Intelligent circuit breaker communication method and system based on edge computing and cloud

Abstract

The application relates to an intelligent circuit breaker communication method and system based on edge computing and the cloud, the method comprising multi-dimensional electrical parameter acquisition, edge adaptive preprocessing, hierarchical uploading, edge pre-judgment and local linkage, cloud deep pre-judgment, hybrid decision communication, multi-channel heterogeneous notification and closed-loop feedback; the system comprises an acquisition layer, a transmission layer, an edge analysis layer, a cloud analysis layer, a hybrid decision layer, a notification layer, a display layer and a closed-loop feedback layer. Through multi-dimensional electrical parameter acquisition and edge adaptive preprocessing, the application significantly improves data real-time performance and processing efficiency, based on double-priority hierarchical uploading and edge local linkage, greatly compresses emergency fault response time delay, adopts cloud trend prediction and hybrid decision, realizes intelligent pre-judgment, effectively identifies response deviation caused by abnormal operation or equipment aging, supports flexible expansion of more terminal types and notification channels, and has remarkable economic benefits.

Description

Technical Field

[0001] This invention relates to the field of intelligent circuit breaker communication technology, specifically to an intelligent circuit breaker communication method and system based on edge computing and cloud computing. Background Technology

[0002] With the rapid development of smart grids and the Industrial Internet of Things (IIoT), traditional circuit breakers are gradually evolving towards intelligence and networking. As core equipment in low-voltage power distribution, smart circuit breakers not only need to perform basic circuit switching and protection functions, but also need to realize composite functions such as power quality monitoring, operational status sensing, remote control, and fault early warning. This places high-speed, real-time, bidirectional, and integrated technical requirements on the communication system of smart circuit breakers.

[0003] Existing intelligent circuit breaker communication systems have the following technical shortcomings: (1) Lack of hierarchical mechanism for data acquisition and transmission: Existing systems usually collect electrical parameters at a fixed frequency and upload them all to the cloud, resulting in high communication bandwidth occupancy, large cloud computing load, and inability to respond quickly to abnormal states. Although some systems have introduced edge computing nodes for data preprocessing, they have not yet achieved hierarchical data acquisition and differentiated transmission based on parameter deviation. (2) Fault response relies on cloud decision-making and lacks real-time performance: In the traditional architecture, data needs to be uploaded to the cloud platform for analysis through the gateway before control commands are issued. The end-to-end delay is difficult to meet the millisecond-level response requirements of emergency faults (such as short circuits and leakage). Although some solutions integrate edge analysis algorithms locally, they lack a differentiated processing mechanism for emergency faults and non-emergency anomalies. (3) Rigid decision-making mechanism for manual intervention: When the system detects an abnormal trend that requires manual intervention, the existing solutions usually adopt a single notification channel or sequential notification method, which results in low decision-making efficiency. At the same time, there is a lack of a sound mechanism for multi-terminal collaborative decision-making and automatic processing of timeouts, which can easily delay the handling time due to personnel response delays. (4) Lack of dynamic optimization in notification delivery: Existing systems mostly use fixed channels to send alarm notifications without considering the terminal's historical reachability, current online status and timeliness requirements, making it difficult to guarantee the effective delivery of notifications within the preset time limit.

[0004] No solutions have yet been proposed for the relevant technical issues. Summary of the Invention

[0005] To address the problems in related technologies, this invention proposes a smart circuit breaker communication method and system based on edge computing and cloud computing to overcome the aforementioned technical issues in existing technologies. The purpose of this invention is to significantly improve data real-time performance and processing efficiency through multi-dimensional electrical parameter acquisition and edge adaptive preprocessing; to greatly reduce emergency fault response latency based on dual-priority hierarchical uploading and edge local linkage; and to achieve intelligent prediction by employing cloud trend prediction and hybrid decision-making, thus avoiding notification loss due to computational delays. It effectively identifies response deviations caused by abnormal operations or equipment aging, supports flexible expansion of more terminal types and notification channels, and is suitable for application scenarios with stringent requirements for power safety and continuous power supply, such as industrial power distribution networks, smart parks, data centers, and high-rise buildings, offering significant economic benefits.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a communication method for intelligent circuit breakers based on edge computing and the cloud, comprising the following steps: S1, Multi-dimensional electrical parameter acquisition: The digital miniature circuit acquires current, voltage, leakage current, power, temperature and switch status parameters at a preset reference frequency to generate the original time-series data stream; S2, Edge Adaptive Preprocessing: The original time-series data stream is received through the smart gateway, and anomaly detection is performed. When the deviation of any parameter exceeds the first-level dynamic threshold of the local cache, a high-frequency sampling mode is triggered and multi-parameter joint analysis is started to generate an anomaly feature vector. When the deviation does not exceed the first-level dynamic threshold, the original time-series data stream is aggregated and compressed to generate a normal statistical package. S3, Hierarchical Upload: The abnormal feature vector is uploaded to the edge analysis node with the first priority channel, and the normal statistical package is uploaded to the edge analysis node with the second priority channel; S4, Edge Prediction and Local Linkage: The edge analysis node performs real-time pattern matching on the abnormal feature vector based on the preset rule engine. If an emergency fault mode is matched, a local linkage command is directly sent to the digital micro-power line to execute an emergency power outage, and the linkage result is synchronized to the cloud platform. If a non-emergency abnormal mode is matched, the abnormal feature vector is forwarded to the cloud platform. S5, Cloud-based Deep Prediction: Through a trend prediction model trained on historical data on a cloud platform, non-urgent abnormal patterns are analyzed to generate prediction results and suggested intervention strategies. S6, Hybrid Decision Communication: When the predicted result reaches the trigger condition for manual intervention, the cloud platform sends a decision request message, including an event identifier, suggested intervention strategy, decision countdown, and current status snapshot, to several management terminals in parallel. Based on a first-come, first-served decision token mechanism, the instruction from the first responding management terminal is taken as a valid decision instruction. If no valid decision instruction is received within the decision countdown, the cloud platform reads the current switch status of the digital mini-power circuit breaker before executing the intervention instruction. If the current status is open, the intervention instruction is converted into a fault locking instruction; if the current status is closed, the intervention instruction is automatically sent to the digital mini-power circuit breaker to power off. Specifically, the decision request message is sent using a broadcast-contention mechanism. After receiving the decision request message, if several management terminals submit a response within their local decision countdown, they send a token request frame to the cloud platform. The cloud platform identifies the management terminal corresponding to the first valid token request frame as the decision-leading terminal and sends an observation mode instruction to the other management terminals. S7, Multi-channel Heterogeneous Notification: Based on the fault level, the historical reachability matrix of the management terminal, and the current online status, dynamically select at least one notification channel combination from SMS, telephone, APP push, WeChat message, and platform alarm within a preset time threshold, and send alarm notifications to the corresponding management terminal; if the channel selection calculation cannot be completed within the time threshold, the preset default channel combination is used directly for notification sending; specifically, the dynamic selection of notification channel combination includes: calculating the expected reachability probability of each channel at the current time point based on the historical reachability matrix of the management terminal, selecting the channel with the highest expected reachability probability as the main notification channel, and selecting at least one channel with the second highest expected reachability probability as a backup notification channel; if the main notification channel does not receive a reply within a preset confirmation time limit, it automatically switches to the backup notification channel; S8, Closed-loop feedback: After the digital micro-power circuit breaker performs an intervention operation, it sends the operation execution result and the electrical parameter change curve before and after the operation back to the cloud platform. The cloud platform updates the trend prediction model and the first-level dynamic threshold based on the operation execution result, and periodically synchronizes the updated first-level dynamic threshold to the local cache of the smart gateway. The cloud platform is also connected to the central management center, which is used to centrally monitor the operation status and abnormal events of the digital micro-power circuit breaker.

