A method and system for intelligent power outage and insulation restoration in the event of an electrical fire.

Abstract

This invention discloses an intelligent power-off and insulation restoration method and system for electrical fires. The method uses multi-source sensing units deployed on-site to collect parameters such as current, temperature, and insulation resistance of electrical lines in real time. Edge computing nodes perform preprocessing and feature extraction locally to form a fire risk feature vector, which is then input into a pre-set risk assessment model for real-time analysis, outputting a dynamic fire risk level. When the risk level reaches high risk, the edge computing node immediately generates a power-off command, controlling the intelligent circuit breaker to cut off the power supply. After power failure, the system continuously monitors the insulation resistance value. Once it recovers to a safe threshold and remains stable, it automatically generates a power restoration command, controlling the circuit breaker to close. The system includes multi-source sensing units, edge computing nodes, and intelligent execution units. This invention achieves closed-loop management from intelligent early warning and rapid power failure to safe restoration, significantly improving the timeliness, accuracy, and automation level of electrical fire protection.

Description

Technical Field

[0001] This invention belongs to the field of electrical safety and protection, and relates to an intelligent power-off and insulation restoration method for electrical fires, as well as an intelligent power-off and insulation restoration system for electrical fires. Background Technology

[0002] Electrical fires are one of the main causes of building fires. They are usually caused by the long-term accumulation of hidden dangers such as circuit overload, poor contact, insulation aging, or grounding faults, which then suddenly erupt. These fires are characterized by their high degree of concealment, rapid development, and great destructive power, and traditional protection methods are clearly inadequate.

[0003] Currently, common electrical fire protection technologies mainly rely on circuit breakers and residual current electrical fire monitoring detectors. Circuit breakers primarily protect against overloads and short circuits, and their operation depends on a fixed current threshold, making them unable to identify early, slowly changing fault symptoms. While residual current detectors can detect leakage current, their judgment criteria are singular, making them susceptible to environmental interference that can lead to false alarms or missed alarms. Furthermore, they cannot comprehensively consider multi-dimensional information such as temperature and harmonics for accurate assessment.

[0004] In recent years, with the development of IoT and AI technologies, some cloud-based intelligent power monitoring systems have emerged. These systems collect data by deploying sensors and upload it to the cloud for analysis, enabling more comprehensive status monitoring. However, this architecture has inherent drawbacks: First, the loop delay for cloud-based decision-making is relatively long, typically taking seconds or even longer from data upload and analysis to command issuance, making it difficult to meet the millisecond-level response requirements for emergency power cut-off in electrical fires; second, it has high network dependence, and protection functions may fail if the network is interrupted; third, most systems only implement risk warnings or remote manual power cut-off, lacking a complete automated closed loop of "diagnosis-protection-recovery." In particular, after a power outage, whether the line insulation has recovered and when it is safe to restore power still requires on-site testing and judgment by maintenance personnel, which is inefficient and carries the risk of human error.

[0005] Therefore, existing technologies suffer from the following prominent problems: inaccurate early warning mechanisms, untimely protective actions, reliance on cloud-based or manual decision-making processes, and a lack of automated capabilities for safe recovery after power outages. There is an urgent need for an electrical fire prevention and control solution that integrates local intelligent real-time analysis, multi-parameter fusion judgment, rapid proactive protection, and automatic safe recovery. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent power outage and insulation restoration method and system for electrical fires. This invention can integrate local intelligent real-time analysis, multi-parameter fusion judgment, rapid active protection and automatic safety restoration.

[0007] The first technical solution adopted in this invention is an intelligent power-off and insulation restoration method for electrical fires, comprising the following steps:

[0008] Step 1: Acquire electrical parameters in real time from the multi-source sensing units connected to the edge computing node. The electrical parameters include at least the current value, temperature value, and insulation resistance value.

[0009] Step 2: At the edge computing node, preprocess and extract features from the acquired electrical parameters to obtain a fire risk feature vector.

[0010] Step 3: Input the fire risk feature vector into the electrical fire risk assessment model pre-set on the edge computing node. The electrical fire risk assessment model performs real-time calculations and outputs the dynamic fire risk level of the current electrical circuit.

[0011] Step 4: If the dynamic fire risk level reaches the preset high-risk threshold, the edge computing node will immediately generate a power-off command and send it to the controlled smart circuit breaker to cut off the power supply to the corresponding electrical circuit.

