the supplemented data packet set.

[0133] S6-4. After supplementing the data packet set, a secondary verification mechanism is executed to determine whether the data integrity meets the preset standard. If there are still missing data, the retransmission mechanism is repeated until complete data is obtained.

[0134] S6-5. Based on the complete data, analyze the specific content of the anomaly report, extract information fields related to fault location, and determine the key description of the fault.

[0135] S6-6. After obtaining the key description of the fault, compare and analyze it with the fault database pre-established by the control center, and use the support vector machine algorithm to classify the fault type to obtain the final fault location result.

[0136] S6-7. Based on the final fault location result, generate an information confirmation record and send it back to the control center through the optimized transmission path to complete the data processing flow.

[0137] For example, when acquiring complete anomaly report data arriving at the control center through the optimized transmission channel, assuming the system transmits data via a high-bandwidth fiber optic network with a bandwidth of 10Gbps and a data packet size of 1MB, using fragmentation technology during transmission with each fragment being 256KB, the system automatically records the sending and arrival times of each data packet and calculates the transmission delay. Assuming a normal delay of 5ms, if a data packet's delay exceeds 10ms, the system automatically marks it as a potential anomaly and logs it. Subsequently, it enters the integrity verification stage. The system uses the CRC32 algorithm to verify the data packets and calculates the checksum of each data packet. For example, the CRC32 value of data packet A is 0x4A17B156. If the value calculated by the receiving end is inconsistent with that of the sending end, it is determined that the data is corrupted, triggering a retransmission mechanism. The system will automatically... The sending end sends a retransmission request, limiting the number of retransmissions to 3. If the retransmission fails after 3 attempts, the data packet is marked as lost and the upper-layer application is notified. At the same time, the percentage of lost data packets is recorded. Assuming there are 1000 data packets in total and 2 are lost, accounting for 0.2%, the system will analyze the cause of the loss, which may be network congestion or hardware failure. This will trigger a switchover to the backup channel to ensure data transmission continuity. When finally confirming the fault location information, the system analyzes the transmission path and timestamp of the abnormal data packets, combined with historical data. Assuming that a node has experienced 5 similar delay anomalies in the past 24 hours, the system determines that the node is the fault point, automatically generates a fault report, including the node ID, the number of anomalies, and the time distribution, and sends the report to the control center through an encrypted channel. At the same time, the fault database is updated to form a closed-loop management system to ensure subsequent transmission optimization.

[0138] S7. Generate a targeted isolation command sequence based on the finally confirmed fault location information. Ensure that the commands are quickly sent to the field equipment through real-time classification speed, execute the automatic isolation operation of the abnormal node of the line, and obtain the final fault isolation result.

[0139] S7-1. Obtain detailed information about abnormal nodes through fault location data to determine the specific location of line abnormalities;

[0140] S7-2. Based on the location of the line anomaly, generate the corresponding isolation command and construct the command sequence to cover all relevant anomaly nodes;

[0141] S7-3. Real-time classification technology is used to prioritize the instruction sequence and determine the urgency of the instruction. If the urgency is higher than a preset threshold, it is immediately transmitted to the field equipment.

[0142] S7-4. Upon receiving the isolation command from the field equipment, perform automatic isolation operations and obtain the isolation status information of the abnormal node;

[0143] S7-5. Analyze the completion status of fault isolation from the isolation status information. If there are any abnormal nodes that have not been isolated, generate a supplementary isolation command and send it to the field equipment.

[0144] S7-6. Based on the isolation operation data fed back by the field equipment, determine the final fault isolation result and record the processing status of all abnormal nodes;

[0145] S7-7. After obtaining the final fault isolation result, update the monitoring data of the line anomaly to form a complete isolation result file.

[0146] For example, in processing fault location information, the system first automatically analyzes the data of the finally confirmed fault point. Suppose that the fault location information of a power line shows that a short circuit occurred at tower No. 10. The system will calculate the scope of the fault impact based on historical data and real-time monitoring data, and analyze and find that there are 3 substations and 5 branch lines within a radius of 2.5 kilometers.