[0007] Preferably, in step S2, the multi-parameter joint analysis includes: constructing a current-temperature joint deviation index and a power-voltage fluctuation coupling coefficient; when the joint deviation index exceeds a comprehensive threshold, it is determined to be a suspected overload fault and a corresponding abnormal feature vector is generated.

[0008] Preferably, in step S4, the emergency fault mode includes short circuit protection triggering, leakage current protection triggering, and over-temperature protection triggering; the non-emergency abnormal mode includes overload trend, voltage sag trend, and power factor abnormal trend.

[0009] Preferably, step S8 further includes generating an operation and response digital fingerprint on the cloud platform, obtaining the operating condition label of the load carried by the current digital micro-transistor, comparing the digital fingerprint with fingerprints with the same operating condition label in the historical successful case fingerprint database, and sending a secondary confirmation alarm to the central management center if the deviation exceeds the threshold.

[0010] To achieve the above objectives, the present invention also provides the following technical solution: A smart circuit breaker communication system based on edge computing and cloud computing includes: The acquisition layer, including digital micro-circuit breakers, is used to acquire current, voltage, leakage current, power, temperature and switch status parameters at a preset reference frequency to generate raw timing data streams. The transport layer, including a smart gateway, is connected to the digital micro-disconnect communication layer. It receives the raw time-series data stream and performs anomaly detection. When the deviation of any parameter exceeds the locally cached first-level dynamic threshold, it triggers a high-frequency sampling mode and initiates multi-parameter joint analysis to generate an anomaly feature vector. When the deviation does not exceed the first-level dynamic threshold, it aggregates and compresses the raw time-series data stream to generate a normal statistical packet. The smart gateway is also used to upload the anomaly feature vector via a first-priority channel and the normal statistical packet via a second-priority channel. Furthermore, the smart gateway periodically receives and caches the first-level dynamic threshold synchronized by the cloud platform. The edge analysis layer, including edge analysis nodes, is communicatively connected to the smart gateway and is used to perform real-time pattern matching on abnormal feature vectors based on a preset rule engine. If an emergency fault mode is matched, a local linkage command is directly sent to the digital micro-power circuit breaker to execute an emergency power outage, and the linkage result is synchronized to the cloud platform. If a non-emergency abnormal mode is matched, the abnormal feature vector is forwarded to the cloud platform. The cloud-based analytics layer, including a cloud platform, is communicatively connected to the edge analytics nodes and is used to perform evolutionary analysis on non-urgent anomaly patterns based on a trend prediction model trained on historical data, generating prediction results and suggested intervention strategies. A hybrid decision-making layer, deployed on the cloud platform, is used to send decision request messages, including event identifiers, suggested intervention strategies, decision countdowns, and current state snapshots, to several management terminals in parallel when the predicted result meets the conditions for manual intervention. Based on a first-come, first-served decision token mechanism, the instruction from the first responding management terminal is taken as a valid decision instruction. If no valid decision instruction is received within the decision countdown, the cloud platform reads the current switch status of the digital mini-circuit breaker before executing the intervention instruction. If the current status is open, the intervention instruction is converted into a fault locking instruction; if the current status is closed, the intervention instruction is automatically sent to the digital mini-circuit breaker execution device to cut off power. The notification layer, deployed on the cloud platform, dynamically selects at least one combination of notification channels—SMS, telephone, APP push, WeChat message, and platform alarm—within a preset time threshold based on the fault level, the historical reachability matrix of the management terminal, and the current online status, and sends alarm notifications to the corresponding management terminals. If the channel selection calculation cannot be completed within the time threshold, the preset default channel combination is used directly for notification sending. The display layer includes a sub-management platform and a central management center. The display layer is communicatively connected to the cloud platform and is used to visually display electrical parameters, abnormal events, and intervention operation records. A closed-loop feedback layer, deployed between the digital micro-circuit and the cloud platform, is used to transmit the operation execution results of the intervention operation and the change curves of electrical parameters before and after the operation back to the cloud platform, so that the cloud platform can update the trend prediction model and the first-level dynamic threshold. The cloud platform periodically synchronizes the updated first-level dynamic threshold to the local cache of the smart gateway.

[0011] Preferably, the smart gateway includes: The data classification module is used to divide the raw time-series data stream into emergency data stream, abnormal data stream, and normal data stream based on the parameter deviation. The adaptive compression module is used for lossless compression of emergency data streams, feature extraction compression of abnormal data streams, and statistical aggregation compression of normal data streams. The dual-channel scheduling module is used to allocate high-bandwidth, low-latency transmission resources to the first priority channel and normal transmission resources to the second priority channel. The threshold synchronization module is used to receive and cache the first-level dynamic threshold update packages periodically issued by the cloud platform, and to mark the version number of the currently effective threshold with a timestamp.

[0012] Preferably, the hybrid decision layer includes: The message generation module is used to encapsulate decision request messages; The token arbitration module is used to receive token request frames from each management terminal and identify the terminal corresponding to the first valid token request frame as the decision-making leading terminal. The timeout execution module is used to first call the status reading interface to obtain the current opening and closing status of the digital micro-circuit if no valid decision instruction is received when the decision countdown ends, and then decide to issue an intervention instruction or a fault lockout instruction based on the status.