[0012] Step 5: After the power is cut off, continuously monitor the insulation resistance value of the electrical circuit;

[0013] Step 6: When the insulation resistance value is detected to recover to the preset safe insulation threshold and remain stable for a preset time, it is determined that the insulation recovery condition is met. The edge computing node generates a power restoration command and sends it to the smart circuit breaker to restore power supply.

[0014] Another technical solution adopted in this invention is an intelligent power outage and insulation restoration system for electrical fires, comprising:

[0015] Multi-source sensing units are deployed on electrical lines to collect electrical parameters, including insulation resistance values, in real time.

[0016] An edge computing node is deployed on-site at the electrical line and is communicatively connected to a multi-source sensing unit. The edge computing node is configured to execute the above-mentioned intelligent power outage and insulation restoration method for electrical fires, so as to perform dynamic fire risk assessment locally and generate corresponding control commands based on the data collected in real time by the multi-source sensing unit.

[0017] The intelligent execution unit is communicatively connected to the edge computing node and is used to respond to the control commands, specifically to perform a power-off operation in response to the power-off command and to perform a power-restoration operation in response to the power-restoration command.

[0018] The control commands include power-off commands and power-restoring commands, with the power-restoring command generated based on a judgment of the insulation resistance recovery status.

[0019] The invention is further characterized by:

[0020] Furthermore, in step 1, the multi-source sensing unit includes a voltage sensor and a leakage current sensor, and the electrical parameters also include voltage value and leakage current value.

[0021] Furthermore, in step 2, the fire risk feature vector includes at least one of the following features: current mutation rate, current harmonic content, temperature rise gradient, leakage current change trend, and load cycle characteristics based on historical data.

[0022] Furthermore, in step 3, the electrical fire risk assessment model is a time-series-based machine learning model.

[0023] Furthermore, in step 6, the determination of insulation restoration conditions includes: before determining the restoration of power supply, simulating a low-voltage test signal by injecting it into the electrical circuit through an edge computing node, and generating a power restoration command after confirming that there is no abnormal leakage current.

[0024] Furthermore, the method also includes uploading electrical parameters, risk levels, action commands, and result data generated by the edge computing nodes during operation to the cloud platform for incremental learning and regular updates of the electrical fire risk assessment model, and then distributing the updated model to the edge computing nodes.

[0025] Furthermore, the intelligent execution unit includes an intelligent circuit breaker and an insulation restoration module;

[0026] Intelligent circuit breakers are used to switch power on and off.

[0027] The insulation restoration module is configured to activate after the smart circuit breaker disconnects the power supply to facilitate the restoration of the insulation condition of the electrical circuit.

[0028] Furthermore, the multi-source sensing unit includes a current sensor, a temperature sensor, and an insulation resistance tester.

[0029] Furthermore, the edge computing nodes and multi-source sensing units are connected in a star topology via industrial fieldbus or industrial Ethernet.

[0030] The beneficial effects of this invention are as follows:

[0031] 1. This invention integrates multi-dimensional parameters such as current, temperature, and insulation resistance to construct a fire risk feature vector and uses time-series models such as LSTM for dynamic risk assessment. This enables earlier and more accurate identification of early signs and emergency states of electrical fires, overcoming the problems of delayed response and high false alarm and missed alarm rates of traditional fixed threshold methods.

[0032] 2. This invention deploys the intelligent risk assessment model on edge computing nodes, eliminating the need for cloud-based data transmission and significantly reducing decision-making latency. The entire process, from risk assessment to issuing a power-off command, can be completed within milliseconds (e.g., ≤50ms), achieving "real-time monitoring and instantaneous response" to electrical fire hazards and significantly shortening protection action time.

[0033] 3. This invention not only focuses on "power outage protection" but also innovatively incorporates "insulation recovery monitoring" and "automatic power restoration" steps. By continuously monitoring the insulation resistance recovery status and automatically restoring power supply after confirming safety, it avoids the cumbersome and uncertain process of "manual troubleshooting and resetting after tripping" required by traditional protection devices, and achieves fully automated closed-loop safety management from early warning and protection to restoration.

[0034] 4. The system designed in this invention establishes exclusive risk benchmarks for different lines through a "personalized baseline generation method" and supports uploading operational data to a cloud platform for incremental learning and periodic updates of the model. This enables the system to adapt to lines with different scenarios and load characteristics, and to continuously optimize evaluation accuracy over time and with data accumulation, exhibiting good scalability and long-term effectiveness.