[0147] Next, a targeted isolation instruction sequence is generated. The system calculates the combination of nodes that need to be isolated based on a preset isolation algorithm (such as the shortest path algorithm). The priority order is the isolation of the main line from tower 10 to tower 11. The time is expected to be 0.5 seconds. The generated instruction sequence includes closing the circuit breaker switches of towers 10 and 11. The instruction code is "ISO_10_11_001".

[0148] Subsequently, the system ensures rapid instruction delivery by real-time classification speed. The system utilizes 5G network for transmission, setting the instruction delivery delay to no more than 0.1 seconds. The classifier prioritizes instructions based on the urgency of the fault, ensuring that they are transmitted to the field equipment within 100 milliseconds. During transmission, the CRC check algorithm is used to verify data integrity, and the bit error rate is controlled below 0.0001.

[0149] Then, the automatic isolation operation of the abnormal node of the line is performed. After receiving the instruction, the field equipment automatically triggers the circuit breaker to close the designated switch. The isolation time is controlled within 0.2 seconds. The system records the isolation action log in real time, including timestamps and equipment status (such as "Circuit breaker No. 10 is closed, time 2023-10-01, 14:30:00").

[0150] Finally, the final fault isolation result is obtained. The system analyzes the isolation effect through sensor feedback data, confirming that the fault current has dropped from 500 amps to 0 amps and the affected area has been reduced to 0.5 kilometers. An isolation report is generated and uploaded to the cloud database. At the same time, the system coordinates with the dispatch system to adjust the power supply path to ensure that the load rate of other lines does not exceed 80%. For example, through algorithm optimization, the load is transferred from line A to line B by a transfer ratio of 30%, forming a complete business closed loop to ensure the stable operation of the power grid.

[0151] In some alternative embodiments, during a typhoon in a coastal city, a 10kV harbor line experienced an intermittent C-phase short-circuit fault near tower No. 15 due to tree branches broken by strong winds touching the ground.

[0152] The system operation process is as follows:

[0153] S1: LoRa sensors deployed upstream (tower 14) and downstream (tower 16) of tower 15 captured an abnormal waveform at a sampling rate of 1000 times per second, showing a sudden drop in C-phase voltage from 10.2kV to 8.5kV, while the C-phase current surged from 85A to 450A. The sensor's built-in temperature module recorded a 15°C instantaneous increase in ambient temperature near the fault point due to the electric arc. All data, after wavelet transform denoising, were timestamped and tagged with location to form the raw monitoring data set, which was then uploaded via the LoRa network.

[0154] The distributed layout ensures that the fault point is perceived by multiple nodes; the high sampling rate and high accuracy capture the complete transient characteristics of the fault, laying the data foundation for subsequent analysis; environmental interference filtering ensures the reliability of the data under severe weather conditions.

[0155] S2: After receiving the data, the edge computing gateway performs Z-score normalization on the voltage and current data to bring them to the same dimension. Subsequently, the system extracts the current surge rate (di / dt) and voltage sag depth as key indicators. Calculations revealed that the current surge rate was as high as 7200 A / s, far exceeding the preset threshold of 1000 A / s; the voltage sag depth reached 16.7%, far exceeding the 5% threshold. The system immediately identified this signal as a "preliminary anomaly."

[0156] Data standardization eliminates the influence of units, making it easier for general algorithms to process; the selection of feature dimensions (instead of simply using the maximum value, but using more representative time-domain features such as mutation rate and indentation depth) improves the sensitivity and accuracy of anomaly identification.

[0157] S3: Abnormal Signal Optimization and Precise Location:

[0158] Operational process: Multiple nodes within the system simultaneously reported minor fluctuations (such as a current spike caused by load switching on tower No. 16). The initial anomaly signal pool contains fault signal No. 15 and these interference signals.

[0159] Classification boundary optimization: The system calls a pre-trained Support Vector Machine (SVM) model to cluster all preliminary abnormal signals based on features such as "fault current harmonic content", "fault duration", and "voltage and current phase changes". The model successfully classified the fault of tower No. 15 (features: rich in odd harmonics, duration > 100 ms, phase abrupt change) into the "permanent fault" layer; while classifying the interference of tower No. 16 (features: low harmonic content, duration < 20 ms) into the "transient disturbance" layer.