[0013] Preferably, the notification layer includes: The reachability prediction module is used to calculate the expected reachability probability of each notification channel at the current time based on the historical reachability matrix of the management terminal. The dynamic scheduling module is used to select the channel with the highest expected reachability as the primary notification channel, select at least one channel with the second highest expected reachability as a backup notification channel, and automatically switch to the backup notification channel if the primary notification channel does not receive a confirmation within the preset confirmation time limit. The timeliness guarantee module is used to monitor the time spent on channel selection calculation. If the timeliness threshold is exceeded, the selection calculation will be interrupted and the preset default channel combination will be called.

[0014] Preferably, the closed-loop feedback layer further includes: A digital fingerprint generation module, deployed on the cloud platform, is used to generate operation and response digital fingerprints based on operation execution results and electrical parameter change curves. The operating condition identification module is used to obtain the operating condition label of the load currently carried by the digital miniature circuit breaker; The integrity verification module is used to compare the digital fingerprint with fingerprints with the same working condition label in the historical successful case fingerprint database. If the deviation exceeds the integrity threshold, a secondary confirmation alarm is sent to the central management center.

[0015] Preferably, the digital micro-segment includes: Multi-dimensional sensors, including current sensors, voltage sensors, leakage current sensors, power metering chips, and temperature sensors; The local execution module is used to execute emergency circuit breaker in response to local linkage commands, and to execute equipment power-off in response to intervention commands; The status feedback module is used to send the operation execution results and electrical parameter change curves back to the cloud platform; The status query interface is used by the cloud platform to read the current opening and closing status of the digital micro-circuit after a timeout decision.