[0035] 5. This invention adopts an architecture that combines edge computing and a cloud platform, ensuring both the independent and reliable operation of core protection functions locally, unaffected by network fluctuations, and leveraging the computing power of the cloud for model optimization and data analysis. A reliable connection between the sensing layer and the execution layer via an industrial network ensures the overall stability and usability of the system. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart of the present invention;

[0038] Figure 2 This is a block diagram of the present invention. Detailed Implementation

[0039] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0040] The following is in conjunction with the appendix Figure 1 To be continued Figure 2 The invention will be described in detail with specific embodiments:

[0041] A method for intelligent power outage and insulation restoration in the event of an electrical fire, referenced Figure 1 This includes the following steps:

[0042] Step 1: Real-time acquisition of electrical parameters collected by the multi-source sensing unit connected to the edge computing node. The electrical parameters include current value, temperature value, insulation resistance value, voltage value, and leakage current value.

[0043] Step 2: At the edge computing node, preprocess and extract features from the acquired electrical parameters to obtain a fire risk feature vector.

[0044] The fire risk feature vector includes at least one of the following features: current mutation rate, current harmonic content, temperature rise gradient, leakage current change trend, and load cycle characteristics based on historical data.

[0045] The current mutation rate is the rate of change of the effective current value within the current sampling period relative to the previous period, and is used to detect transient events such as short circuits or heavy load switching.

[0046] The harmonic content of current reflects the degree of distortion of the current waveform. Aging of the line, loose connection or abnormal nonlinear load often leads to an increase in specific harmonics.

[0047] The temperature gradient is the change in temperature sensor readings per unit time, used to identify sustained overheating trends rather than fluctuations in ambient temperature.

[0048] The effective value of leakage current and its changing trend directly reflect the insulation degradation of the line or equipment.

[0049] The load cycle characteristic is the load rate deviation calculated based on historical data from the same period, which is used to identify abnormal load patterns.

[0050] The above features together constitute a multi-dimensional "fire risk feature vector", and its construction method is a key component of this invention, enabling the general LSTM model to be effectively applied to specific electrical safety fields.

[0051] Step 3: Input the fire risk feature vector into the electrical fire risk assessment model pre-set on the edge computing node. The electrical fire risk assessment model performs real-time calculations and outputs the dynamic fire risk level of the current electrical circuit.

[0052] This invention replaces the traditional fixed threshold with a "dynamic fire risk level" to achieve accurate and intelligent early warning.

[0053] Dynamic fire risk levels include low risk, medium risk, and high risk, which correspond to different stages in the development of electrical faults into fires.

[0054] Low risk is used to characterize an electrical system as being in a safe operating state, where all parameter fluctuations are within the normal historical range.

[0055] Medium risk is used to characterize early signs of abnormality in the system, which may be a slight and continuous deviation of a single parameter or a short-term disturbance of multiple parameters, requiring attention and enhanced monitoring.

[0056] High risk is used to characterize a system in an emergency state nearing a fault or fire. It is usually manifested by multiple key parameters being seriously out of control or showing a clear trend of rapid deterioration, requiring immediate power outage protection actions.

[0057] After entering self-learning mode, the edge computing node automatically executes a personalized baseline generation method to establish a personalized risk benchmark model for the normal operation of the electrical line. The specific implementation steps are as follows:

[0058] Step 3.1: Set a complete learning cycle and data sampling frequency, and continuously collect raw data from the multi-source sensing unit within the learning cycle to form the raw data sequence of each parameter.

[0059] Suppose there are M parameters to be monitored, denoted as set M. , specifically, .

[0060] Step 3.2, each parameter The original data sequence The length is N total The following steps are performed to apply sliding window filtering and outlier removal:

[0061] Set window width For each primary index in the sequence (from arrive ), calculate its smoothing value:

[0062]

[0063] In the formula, This is the offset within the window (a temporary variable). Indicates parameters The original sequence of the first The value of each sampling point.

[0064] This process yields a smooth sequence. Its effective index range is .

[0065] Calculate smooth sequence mean and standard deviation For each index ( If the following conditions are met:

[0066]

[0067] This point is then considered an outlier. It is replaced using linear interpolation of the valid data points before and after it, ultimately yielding the cleaned sequence. Its effective length is denoted as ( ).

[0068] Step 3.3: For each parameter Cleaned sequence (length is) ),calculate:

[0069] Mean:

[0070]

[0071] Standard deviation:

[0072]

[0073] in, For the primary index of data points, superscript Explicitly indicating that this statistic belongs to the parameter category .