[0160] False alarm rate control: The system sets the confidence threshold to 98%. Data from the "transient disturbance" layer is verified, and its confidence level is only 65%-85%, so it is completely rejected. Ultimately, only the fault signal from tower No. 15 is retained with a confidence level of 99.5%, forming a set of precisely located abnormal signals.

[0161] Key takeaway: This step is crucial for improving the reliability of this invention. By optimizing the classification boundaries, refined fault grading is achieved; through a false alarm rate control mechanism, over 30% of interference signals are effectively filtered out, ensuring that subsequent operations target only high-confidence genuine faults, greatly avoiding unnecessary power outages and maintenance deployments.

[0162] S4: Generate Location Report: Based on the unique ID of the sensor on tower 15, the system locates its latitude and longitude (E121.50, N31.25) in the GIS map and generates an anomaly report data packet according to the standard protocol. The packet contains: {Event ID: F-20241015-001, Location: Tower 15, Latitude and Longitude: (121.50, 31.25), Fault Type: C-phase grounding, Severity Level: Critical}.

[0163] It achieves seamless association between fault information and geographic information, providing precise "targets" for subsequent isolation and inspection.

[0164] S5: Reliable long-distance transmission;

[0165] Operational Process: During typhoon weather, the communication link quality from tower 15 to the main gateway deteriorated (RSSI < -125dBm). The system immediately activated power management, and the sensors switched to maximum transmit power mode to ensure signal penetration. Simultaneously, the transmission management module detected a packet loss rate exceeding 60% on the main path and automatically activated the backup path: data packets were first sent to the relay node on tower 12 with better signal strength, and then relayed to the control center. The data packets employed forward error correction (FEC) coding, ensuring that the receiving end could recover complete information even with packet loss.

[0166] Dynamic power management and adaptive routing adjustment ensure the survivability and stability of communication links in extreme environments, which is crucial for the transmission of fault information and solves the pain point of communication interruption in remote or harsh environments.

[0167] S6: The control center receives the data packet, and the CRC check passes. After parsing, the system calls the fault knowledge base and uses the SVM algorithm again to compare the fault characteristics, finally confirming that the fault type is "C-phase permanent grounding fault", and the location information is correct.

[0168] End-to-end integrity verification (CRC) and secondary classification confirmation by the control center constitute a double guarantee, ensuring the absolute reliability of the decision-making basis.

[0169] S7: Based on the confirmed fault location information, the control center automatically generates the optimal isolation command sequence within 200 milliseconds:

[0170] Disconnect the sectionalizing switch on tower No. 14 (isolate the upstream of the fault point).

[0171] Close the connecting switch on tower No. 13 (to restore power from other lines to the non-faulty area).

[0172] The command was issued with the highest priority via the 5G network, and the on-site smart switch completed the execution within 150 milliseconds. Ultimately, the fault point was quickly isolated, the affected area was controlled between towers 14 and 16, and downstream users quickly had their power restored through power transfer.

[0173] The "real-time classification speed" and "automatic closed-loop control" from fault confirmation to instruction generation shorten the traditional manual handling process that takes tens of minutes to seconds, greatly reducing the scope and duration of power outages, demonstrating the great value of this system in improving power supply reliability.

[0174] Through this coherent scenario, it is clear that this invention is not a collection of scattered technologies, but a complete, intelligent, and efficient automated system from "sensing" to "handling." Among these, the intelligent screening and positioning of S3 and the reliable transmission of S5 play a decisive role in key stages, jointly ensuring the accuracy, reliability, and speed of fault handling, perfectly solving the industry problems of "high false alarm rate, unstable transmission, and slow response."