[0016] Compared with the prior art, the beneficial effects of the present invention are: (1) This invention is a communication method and system for intelligent circuit breakers based on edge computing and cloud computing. By synchronously collecting six-dimensional parameters of current, voltage, leakage current, power, temperature and switch status at a preset reference frequency, the original time-series data stream is generated, which overcomes the limitations of single parameter monitoring. The intelligent gateway can perform anomaly detection at the edge. It only triggers high-frequency sampling and multi-parameter joint analysis when the parameter deviation exceeds the first-level dynamic threshold. Under normal conditions, it performs aggregation compression, which greatly reduces the amount of invalid data uploaded and reduces network bandwidth occupation, while ensuring the timely capture of high-density data under abnormal operating conditions. (2) This invention is a communication method and system for intelligent circuit breakers based on edge computing and cloud computing. By setting a first priority channel and a second priority channel, it ensures that key abnormal data is uploaded with priority in low latency and high bandwidth resources. The edge analysis node has a built-in rule engine that can directly identify emergency fault modes such as short circuit, leakage, and overheating locally and issue circuit breaker commands in real time without waiting for cloud feedback. This reduces the emergency response delay from seconds to milliseconds, significantly improving the power safety protection capability. (3) This invention is a communication method and system for intelligent circuit breakers based on edge computing and cloud computing. By setting up a trend prediction model trained on historical data of the cloud platform, it performs evolutionary analysis on non-emergency abnormal modes, generates prediction strategies, and gives the system forward-looking maintenance capabilities. It supports multiple management terminals to receive decision requests in parallel and determines the decision-making terminal based on the first-come-first-served principle, ensuring the timeliness of decision-making while being compatible with human authoritative judgment. In the case of timeout, the system automatically reads the current status of the switch and intelligently converts it into an intervention command or fault locking command, effectively preventing repeated operations or misoperations, and significantly improving the robustness and safety of decision-making in complex scenarios. (4) This invention is a communication method and system for intelligent circuit breakers based on edge computing and cloud computing. By dynamically calculating the expected reachability probability of multiple channels such as SMS, telephone, APP push, WeChat message, and platform alarm based on the historical reachability matrix of the management terminal and the current online status, the main channel and backup channel are selected in the best way and automatically switched within the preset confirmation time limit, which significantly improves the delivery success rate of alarm notifications. If the channel selection calculation timeout, the default channel combination is activated as a backup guarantee, which takes into account the intelligence and timeliness of the notification and avoids the loss of notification due to calculation delay. (5) This invention is a communication method and system for intelligent circuit breakers based on edge computing and cloud computing. The operation execution results and the change curves of electrical parameters before and after are transmitted back to the cloud platform through digital micro-circuit to dynamically update the trend prediction model and the first-level dynamic threshold. It is also periodically synchronized to the local cache of the intelligent gateway, so that the system has the ability to learn online and optimize adaptively. The cloud platform generates digital fingerprints of operation and response and compares them with the fingerprint database of historical successful cases under the same working condition label. If the deviation exceeds the threshold, a secondary confirmation alarm is sent to the central management center to effectively identify response deviations caused by abnormal operation or equipment aging. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the framework structure of the system of the present invention; Figure 3 This is a schematic diagram of the intelligent gateway structure of the system of the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0019] Example 1 Please see Figure 1 As shown, this invention proposes a communication method for intelligent circuit breakers based on edge computing and the cloud, including the following steps: S1, Multi-dimensional Electrical Parameter Acquisition: Current, voltage, leakage current, power, temperature, and switch status parameters are acquired by a digital miniature circuit breaker at a preset reference frequency to generate a raw time-series data stream. Specifically, the preset reference frequency is preferably the power frequency (40Hz~80Hz) with 128 or 256 sampling points per cycle to ensure that the raw data contains sufficient harmonic component information, providing high-fidelity input for subsequent joint analysis. The metering chip built into the digital miniature circuit breaker performs analog-to-digital conversion at this reference frequency and pushes the raw data packets with timestamps to the local buffer queue through the SPI or UART interface to ensure the continuity and low latency of the data stream. S2, Edge Adaptive Preprocessing: The original time-series data stream is received through the intelligent gateway, and anomaly detection is performed. When the deviation of any parameter exceeds the first-level dynamic threshold of the local cache, a high-frequency sampling mode is triggered and multi-parameter joint analysis is initiated to generate an anomaly feature vector. When the deviation does not exceed the first-level dynamic threshold, the original time-series data stream is aggregated and compressed to generate a normal statistical package. Specifically, the deviation calculation adopts the sliding window Z-score normalization method based on time series and is compared with the dynamic threshold in real time. The high-frequency sampling mode refers to increasing the sampling frequency to 4 or 8 times the reference frequency for a duration of 200ms to capture transient features. The multi-parameter joint analysis specifically includes: constructing the current-temperature joint deviation index and the power-voltage fluctuation coupling coefficient. The normal statistical package calculates the mean, peak value, effective value, and standard deviation of each parameter within every 10 cycles and packages them, achieving a data compression ratio of up to 50:1, significantly reducing bandwidth usage under normal conditions. S3, Hierarchical Upload: The abnormal feature vector is uploaded to the edge analysis node via the first priority channel, and the normal statistical packet is uploaded to the edge analysis node via the second priority channel. Specifically, the first priority channel and the second priority channel can use different VLAN tags or different MQTT topics at the physical layer. In terms of transmission resource scheduling, the first priority channel enjoys an independent token bucket and bandwidth reservation, and its maximum transmission latency is strictly limited to within 50ms, while the second priority channel allows a maximum transmission latency of 500ms, thereby realizing a deterministic communication strategy of prioritizing urgent data. S4, Edge Prediction and Local Linkage: Edge analysis nodes perform real-time pattern matching on abnormal feature vectors based on a preset rule engine. If an emergency fault mode is matched, a local linkage command is directly sent to the digital mini-circuit breaker (DMB) to execute an emergency trip, and the linkage result is synchronized to the cloud platform. If a non-emergency abnormal mode is matched, the abnormal feature vector is forwarded to the cloud platform. Specifically, the rule engine uses the RETE algorithm for efficient matching, and the rule library is defined in JSON format and supports hot updates. In the determination of emergency fault modes, the short-circuit protection trigger condition is: the current change rate di / dt exceeds a set threshold (e.g., 100A / ms) within 1ms and the peak current exceeds 10 times the rated current; the leakage current protection trigger condition is: the residual current mutation exceeds 30mA and lasts for more than 40ms; the over-temperature protection trigger condition is: the temperature sensor detection value exceeds 90℃ and the temperature rise rate is >2℃ / s; the local linkage command is encapsulated as a standard MODBUS or IEC 61850 general service command, and the tripping action is completed within 5ms by the DM's actuator. S5, Cloud-based Deep Prediction: This feature utilizes a trend prediction model trained on historical data within a cloud platform to analyze the evolution of non-emergency anomaly patterns, generating prediction results and suggested intervention strategies. Specifically, the trend prediction model employs a time-series prediction model combining LSTM (Long Short-Term Memory) networks and an attention mechanism. Input features include the frequency of voltage sags, power factor change slope, temperature accumulation effect, and load rate change curves within the past 72 hours. The model outputs the probability and time window that the anomaly pattern may evolve into an emergency fault within the next 30 minutes (e.g., an overload trend will trigger overheat protection after 25 minutes). The suggested intervention strategies include, but are not limited to, recommending load reduction, peak-shifting startup, line maintenance, and switching to backup power, and are appended to the prediction results in structured JSON format. S6, Hybrid Decision Communication: When the predicted result reaches the trigger condition for manual intervention, the cloud platform sends a decision request message, including an event identifier, suggested intervention strategy, decision countdown, and current status snapshot, to several management terminals in parallel. Based on a first-come, first-served decision token mechanism, the instruction from the first responding management terminal is taken as a valid decision instruction. If no valid decision instruction is received within the decision countdown, the cloud platform reads the current switch status of the digital micro-circuit breaker before executing the intervention instruction. If the current status is open, the intervention instruction is converted into a fault locking instruction; if the current status is closed, the intervention instruction is automatically sent to the digital micro-circuit breaker execution device. Specifically, the decision request message is sent using a broadcast-contention mechanism. After receiving the decision request message, several management terminals, if they submit a response within the local decision countdown, send a token request frame to the cloud platform. The cloud platform identifies the management terminal corresponding to the first valid token request frame as the decision-making leader terminal and sends an observation mode command to the other management terminals. Specifically, the manual intervention trigger condition is set to a prediction probability ≥ 85% or a remaining intervention time window ≤ 5 minutes. The decision request message is broadcast using a WebSocket full-duplex channel and carries a digital signature to prevent forgery. The local decision countdown is set to (global countdown - 2s) to compensate for network transmission delay. The fault lockout command includes an irrevocable lockout bit to prevent the digital micro-circuit from being accidentally closed remotely or locally before the fault is resolved. This lockout state will continue until the central management center manually unlocks it or a maintenance completion signal is received. The observation mode command causes non-leading terminals to enter a read-only monitoring interface, displaying the leader terminal's countdown and decision-making process, avoiding conflicts caused by simultaneous operation of multiple terminals. S7, Multi-channel Heterogeneous Notification: Based on the fault level, the historical reachability matrix of the management terminal, and the current online status, dynamically select at least one notification channel combination from SMS, telephone, APP push, WeChat message, and platform alarm within a preset time threshold, and send alarm notifications to the corresponding management terminal; if the channel selection calculation cannot be completed within the time threshold, the preset default channel combination is directly used for notification sending; specifically, the dynamic selection of notification channel combination includes: calculating the expected reachability probability of each channel at the current time point based on the historical reachability matrix of the management terminal, selecting the channel with the highest expected reachability probability as the main notification channel, and selecting at least one channel with the second highest expected reachability probability as a backup notification channel; if the main notification channel fails to reach the desired channel, the channel selection is performed as the primary notification channel. If no confirmation is received within the preset confirmation time limit, the system will automatically switch to the backup notification channel. Specifically, the historical reachability matrix is ​​an M×N two-dimensional table (M is the number of management terminals, and N is the channel type), storing the response rate, average response time, and packet loss rate of each channel every half hour within each working day. The expected reachability probability uses a weighted moving average method, giving higher weight to recent data (0.6 for this week, 0.3 for last week, and 0.1 for last month). The timeliness threshold is set to 3 seconds by default. If the channel selection calculation is not completed within this time, the preset default channel combination will be used immediately to avoid delays. The preset confirmation time limit is set to 20 seconds. If there is no confirmation from the primary channel within this time, the backup channel will immediately take over to ensure the reliability and timeliness of the notification. S8, Closed-loop feedback: After the digital micro-circuit interrupter (DSP) performs an intervention operation, it transmits the operation result and the electrical parameter change curves before and after the operation back to the cloud platform. The cloud platform updates the trend prediction model and the first-level dynamic threshold based on the operation result, and periodically synchronizes the updated first-level dynamic threshold to the local cache of the smart gateway. The cloud platform is also communicatively connected to a central management center, which centrally monitors the operating status and abnormal events of the DSP. Specifically, the electrical parameter change curves before and after the operation include raw waveform segments from 200ms before and 500ms after the operation, used for analyzing the moment of circuit breaker failure. Arc energy and system disturbances; the trend prediction model adopts an online incremental learning strategy, using each data transmission as a new training sample, and fine-tuning the model when computing resources are idle; the update of the first-level dynamic threshold adopts an adaptive sliding window method, using the parameter values ​​at the current abnormal moment as historical samples to recalculate the moving average and standard deviation, generating a new threshold (μ±3σ), with the synchronization cycle preset to once every 24 hours or triggered immediately when the threshold change exceeds 5%; the central management center has a global topology display function across regions and sites, and can present the health status of the entire network of digital micro-interrupts in a heat map and provide comprehensive situational awareness.