[0074] Step 3.4, based on the results obtained in Step 3.3 and According to statistical principles, for each parameter Define the initial risk scope:

[0075] Low risk range: Medium risk range: High-risk areas (statistics):

[0076] Step 3.5: Read parameters from the built-in electrical safety specification library Legal security limit and lower limit .

[0077] Compare the high-risk area in step 3.4 with the legally defined danger area. Take the union of the sets to obtain the final high-risk range, ensuring compliance with mandatory standards.

[0078] Step 3.6, for each parameter Assign initial weights The initial weights can be set based on the contribution of parameters to electrical fire risk or relevant safety standards, to meet the following requirements. It can be optimized and adjusted based on actual data during subsequent operation.

[0079] Risk score mapping function: Definition function parameters The value of Mapped to risk score:

[0080]

[0081] Comprehensive Risk Score (CRS) calculation: For any given time, given the set of values ​​for each parameter. ,calculate:

[0082]

[0083] Set initial thresholds, including a low-risk threshold. Medium risk threshold High-risk threshold .

[0084] Step 3.7: Set each parameter The final risk scope and weight The data, along with initial thresholds, are packaged to form a personalized risk benchmark model for the loop, which is then serialized and stored in the non-volatile memory of the edge computing node.

[0085] Step 3.8: Check the percentage of effective data in the sequence after cleaning for each parameter. If it is lower than the preset threshold, issue a warning and suggest extending the learning time or checking the sensor.

[0086] In step 3, the electrical fire risk assessment model is a time-series-based machine learning model.

[0087] Specifically, the time-series-based machine learning model uses an LSTM model. The trained long LSTM model is deployed in the memory of the edge computing node for real-time risk assessment. The specific application process is as follows:

[0088] Data preparation and input: The edge computing nodes continuously collect electrical parameters such as temperature and insulation resistance values, which are then preprocessed and feature extracted to form a time-series feature vector sequence of fixed length L. , as the input to the model. Where This indicates the temporal position of the feature vector in the sequence. 1. The earliest moment, This refers to the current moment.

[0089] Specifically, the edge computing node will continuously collect the current, temperature, and insulation resistance values ​​per second over the past 60 seconds, so L=60.

[0090] Forward Propagation and Risk Calculation: LSTM Model by Time Step The input sequence X is processed sequentially from 1 to L.

[0091] At every step The LSTM model is based on the input features at this moment. The hidden state of the previous moment The hidden state at the current time is updated by calculating according to formulas (1) to (6). and cell state

[0092] Forget Gate Formula: （1） Input gate formula: （2） Candidate cell state formula: （3） Update cell state formula: （4） Output gate formula: （5） Update hidden state formula: （6）

[0093] After processing the last time step k The final hidden state obtained afterwards It encodes the context information of the entire L-length sequence.

[0094] Will The input is fed into subsequent fully connected layers and Softmax layers to calculate a probability distribution of risk levels:

[0095]

[0096] In the formula, c represents three categories: low risk, medium risk, and high risk.

[0097] The category with the highest probability is taken as the dynamic fire risk level at the current moment (i.e., the moment at the end of the sequence).

[0098] Step 4: If the dynamic fire risk level reaches the preset high-risk threshold, the edge computing node will immediately generate a power-off command and send it to the controlled smart circuit breaker to cut off the power supply to the corresponding electrical circuit.

[0099] Specifically, the decision time from when the risk level reaches the high-risk threshold to when a power-off command is generated is less than or equal to 50 milliseconds.

[0100] Step 5: After the power is cut off, continuously monitor the insulation resistance value of the electrical circuit;

[0101] Step 6: When the insulation resistance value is detected to return to the preset safe insulation threshold and remain stable for a preset time, it is determined that the insulation recovery condition is met. The edge computing node generates a power restoration command and sends it to the smart circuit breaker to restore power supply.

[0102] Specifically, the determination of insulation restoration conditions includes: before determining whether to restore power supply, simulating a low-voltage test signal by injecting it into the electrical circuit through an edge computing node, and confirming that there is no abnormal leakage current before generating a power restoration command, which is implemented according to the following steps:

[0103] Step 6.1: The edge computing node compares the real-time collected insulation resistance value with the preset safety insulation threshold in real time.

[0104] Step 6.2: While satisfying the threshold condition that the insulation resistance value is ≥ the preset safety insulation threshold, the edge computing node further calculates the rate of change of the insulation resistance value within a continuous preset stable time period, for example, the rate of change per minute does not exceed 2%, and determines whether the rate of change is lower than a preset stable threshold, which can be configured according to the line characteristics and safety requirements.