[0175] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A LoRa communication system for detecting faults in power distribution network lines, characterized in that, include: Distributed sensor networks are used to deploy distributed sensors at key nodes of power distribution lines. They continuously monitor changes in voltage, current, and temperature within the monitoring range of each node with high voltage acquisition accuracy and specific current signal sampling rate. Combined with environmental interference filtering technology, noise is removed, and raw monitoring data sets are generated according to a preset data acquisition frequency. The data preprocessing module is used to standardize the collected signals based on the original monitoring data set, extract key indicators after feature dimension selection for voltage and current fluctuations, and determine preliminary abnormal signals when the processed indicators deviate from the abnormal threshold set range. The abnormal signal optimization module is used to refine and stratify the abnormal data based on the initial abnormal signals using classification boundary optimization technology, and combined with the false alarm rate control mechanism to filter out high-confidence abnormal events, thereby obtaining a set of precisely located abnormal signals. The report generation module is used to associate abnormal events with specific line nodes from a set of precisely located abnormal signals according to the node communication protocol, and generate an abnormal report data packet containing location information. The transmission management module is used to ensure the stability of sensor nodes during long-distance transmission through power management technology, convert abnormal report data packets into transmission sequences through a specific encoding method, and adjust the transmission route through backup communication paths when signal interruption or delay is detected during transmission to determine the optimized transmission channel. The data verification module is used to obtain complete anomaly report data arriving at the control center based on the optimized transmission channel, perform integrity verification on the data packets, and trigger a retransmission mechanism when data is found to be missing during verification, so as to obtain the final confirmed fault location information. The isolation command execution module is used to generate a targeted isolation command sequence based on the finally confirmed fault location information. Through real-time classification speed, the commands are quickly sent to the field equipment to perform automatic isolation operations on the abnormal nodes of the line and obtain the final fault isolation result.

2. A method for using a LoRa communication system for fault detection in power distribution networks, characterized in that, The method, applied to the LoRa communication system for power distribution line fault detection as described in claim 1, comprises: By deploying distributed sensors at key nodes of power distribution lines, the voltage, current and temperature of the lines are continuously monitored within the monitoring range with high voltage acquisition accuracy and specific current signal sampling rate. Combined with environmental interference filtering technology to remove noise influence, the original monitoring data sets are generated according to the preset data acquisition frequency. Based on the original monitoring data set, the collected signals are standardized using a data preprocessing process. Key indicators are extracted based on the selected feature dimensions for voltage and current fluctuations. If the processed indicators deviate from the abnormal threshold set range, they are judged as preliminary abnormal signals. For the initial abnormal signals, the classification boundary optimization technique is used to refine and stratify the abnormal data. Combined with the false alarm rate control mechanism, high-confidence abnormal events are selected to obtain a set of precisely located abnormal signals. From the set of precisely located abnormal signals, the abnormal events are associated with specific line nodes according to the node communication protocol, and an abnormal report data packet containing location information is generated. Power management technology ensures the stability of sensor nodes during long-distance transmission. Abnormal report data packets are converted into transmission sequences through a specific encoding method. If signal interruption or delay is detected during transmission, the transmission route is adjusted through a backup communication path to determine the optimized transmission channel. Complete anomaly report data is obtained from the control center through the optimized transmission channel. The integrity of the data packets is checked. If the check finds that the data is missing, the retransmission mechanism is triggered to obtain the final confirmed fault location information. Based on the finally confirmed fault location information, a targeted isolation command sequence is generated. The command is quickly sent to the field equipment through real-time classification speed to perform automatic isolation operation of the abnormal node of the line and obtain the final fault isolation result.

3. The method of using the LoRa communication system for fault detection in power distribution networks according to claim 2, characterized in that, The method involves deploying a distributed sensor network at key nodes of the power distribution line. For each node, high voltage acquisition accuracy and a specific current signal sampling rate are used to continuously monitor changes in voltage, current, and temperature within the monitoring range. Environmental interference filtering technology is used to remove noise influence, and raw monitoring data sets are generated according to a preset data acquisition frequency. These data sets include: By acquiring voltage, current, and temperature change data from distributed sensors at key nodes of power distribution lines, and using a pre-set processing module to preliminarily organize the acquired signals, a structured monitoring data set is obtained. Based on the structured monitoring data set, environmental interference filtering technology is used to remove noise from the data, generating a denoised monitoring data set.