[0020] Furthermore, in step S2, the multi-parameter joint analysis includes: constructing a current-temperature joint deviation index and a power-voltage fluctuation coupling coefficient; when the joint deviation index exceeds a comprehensive threshold, it is determined to be a suspected overload fault and a corresponding abnormal feature vector is generated.

[0021] In this embodiment, the current-temperature joint deviation index The specific calculation formula is as follows: ,in , satisfy + =1 and dynamically adjusted based on the load inertia time constant; The power-voltage fluctuation coupling coefficient Defined as: ,when When the value is greater than 1.5, the load exhibits abnormal constant power characteristics, which may indicate a rectifier failure; when... A value less than 0.5 indicates abnormal impedance fluctuations, potentially indicating poor contact or arcing. The combined deviation index threshold is calibrated on-site and is set to 0.75 by default.

[0022] Furthermore, in step S4, the emergency fault mode includes short circuit protection triggering, leakage current protection triggering, and over-temperature protection triggering; the non-emergency abnormal mode includes overload trend, voltage sag trend, and power factor abnormal trend.

[0023] In this embodiment, the overload trend is specifically defined as follows: within a 120s rolling window, the average current continuously exceeds 80% of the rated value and the rate of increase every 20s is ≥2% of the rated current; the voltage sag trend is defined as follows: within 60 seconds, the effective voltage value drops below 85% of the rated value 3 or more times and lasts for more than 20ms; the power factor anomaly trend is defined as follows: within 10 minutes, the average power factor deviates from the historical baseline and the fluctuation variance exceeds 3 times the baseline; all of the above non-emergency modes are equipped with independent feature codes, which facilitates the cloud platform to perform classification aggregation analysis and historical case retrieval.

[0024] Furthermore, step S8 also includes generating an operation and response digital fingerprint on the cloud platform, obtaining the operating condition label of the load carried by the current digital micro-power unit, comparing the digital fingerprint with fingerprints with the same operating condition label in the historical successful case fingerprint database, and if the deviation exceeds the threshold, sending a secondary confirmation alarm to the central management center.

[0025] In this embodiment, the operation and response digital fingerprint is generated by hash encoding of multi-dimensional features such as circuit breaker action time, di / dt characteristics at the tripping moment, arc duration, recovery voltage establishment time, and temperature change slope, resulting in a 64-bit hexadecimal string. The operating condition label includes the load category (such as motor, rectifier, resistive) and the current operating stage (such as startup, steady state, shutdown). The deviation is calculated using Hamming distance to determine the difference in the number of bits of the digital fingerprint. The secondary confirmation alarm includes details of the fingerprint deviation, possible cause speculation (such as load change, aging anomaly, or algorithm misjudgment), and suggested manual review operation guidelines, which are highlighted in red on the large screen of the central management center until manual confirmation is completed.

[0026] Example 2 Please see Figure 2-3 As shown, this invention proposes an intelligent circuit breaker communication system based on edge computing and the cloud, comprising: The acquisition layer, including a digital micro-circuit, is used to acquire current, voltage, leakage current, power, temperature, and switching status parameters at a preset reference frequency, generating raw timing data streams. Specifically, the digital micro-circuit in the acquisition layer embeds a dual-core architecture main control chip (such as Cortex-A). + Cortex- ),in The core is responsible for the high-precision measurement algorithm. The core is responsible for the communication protocol stack and real-time response tasks; each sensor has undergone factory temperature compensation calibration and has a self-diagnostic function. When a sensor self-test is abnormal, it will actively report an alarm flag. The transport layer, including the smart gateway, is connected to the digital micro-disconnect communication network. It receives raw time-series data streams and performs anomaly detection. When the deviation of any parameter exceeds the first-level dynamic threshold in the local cache, it triggers a high-frequency sampling mode and initiates multi-parameter joint analysis to generate anomaly feature vectors. When the deviation does not exceed the first-level dynamic threshold, it aggregates and compresses the raw time-series data streams to generate normal statistical packets. The smart gateway is also used to upload the anomaly feature vectors via the first priority channel and the normal statistical packets via the second priority channel. The smart gateway is also used to periodically receive and cache the first-level dynamic threshold synchronized by the cloud platform. Specifically, the smart gateway hardware is equipped with an AI acceleration coprocessor (such as an NPU or TPU) to accelerate matrix operations for anomaly detection and multi-parameter joint analysis. Its local cache uses non-volatile MRAM to ensure that the first-level dynamic threshold is not lost after power failure. At the same time, the gateway has a built-in hardware watchdog that can automatically reset within 200ms when the main program malfunctions. After the reset, it immediately restores the threshold from the local cache and reconnects to the cloud platform. The edge analysis layer, including edge analysis nodes, communicates with the smart gateway and is used to perform real-time pattern matching on abnormal feature vectors based on a preset rule engine. If an emergency fault mode is matched, a local linkage command is directly sent to the digital micro-power line to execute an emergency power outage, and the linkage result is synchronized to the cloud platform. If a non-emergency abnormal mode is matched, the abnormal feature vector is forwarded to the cloud platform. Specifically, the edge analysis nodes are deployed in a containerized manner on industrial-grade edge servers close to the smart gateway, supporting Docker / K3s orchestration. Their rule engine supports dynamic loading of rule packages, which are uniformly compiled, signed, and distributed by the cloud platform. Each node maintains a local image library, retaining the matching records of the most recent 1,000 abnormal feature vectors for rapid repetitive pattern recognition and rule hit rate statistics. The cloud analytics layer, including the cloud platform, communicates and connects with edge analytics nodes. It is used to perform evolutionary analysis of non-urgent abnormal patterns based on trend prediction models trained on historical data, generating prediction results and suggested intervention strategies. Specifically, the cloud platform is deployed on a hybrid cloud architecture, the core inference service is accelerated using a GPU cluster, the data storage layer uses a time-series database (such as InfluxDB) and object storage separation architecture, and the training job of the trend prediction model adopts a distributed Spark ML pipeline, which automatically performs incremental training every morning and generates an A / B test version. When the AUC-ROC improvement on the validation set exceeds 2%, it is automatically released in a canary release. The hybrid decision-making layer, deployed on a cloud platform, is used to send decision request messages in parallel to multiple management terminals when the predicted result meets the conditions for manual intervention. These messages include an event identifier, suggested intervention strategy, decision countdown, and a current status snapshot. Based on a first-come, first-served decision token mechanism, the first instruction from a responding management terminal is considered the valid decision instruction. If no valid decision instruction is received within the decision countdown, the cloud platform reads the current switch status of the digital mini-circuit before executing the intervention instruction. If the current status is open, the intervention instruction is converted into a fault locking instruction; if the current status is closed, the intervention instruction is automatically sent to the digital mini-circuit execution device to cut off power. Specifically, the hybrid decision-making layer's request messages are serialized using Protocol Buffers to reduce size and use the QUIC protocol for low-latency transmission to reduce handshake overhead. The decision countdown can be configured differently on the cloud platform console according to the fault level (60s for emergency cases, 300s for general cases). The system also supports complete audit traceability of decision logs, recording the receiving timestamp, terminal IP, decision-maker ID, and instruction content of each token request frame, meeting the safety audit requirements of the power industry. The notification layer, deployed on the cloud platform, dynamically selects at least one notification channel combination from SMS, telephone, APP push, WeChat message, and platform alarm within a preset time threshold based on the fault level, the historical reachability matrix of the management terminal, and the current online status, and sends alarm notifications to the corresponding management terminals. If the optimal channel selection cannot be completed within the time threshold, the preset default channel combination is used directly for notification sending. Specifically, the notification layer includes an independent channel health monitoring microservice that actively detects the API response latency and error codes of each third-party channel (SMS gateway, WeChat interface, telephone voice platform) every 30 seconds and feeds back the health status to the reachability matrix in real time. The preset default channel combination is set according to the fault level: Class A faults (urgent) default to telephone + SMS dual channels, Class B faults (important) default to APP push + WeChat message, and Class C faults (general) default to platform in-site alarm. The display layer, comprising sub-management platforms and a central management center, communicates with the cloud platform to visualize electrical parameters, abnormal events, and intervention records. Specifically, the sub-management platforms are compatible with both mobile and PC devices, using ECharts and WebGL to render electrical parameter curves in real time and display 3D topology maps. The central management center is equipped with a large-screen visualization system that supports multi-dimensional filtering (by region, by site, by time, and by fault type) and can export statistical reports and compliance reports with one click. The display layer maintains a long-term connection with the cloud platform using WebSocket to ensure that the push latency of alarm pop-ups is less than 1 second. The closed-loop feedback layer, deployed between the digital micro-network and the cloud platform, is used to transmit the operation execution results and the electrical parameter change curves before and after the operation back to the cloud platform. This allows the cloud platform to update the trend prediction model and the first-level dynamic threshold. The cloud platform periodically synchronizes the updated first-level dynamic threshold to the local cache of the smart gateway. Specifically, the data upload of the closed-loop feedback layer adopts a breakpoint resume mechanism. When the network is temporarily interrupted, the operation results and waveform data are temporarily stored in the local Flash of the digital micro-network. After the network is restored, they are re-transmitted in the order of timestamps. Each time the cloud platform receives feedback, it will perform data integrity verification. Only after the verification is passed will the model update process be triggered to ensure that the training data is clean and reliable.