[0105] Step 6.3: When the threshold condition and stability condition are met simultaneously, the edge computing node controls the generation and injection of a low-power power frequency test voltage signal into the electrical circuit.

[0106] Step 6.4: During the injection of the power frequency test voltage signal, the edge computing node synchronously monitors the leakage current value of the line and determines whether it is less than the preset safe leakage current threshold.

[0107] Step 6.5, Instruction Generation: The edge computing node generates a formal power restoration instruction and sends it to the smart circuit breaker only when the leakage current value is confirmed to be less than the safe leakage current threshold.

[0108] Step 6.6: After receiving the power restoration command, the intelligent circuit breaker performs the closing operation and feeds back the "closed" status signal to the edge computing node.

[0109] The method also includes uploading electrical parameters, risk levels, action commands, and result data generated by the edge computing nodes during operation to the cloud platform for incremental learning and periodic updates of the electrical fire risk assessment model, and then distributing the updated model to the edge computing nodes.

[0110] The intelligent power outage and insulation restoration system for electrical fires provided in this embodiment of the invention is referenced. Figure 2 It mainly includes multi-source sensing units, edge computing nodes, intelligent execution units, and an optional cloud platform.

[0111] The multi-source sensing unit is deployed on the monitored electrical circuit to collect key electrical parameters in real time. Specifically, this unit includes: a current sensor to collect the current value of the circuit; a temperature sensor to collect the temperature value of electrical connection points or cables; and an insulation resistance tester to collect the insulation resistance value of the circuit.

[0112] It may also include voltage transformers, leakage current sensors, etc., for collecting voltage and leakage current values.

[0113] The aforementioned sensors collectively correspond to the data acquisition stage in step 1 of the method, providing real-time, multi-dimensional data sources for subsequent analysis. In this scheme, the multi-source sensing unit is connected to the edge computing node via an industrial fieldbus or industrial Ethernet in a star topology to ensure reliable and low-latency data transmission.

[0114] Edge computing nodes are deployed in the field of electrical wiring, and can be installed inside distribution boxes. They are the core of the system's intelligent analysis. They are connected to multi-source sensing units and intelligent execution units via wired connections and are configured to fully execute the aforementioned methods. Their internal processing flow and method steps strictly correspond:

[0115] Data access and processing correspond to steps 1 and 2. The node receives the raw electrical parameters uploaded by the multi-source sensing unit in real time, performs preprocessing and feature extraction locally, and constructs a fire risk feature vector.

[0116] The dynamic risk assessment corresponds to step 3. The edge computing node has a built-in electrical fire risk assessment model. The feature vector is input into the model for real-time calculation, and the dynamic fire risk level is output.

[0117] During the initialization phase, edge computing nodes execute a "personalized baseline generation method" to establish a risk benchmark model for specific lines.

[0118] The decision-making and instruction generation correspond to steps 4 and 6. When the risk level reaches the high-risk threshold, the edge computing node immediately generates a power-off instruction. After the power outage, the node continuously monitors the insulation resistance value and, based on the logic of steps 5 and 6, determines whether the insulation has been restored to a safe state. When all restoration conditions are met, the node generates a power-restoration instruction.

[0119] Edge computing nodes also upload the data generated during operation to the cloud platform and can receive model updates issued by the cloud platform to achieve continuous optimization of capabilities.

[0120] The intelligent execution unit receives and executes control commands from the edge computing node, specifically including intelligent circuit breakers and insulation restoration modules.

[0121] The intelligent circuit breaker is connected in series in the electrical circuit. When a power outage command is received, it immediately performs a tripping operation to disconnect the power supply. When a power restoration command is received, it performs a closing operation to restore the power supply and feeds back the status to the edge computing node.

[0122] The insulation restoration module is activated by the edge computing node after the smart circuit breaker trips. Its core function is to apply a safe test voltage to the de-energized line according to the node's instructions, so as to assist the insulation resistance tester in continuously and accurately monitoring the insulation resistance restoration process and providing data for power restoration judgment.

[0123] The cloud platform and edge computing nodes are connected via a wide area network. It receives historical operational data uploaded by each edge node, which is used for incremental learning and periodic updates to the risk assessment model, resulting in a more accurate general model. The updated model can then be redeployed to the edge computing nodes to complete closed-loop optimization, improving the overall accuracy and adaptability of the system's early warning system.