4. The method of using the LoRa communication system for fault detection in power distribution network lines according to claim 2, characterized in that: The process involves standardizing the collected signals using a data preprocessing workflow based on the original monitoring data set. Key indicators, selected based on feature dimensions, are extracted to address voltage and current fluctuations. If the processed indicators deviate from the set abnormal threshold range, they are identified as preliminary abnormal signals, including: Raw data is obtained from monitoring equipment. The collected signal data is then preliminarily cleaned to remove invalid values ​​and noise interference, resulting in a cleaned signal dataset. For the cleaned signal dataset, a standardization operation is used to convert voltage and current data of different dimensions into a unified scale, resulting in a standardized signal matrix; From the standardized signal matrix, characteristic indicators of voltage and current fluctuations are extracted, and key dimension values ​​related to fluctuations are determined by calculating the statistical parameters of the time series. Based on the key dimension values, anomaly detection rules are constructed. If the calculated dimension values ​​exceed the preset anomaly threshold range, it is determined as a preliminary anomaly signal and an anomaly marker is output.

5. The method of using the LoRa communication system for fault detection in power distribution network lines according to claim 2, characterized in that: The process involves refining and stratifying the anomalous data using classification boundary optimization techniques for the initial anomalous signals, and then filtering out high-confidence anomalous events using a false alarm rate control mechanism to obtain a precisely located set of anomalous signals, including: Preliminary abnormal signals are obtained from the raw data, and a pre-established classification model is used to perform preliminary segmentation of the signals. Abnormal data parts are extracted from the segmentation results to obtain a preliminary abnormal data set. For the initial abnormal data set, the data is stratified by the classification boundary optimization method to obtain abnormal data groups at different levels and determine the abnormal data structure after stratification. For the hierarchical abnormal data structure, a false alarm control strategy is used to filter the data. If the abnormal data of a certain level is lower than the preset confidence threshold, the data of that level is removed to obtain a subset of abnormal data with high confidence. For a subset of anomalous data with high confidence, an event correlation analysis is performed on each set of data through a screening strategy. If the correlation of events in a certain data set is higher than the preset standard, the data set is retained, and the set of anomalous events that meet the conditions is identified. For a set of abnormal events that meet the criteria, the location is mapped based on the results of the classification boundary optimization to obtain the specific location information of each abnormal event and determine the list of abnormal events with precise location. For the precisely located list of abnormal events, all relevant data are integrated to form a unified signal set. Data verification tools are used to perform consistency checks on the signal set to obtain the final abnormal signal set.

6. The method of using the LoRa communication system for fault detection in power distribution networks according to claim 2, characterized in that: The process of associating abnormal events with specific line nodes from a precisely located set of abnormal signals, according to the node communication protocol, and generating an abnormal report data packet containing location information includes: Abnormal signal data is obtained from the stored signal set. Each group of signal data is initially screened to determine whether there are abnormal fluctuations. If abnormal fluctuations are detected, it is identified as an abnormal signal group to be processed. For a given abnormal signal group, a preset node communication rule is used to parse the signal source through the communication protocol and obtain the corresponding line node identifier. Based on the line node identifier, obtain the associated geographical location information, and combine it with the characteristics of the abnormal signal group to generate an abnormal event record containing location information; By classifying and processing abnormal event records, and binding abnormal events to line nodes according to the format requirements of communication protocols, structured abnormal data units are obtained. For structured abnormal data units, a unified data encapsulation method is adopted to generate abnormal data packets that conform to the protocol specifications; The generated abnormal data packets are forwarded to the target system through a preset transmission channel, thus completing the transmission and storage of abnormal information.