[0027] Please see Figure 3 As shown, further, the smart gateway includes: The data classification module is used to divide the raw time-series data stream into emergency data stream, abnormal data stream, and normal data stream based on the parameter deviation. The adaptive compression module is used for lossless compression of emergency data streams, feature extraction compression of abnormal data streams, and statistical aggregation compression of normal data streams. The dual-channel scheduling module is used to allocate high-bandwidth, low-latency transmission resources to the first priority channel and normal transmission resources to the second priority channel. The threshold synchronization module is used to receive and cache the first-level dynamic threshold update packages periodically issued by the cloud platform, and to mark the version number of the currently effective threshold with a timestamp.

[0028] In this embodiment, the adaptive compression module uses the LZMA lossless compression algorithm for emergency data streams to ensure no critical waveform details are lost; for abnormal data streams, wavelet transform is used to extract key feature points (peaks, troughs, abrupt changes) for compression, achieving a compression ratio of up to 10:1; and for normal data streams, dead-zone compression (i.e., only recording changes outside the dead zone) is used, achieving a compression ratio of up to 100:1. The threshold synchronization module immediately sends an acknowledgment frame (ACK) back to the cloud platform after each update. If the cloud platform does not receive the ACK, it retransmits after 10 seconds until successful acknowledgment.

[0029] Furthermore, the hybrid decision-making layer includes: The message generation module is used to encapsulate decision request messages; The token arbitration module is used to receive token request frames from each management terminal and identify the terminal corresponding to the first valid token request frame as the decision-making leading terminal. The timeout execution module is used to first call the status reading interface to obtain the current opening and closing status of the digital micro-circuit if no valid decision instruction is received when the decision countdown ends, and then decide to issue an intervention instruction or a fault lockout instruction based on the status.

[0030] In this embodiment, the token arbitration module is implemented using an atomic counter + distributed lock (based on the Redis Redlock algorithm) to ensure strong consistency of token arbitration when deployed across availability zones. A valid token request frame must meet the following requirements: ① it contains the correct session ID; ② the digital signature verification is successful; ③ the timestamp is within the allowable deviation range (±500ms). When the timeout execution module calls the status read interface, if the network times out, it will retry 3 times, with an interval of 200ms each time. If it still fails, it will issue a trip command by default according to the security priority principle.

[0031] Furthermore, the notification layer includes: The reachability prediction module is used to calculate the expected reachability probability of each notification channel at the current time based on the historical reachability matrix of the management terminal. The dynamic scheduling module is used to select the channel with the highest expected reachability as the primary notification channel, select at least one channel with the second highest expected reachability as a backup notification channel, and automatically switch to the backup notification channel if the primary notification channel does not receive a confirmation within the preset confirmation time limit. The timeliness guarantee module is used to monitor the time spent on channel selection calculation. If the timeliness threshold is exceeded, the selection calculation will be interrupted and the preset default channel combination will be called.

[0032] In this embodiment, the reachability prediction module pre-calculates and caches the expected probability matrix for the next time period every 5 minutes to reduce the calculation delay when the notification is triggered; the dynamic scheduling module supports dynamic weighting of channel combinations. For example, when the main channel is APP push, the backup channels can simultaneously select SMS and phone calls to form an umbrella notification structure with one main and multiple backups; the timeliness guarantee module has a built-in real-time timer. When the time reaches 90% of the timeliness threshold, an early warning signal is issued. When it reaches 100%, it immediately interrupts and switches to the default channel combination. At the same time, the timeout event is recorded for subsequent optimization of the channel selection algorithm.