[0124] The present invention has been further described above with reference to specific embodiments. However, it should be understood that the specific description herein should not be construed as limiting the nature and scope of the present invention. Various modifications made to the above embodiments by those skilled in the art after reading this specification are all within the scope of protection of the present invention.

Claims

1. A method for intelligent power outage and insulation restoration in the event of an electrical fire, characterized in that, Includes the following steps: Step 1: Real-time acquisition of electrical parameters collected by the multi-source sensing unit connected to the edge computing node, wherein the electrical parameters include at least current value, temperature value and insulation resistance value; Step 2: At the edge computing node, preprocess and extract features from the acquired electrical parameters to obtain a fire risk feature vector. Step 3: Input the fire risk feature vector into the electrical fire risk assessment model pre-set on the edge computing node, and the electrical fire risk assessment model performs real-time calculation and outputs the dynamic fire risk level of the current electrical circuit. Step 4: If the dynamic fire risk level reaches the preset high-risk threshold, the edge computing node immediately generates a power-off command and sends it to the controlled smart circuit breaker to cut off the power supply to the corresponding electrical circuit. Step 5: After the power is cut off, continuously monitor the insulation resistance value of the electrical circuit; Step 6: When the insulation resistance value is detected to recover to the preset safe insulation threshold and remain stable for a preset time, it is determined that the insulation recovery condition is met. The edge computing node generates a power restoration command and sends it to the smart circuit breaker to restore power supply.

2. The intelligent power-off and insulation restoration method for electrical fires according to claim 1, characterized in that, In step 1, the multi-source sensing unit includes a voltage sensor and a leakage current sensor, and the electrical parameters also include voltage value and leakage current value.

3. The intelligent power-off and insulation restoration method for electrical fires according to claim 1 or 2, characterized in that, In step 2, the fire risk feature vector includes at least one of the following features: current mutation rate, current harmonic content, temperature rise gradient, leakage current change trend, and load cycle characteristics based on historical data.

4. The intelligent power-off and insulation restoration method for electrical fires according to claim 1, characterized in that, In step 3, the electrical fire risk assessment model is a time-series-based machine learning model.

5. The intelligent power-off and insulation restoration method for electrical fires according to claim 1, characterized in that, In step 6, the determination of the insulation restoration condition includes: before determining the restoration of power supply, simulating a low-voltage test signal by injecting it into the electrical circuit through the edge computing node, and generating a power restoration command after confirming that there is no abnormal leakage current.

6. The intelligent power-off and insulation restoration method for electrical fires according to claim 1, characterized in that, The method further includes: uploading electrical parameters, risk levels, action commands and result data generated by the edge computing node during operation to the cloud platform for incremental learning and periodic updates of the electrical fire risk assessment model, and distributing the updated model to the edge computing node.

7. An intelligent power outage and insulation restoration system for electrical fires, characterized in that, include: Multi-source sensing units are deployed on electrical lines to collect electrical parameters, including insulation resistance values, in real time. An edge computing node is deployed at the site of the electrical line and is communicatively connected to the multi-source sensing unit. The edge computing node is configured to execute the intelligent power outage and insulation restoration method for electrical fires as described in claims 1 to 6, so as to perform dynamic fire risk assessment locally and generate corresponding control commands based on the data collected in real time by the multi-source sensing unit. The intelligent execution unit is communicatively connected to the edge computing node and is used to respond to the control commands, specifically to perform a power-off operation in response to the power-off command and a power-restoring operation in response to the power-restoring command. The control commands include power-off commands and power-restoring commands, and the power-restoring commands are generated based on the judgment of the insulation resistance recovery status.

8. The intelligent power-off and insulation restoration system for electrical fires according to claim 7, characterized in that, The intelligent execution unit includes an intelligent circuit breaker and an insulation restoration module; The intelligent circuit breaker is used to switch the power supply on and off. The insulation restoration module is configured to activate after the smart circuit breaker cuts off the power supply, so as to facilitate the restoration of the insulation state of the electrical circuit.

9. The intelligent power-off and insulation restoration system for electrical fires according to claim 7, characterized in that, The multi-source sensing unit includes a current sensor, a temperature sensor, and an insulation resistance tester.

10. The intelligent power-off and insulation restoration system for electrical fires according to claim 7, characterized in that, The edge computing node and the multi-source sensing unit are connected in a star topology via industrial fieldbus or industrial Ethernet.

Images