7. The method of using the LoRa communication system for fault detection in power distribution network lines according to claim 2, characterized in that: The power management technology ensures the stability of sensor nodes during long-distance transmission. Abnormal report data packets are converted into transmission sequences using a specific encoding method. If signal interruption or delay is detected during transmission, the transmission route is adjusted using a backup communication path to determine an optimized transmission channel, including: The power management module is pre-established to monitor the energy consumption of the sensor nodes in real time, and the power supply mode is adjusted by a dynamic allocation strategy. The power supply status data of the current node is obtained to determine whether it meets the energy requirements for long-distance transmission. Based on the remaining power status data, if the energy reserve is detected to be lower than the preset threshold, the energy-saving mode is triggered, non-essential data processing tasks are restricted, an adjusted energy allocation scheme is obtained, and it is determined whether the node can maintain stable operation. For stable sensor nodes, acquire anomaly report data packets, convert them into transmission sequences using specific encoding rules, complete the structured processing of data packets, and determine whether the encoded sequence conforms to the transmission protocol requirements; The communication channel status is monitored in real time during long-distance transmission using the encoded transmission sequence. If signal interruption or transmission delay exceeds a preset range, the current channel status information is recorded to obtain the abnormal transmission judgment result. Based on the abnormal transmission determination result, switch to the pre-configured backup path, recalculate the routing scheme, obtain new transmission channel parameters, and determine whether the normal transmission rate can be restored. For the new transmission channel parameters, a routing adjustment operation is performed to optimize the bandwidth allocation of the communication path, obtain the final optimized channel configuration, and determine whether the stability of long-distance transmission is maintained. Through the optimized channel configuration, the operating status of sensor nodes and the integrity of transmission sequences are continuously monitored. If a new abnormal report data packet is detected, the above encoding and path adjustment process is executed repeatedly to obtain the latest transmission status information and determine whether the overall system operation meets expectations.

8. The method of using the LoRa communication system for fault detection in power distribution network lines according to claim 2, characterized in that: The process involves obtaining complete anomaly report data arriving at the control center through the optimized transmission channel, performing integrity checks on the data packets, and triggering a retransmission mechanism if data is found to be missing, to obtain the final confirmed fault location information, including: The abnormal report data is obtained through the transmission channel, and the received data packets are initially parsed to obtain the initial data structure content; Based on the initial data structure content, a verification mechanism is used to check the integrity of the data packet. If data is missing, the identification information of the missing part is recorded to determine the range of missing data. For the range of missing data, a retransmission mechanism is initiated to retrieve the corresponding data packet content from the transmission channel, resulting in a supplemented set of data packets. The supplemented data packet set is used to perform a secondary verification mechanism to determine whether the data integrity meets the preset standard. If there are still missing data, the retransmission mechanism is repeated until the complete data is obtained. Based on the complete data, analyze the specific content of the anomaly report, extract information fields related to fault location, and determine the key description of the fault. After obtaining the key description of the fault, it is compared and analyzed with the fault database pre-established by the control center. The support vector machine algorithm is used to classify the fault type and obtain the final fault location result. Based on the final fault location results, an information confirmation record is generated and transmitted back to the control center through the optimized transmission path to complete the data processing flow.

9. The method of using the LoRa communication system for fault detection in power distribution network lines according to claim 2, characterized in that: The process involves generating a targeted isolation command sequence based on the finally confirmed fault location information. Real-time classification ensures the commands are rapidly sent to field equipment to execute automatic isolation of abnormal line nodes, resulting in the final fault isolation outcome. This includes: By using fault location data, detailed information about abnormal nodes can be obtained to determine the specific location of the line abnormality. Based on the location of the line anomaly, generate corresponding isolation instructions and construct an instruction sequence to cover all relevant anomaly nodes; Real-time classification technology is used to prioritize the instruction sequence and determine the urgency of the instruction. If the urgency is higher than a preset threshold, it is immediately transmitted to the field equipment. Upon receiving isolation commands from field devices, perform automatic isolation operations and obtain isolation status information for abnormal nodes; Analyze the completion status of fault isolation from the isolation status information. If there are any abnormal nodes that have not been isolated, generate a supplementary isolation command and send it to the field equipment. The final fault isolation result is determined by the isolation operation data fed back by the field equipment, and the processing status of all abnormal nodes is recorded. After obtaining the final fault isolation results, update the monitoring data of the abnormal line to form a complete isolation result archive.

Images