[0033] Furthermore, the closed-loop feedback layer also includes: The digital fingerprint generation module, deployed on a cloud platform, is used to generate digital fingerprints of operations and responses based on the operation execution results and electrical parameter change curves. The operating condition identification module is used to obtain the operating condition label of the load currently carried by the digital miniature circuit breaker; The integrity verification module compares the digital fingerprint with fingerprints in the historical successful case fingerprint database that have the same working condition label. If the deviation exceeds the integrity threshold, a secondary confirmation alarm is sent to the central management center.

[0034] In this embodiment, the digital fingerprint generation module uses the perceptual hash algorithm (pHash) to extract hash features after waveform visualization, which has robustness against small waveform drift. The operating condition identification module automatically identifies the load type through current harmonic spectrum analysis (FFT) during steady-state periods and extracts features such as start-stop frequency and running time ratio as auxiliary labels. The deviation calculation of the integrity verification module uses normalized Hamming distance, and the integrity threshold is preset to 0.15. When the deviation exceeds this value and there are more than 3 highly similar (deviation < 0.05) pre-fault cases in the historical case library, the system will automatically prioritize pushing highly similar historical handling solutions to assist the central management center in decision-making.

[0035] Furthermore, digital micro-segments include: Multi-dimensional sensors, including current sensors, voltage sensors, leakage current sensors, power metering chips, and temperature sensors; The local execution module is used to execute emergency circuit breaker in response to local linkage commands, and to execute equipment power-off in response to intervention commands; The status feedback module is used to send the operation execution results and electrical parameter change curves back to the cloud platform; The status query interface is used by the cloud platform to read the current opening and closing status of the digital micro-circuit after a timeout decision.

[0036] In this embodiment, the current sensor adopts a closed-loop Hall effect sensor with a range of 0~1000A and an accuracy class of 0.5; the voltage sensor adopts a resistive-capacitive voltage divider type with an input impedance ≥2MΩ; the leakage current sensor adopts a zero-sequence current transformer with a minimum detection leakage current of 1mA; the temperature sensor adopts an NTC thermistor and is mounted on the copper busbar connection; the tripping mechanism of the local execution module is an electromagnetic bistable structure with an action response time ≤5ms and a mechanical position indicator window; the status query interface supports both synchronous query (Modbus function code 0x02) and asynchronous subscription (automatic reporting via periodic status messages) modes to adapt to scenarios with different latency requirements.

[0037] 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 communication method for intelligent circuit breakers based on edge computing and the cloud, characterized in that, Includes the following steps: S1, Multi-dimensional electrical parameter acquisition: The digital miniature circuit acquires current, voltage, leakage current, power, temperature and switch status parameters at a preset reference frequency to generate the original time-series data stream; S2, Edge Adaptive Preprocessing: The original time-series data stream is received through the smart gateway, and anomaly detection is performed. When the deviation of any parameter exceeds the first-level dynamic threshold of the local cache, a high-frequency sampling mode is triggered and multi-parameter joint analysis is started to generate an anomaly feature vector. When the deviation does not exceed the first-level dynamic threshold, the original time-series data stream is aggregated and compressed to generate a normal statistical package; S3, Hierarchical Upload: The abnormal feature vector is uploaded to the edge analysis node with the first priority channel, and the normal statistical package is uploaded to the edge analysis node with the second priority channel; S4, Edge Prediction and Local Linkage: The edge analysis node performs real-time pattern matching on the abnormal feature vector based on the preset rule engine. If an emergency fault mode is matched, a local linkage command is directly sent to the digital micro-power line to execute an emergency power outage, and the linkage result is synchronized to the cloud platform. If a non-emergency abnormal mode is matched, the abnormal feature vector is forwarded to the cloud platform. S5, Cloud-based Deep Prediction: Through a trend prediction model trained on historical data on a cloud platform, non-urgent abnormal patterns are analyzed to generate prediction results and suggested intervention strategies. S6, Hybrid Decision Communication: When the prediction result reaches the trigger condition for manual intervention, the cloud platform sends a decision request message including an event identifier, suggested intervention strategy, decision countdown, and current status snapshot to several management terminals in parallel; based on the first-come, first-served decision token mechanism, the instruction of the first responding management terminal is taken as the valid decision instruction; If no valid decision instruction is received within the decision countdown, the cloud platform will first read the current switch status of the digital miniature circuit breaker before executing the intervention instruction. If the current status is open, the intervention instruction will be converted into a fault lockout instruction. If the current status is closed, the intervention instruction will be automatically sent to the digital miniature circuit breaker execution device to cut off the power. S7, Multi-channel heterogeneous notification: Based on the fault level, the historical reachability matrix of the management terminal and the current online status, dynamically select at least one notification channel combination from SMS, telephone, APP push, WeChat message and platform alarm within the preset time threshold, and send alarm notification to the corresponding management terminal; if the channel selection calculation cannot be completed within the time threshold, the preset default channel combination is directly used to send the notification. S8, Closed-loop feedback: After the digital micro-power circuit breaker performs an intervention operation, it sends the operation execution result and the electrical parameter change curve before and after the operation back to the cloud platform. The cloud platform updates the trend prediction model and the first-level dynamic threshold based on the operation execution result, and periodically synchronizes the updated first-level dynamic threshold to the local cache of the smart gateway. The cloud platform is also connected to the central management center, which is used to centrally monitor the operation status and abnormal events of the digital micro-power circuit breaker.

2. The intelligent circuit breaker communication method based on edge computing and cloud as described in claim 1, characterized in that: In step S2, the multi-parameter joint analysis includes: constructing a current-temperature joint deviation index and a power-voltage fluctuation coupling coefficient. When the joint deviation index exceeds the comprehensive threshold, it is determined to be a suspected overload fault and a corresponding abnormal feature vector is generated.

3. The intelligent circuit breaker communication method based on edge computing and cloud as described in claim 1, characterized in that: In step S4, the emergency fault modes include short circuit protection triggering, leakage current protection triggering, and over-temperature protection triggering; the non-emergency abnormal modes include overload trend, voltage sag trend, and power factor abnormal trend.

4. The intelligent circuit breaker communication method based on edge computing and cloud as described in claim 1, characterized in that: Step S8 also includes generating an operation and response digital fingerprint on the cloud platform, obtaining the operating condition label of the load carried by the current digital micro-segment, comparing the digital fingerprint with fingerprints with the same operating condition label in the historical successful case fingerprint database, and sending a secondary confirmation alarm to the central management center if the deviation exceeds the threshold.

5. A system for a smart circuit breaker communication method based on edge computing and cloud computing as described in any one of claims 1 to 4, characterized in that, include: The acquisition layer, including digital micro-circuit breakers, is used to acquire current, voltage, leakage current, power, temperature and switch status parameters at a preset reference frequency to generate raw timing data streams. The transport layer, including a smart gateway, is connected to the digital micro-disconnect communication layer. It is used to receive the raw time-series data stream and perform anomaly detection. When the deviation of any parameter exceeds the first-level dynamic threshold of the local cache, a high-frequency sampling mode is triggered and multi-parameter joint analysis is started to generate an anomaly feature vector. When the deviation does not exceed the first-level dynamic threshold, the raw time-series data stream is aggregated and compressed to generate a normal statistical packet. The smart gateway is also used to upload the anomaly feature vector with a first priority channel and the normal statistical packet with a second priority channel. The smart gateway is also used to periodically receive and cache the first-level dynamic threshold synchronized by the cloud platform; The edge analysis layer, including edge analysis nodes, is communicatively connected to the smart gateway and is used to perform real-time pattern matching on abnormal feature vectors based on a preset rule engine. If an emergency fault mode is matched, a local linkage command is directly sent to the digital micro-power circuit breaker to execute an emergency power outage, and the linkage result is synchronized to the cloud platform. If a non-emergency abnormal mode is matched, the abnormal feature vector is forwarded to the cloud platform. The cloud-based analytics layer, including a cloud platform, is communicatively connected to the edge analytics nodes and is used to perform evolutionary analysis on non-urgent anomaly patterns based on a trend prediction model trained on historical data, generating prediction results and suggested intervention strategies. A hybrid decision-making layer, deployed on the cloud platform, is used to send decision request messages, including event identifiers, suggested intervention strategies, decision countdowns, and current state snapshots, to several management terminals in parallel when the prediction result meets the conditions for manual intervention. Based on a first-come, first-served decision token mechanism, the instruction from the first responding management terminal is taken as the valid decision instruction. If no valid decision instruction is received within the decision countdown, the cloud platform will first read the current switch status of the digital miniature circuit breaker before executing the intervention instruction. If the current status is open, the intervention instruction will be converted into a fault lockout instruction. If the current status is closed, the intervention instruction will be automatically sent to the digital miniature circuit breaker execution device to cut off the power. The notification layer, deployed on the cloud platform, dynamically selects at least one notification channel combination from SMS, telephone, APP push, WeChat message and platform alarm within a preset time threshold based on the fault level, the historical reachability matrix of the management terminal and the current online status, and sends alarm notifications to the corresponding management terminals. If the channel selection calculation cannot be completed within the time limit, the preset default channel combination will be used directly for notification sending. The display layer includes a sub-management platform and a central management center. The display layer is communicatively connected to the cloud platform and is used to visually display electrical parameters, abnormal events, and intervention operation records. A closed-loop feedback layer, deployed between the digital micro-circuit and the cloud platform, is used to transmit the operation execution results of the intervention operation and the change curves of electrical parameters before and after the operation back to the cloud platform, so that the cloud platform can update the trend prediction model and the first-level dynamic threshold. The cloud platform periodically synchronizes the updated first-level dynamic threshold to the local cache of the smart gateway.

6. The intelligent circuit breaker communication system based on edge computing and cloud computing according to claim 4, characterized in that, The smart gateway includes: The data classification module is used to divide the raw time-series data stream into emergency data stream, abnormal data stream, and normal data stream based on the parameter deviation. The adaptive compression module is used for lossless compression of emergency data streams, feature extraction compression of abnormal data streams, and statistical aggregation compression of normal data streams. The dual-channel scheduling module is used to allocate high-bandwidth, low-latency transmission resources to the first priority channel and normal transmission resources to the second priority channel. The threshold synchronization module is used to receive and cache the first-level dynamic threshold update packages periodically issued by the cloud platform, and to mark the version number of the currently effective threshold with a timestamp.

7. The intelligent circuit breaker communication system based on edge computing and cloud computing according to claim 4, characterized in that, The hybrid decision-making layer includes: The message generation module is used to encapsulate decision request messages; The token arbitration module is used to receive token request frames from each management terminal and identify the terminal corresponding to the first valid token request frame as the decision-making leading terminal. The timeout execution module is used to first call the status reading interface to obtain the current opening and closing status of the digital micro-circuit if no valid decision instruction is received when the decision countdown ends, and then decide to issue an intervention instruction or a fault lockout instruction based on the status.

8. The intelligent circuit breaker communication system based on edge computing and cloud computing according to claim 4, characterized in that, The notification layer includes: The reachability prediction module is used to calculate the expected reachability probability of each notification channel at the current time based on the historical reachability matrix of the management terminal. The dynamic scheduling module is used to select the channel with the highest expected reachability as the primary notification channel, select at least one channel with the second highest expected reachability as a backup notification channel, and automatically switch to the backup notification channel if the primary notification channel does not receive a confirmation within the preset confirmation time limit. The timeliness guarantee module is used to monitor the time spent on channel selection calculation. If the timeliness threshold is exceeded, the selection calculation will be interrupted and the preset default channel combination will be called.

9. The intelligent circuit breaker communication system based on edge computing and cloud computing according to claim 4, characterized in that, The closed-loop feedback layer also includes: A digital fingerprint generation module, deployed on the cloud platform, is used to generate operation and response digital fingerprints based on operation execution results and electrical parameter change curves. The operating condition identification module is used to obtain the operating condition label of the load currently carried by the digital miniature circuit breaker; The integrity verification module is used to compare the digital fingerprint with fingerprints with the same working condition label in the historical successful case fingerprint database. If the deviation exceeds the integrity threshold, a secondary confirmation alarm is sent to the central management center.

10. The intelligent circuit breaker communication system based on edge computing and cloud computing according to claim 4, characterized in that, The digital micro-fraction includes: Multi-dimensional sensors, including current sensors, voltage sensors, leakage current sensors, power metering chips, and temperature sensors; The local execution module is used to execute emergency circuit breaker in response to local linkage commands, and to execute equipment power-off in response to intervention commands; The status feedback module is used to send the operation execution results and electrical parameter change curves back to the cloud platform; The status query interface is used by the cloud platform to read the current opening and closing status of the digital micro-circuit after a timeout decision.

Images