Iot-based pole-mounted circuit breaker primary and secondary fusion intelligent monitoring method

Abstract

The application discloses a pole-mounted circuit breaker primary and secondary fusion intelligent monitoring method based on Internet of Things and concretely relates to the technical field of circuit breakers; through integrated sensing collection, edge calculation, structured data communication and master station protocol analysis, in combination with feature parameters extracted from a physical layer and a protocol layer, an abnormal scoring model is constructed, quantitative evaluation of data frame legality risk and dynamic monitoring of resolver identification accuracy are realized; when accuracy decreases, the system can adaptively adjust the protocol verification logic according to the scoring result, and feedback the optimization rules to the master station system to form a continuous learning mechanism, thereby significantly improving the robustness of data identification and the intelligent level of power grid monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of circuit breakers, in particular to a primary and secondary integrated intelligent monitoring method for pole-mounted circuit breakers based on the Internet of Things. BACKGROUND

[0002] The primary and secondary integrated intelligent monitoring of pole-mounted circuit breakers refers to the integration of primary devices (such as switch bodies) and secondary devices (such as sensors, communication and control units) in the power system, real-time monitoring and data analysis of operating status, current, voltage, temperature and other parameters through intelligent means, thereby realizing fault early warning, remote control and state diagnosis, and improving the operating efficiency and fault response capability of the distribution network.

[0003] The prior art has the following shortcomings:

[0004] In the process of receiving data frames by the master station through protocol analysis services or SCADA systems, due to slight differences in the implementation of communication protocols (such as Modbus and IEC 104) by different manufacturers' devices, and the fact that the fault tolerance mechanism of the master station protocol analyzer is not strict, the malformed or spliced incorrect data frames may be misjudged as legal information, thereby causing false alarms, incorrect control and even shielding of real faults, especially in unstable wireless communication environments, which seriously threatens the accuracy and safety of power grid monitoring. SUMMARY

[0005] The purpose of the present application is to provide a primary and secondary integrated intelligent monitoring method for pole-mounted circuit breakers based on the Internet of Things to solve the problems in the background art.

[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical solution: a primary and secondary integrated intelligent monitoring method for pole-mounted circuit breakers based on the Internet of Things, comprising:

[0007] integrating primary devices and secondary devices on the pole-mounted circuit breaker;

[0008] collecting operating parameters of the pole-mounted circuit breaker using various sensors;

[0009] performing preliminary processing of the raw data through an embedded edge computing module to generate structured monitoring data;

[0010] selecting a communication mode to transmit the processed structured monitoring data to the distribution automation master station;

[0011] performing protocol analysis on the received data frames at the master station side, and judging the legality of the data frames based on the standard communication protocol;

[0012] extracting physical layer communication quality characteristic parameters and protocol layer behavior characteristic parameters from the received communication data;

[0013] An anomaly scoring model is established based on the extracted feature parameters. A legality risk assessment is performed on each data frame, and an anomaly score is output.

[0014] The judgment results of the main station protocol parser are compared with the abnormal score value to evaluate its accuracy in identifying malformed or spliced ​​data frames;

[0015] When the recognition accuracy is lower than the set threshold, the verification logic in the protocol parser is adaptively adjusted based on the output of the scoring model.

[0016] The adjusted protocol parsing rules are fed back to the main station system and continuously optimized in conjunction with subsequent communication data.

[0017] Preferably, the secondary equipment includes: a signal acquisition unit, a control execution unit, a communication and protocol processing unit, an embedded edge computing unit, and a power management module. All units are encapsulated in the control box or housing of the pole-mounted circuit breaker, forming an intelligent monitoring terminal.

[0018] Preferably, the signal acquisition unit includes a current sensor, a voltage sensor, a thermal temperature sensor, a humidity sensor, a displacement sensor, and an acceleration sensor, used to acquire the operating status, environmental parameters, and mechanical characteristic parameters of the circuit breaker.

[0019] Preferably, the failure rate of data retransmission requests from the main station within a fixed time period is analyzed to obtain abnormal values ​​for main station retransmission requests. Specifically:

[0020] Within a fixed time period T, the master station monitors N pole-mounted circuit breaker devices. For each device, the retransmission failure rate is recorded once. For each device i∈{1,2,...,N}, the retransmission failure feature x is extracted. i The expression is: in: This represents the number of retransmission failures of device i within the time period. x represents the total number of retransmission requests for device i within the time period. i ∈[0,1]: represents the retransmission failure rate of the device; construct a similarity graph G=(V,E), where each node v i ∈V represents a device; the similarity between nodes is defined using a distance function, and the edge weights w are used. ij ; where: w ij Let the degree matrix D∈R be the edge weight between nodes i and j. N×N :D ii =∑ j w ij R is the set of real numbers; the Laplace matrix L is: L = DW; where W is the edge weight matrix [w ij The expression for calculating the retransmission exception value requested by the main station is: in: WD represents the k nearest neighbors of node i; WD indicates that the master station requests a retransmission of the abnormal value, and k is the number of nearest neighbors.

[0021] Preferably, after analyzing the changes in the ratio of abnormal frames within a unit time window, anomaly values ​​for the increase in the ratio of abnormal frames are generated, specifically as follows:

[0022] The monitoring period is divided into n equal-length windows T1, T2, ..., T. n ; in each time window T n Within, record the abnormal frame rate 'a' of the device. n : in: For window T n The number of data frames that failed to resolve internally or were deemed illegal; For window T n The total number of data frames received internally;

[0023] Calculate the adjacent window increment Δa for the abnormal frame ratio i The expression is: Δa i =a i -a i-1 ,i=2,3,...,n; indicates that from window T i To T i-1 The increase in the ratio of abnormal frames between them, and the mean μ of the historical increase sequence. Δa With standard deviation σ Δa The abnormal frame rate increase is calculated using the following expression: SD represents the abnormal increase in the abnormal frame rate.

[0024] Preferably, an anomaly scoring model is established based on the extracted feature parameters. A legality risk assessment is performed on each data frame, and an anomaly score value is output. Specifically, the main station request retransmission anomaly value and the anomaly frame ratio increase anomaly value are converted into a comprehensive feature vector. The comprehensive feature vector is used as the input of the machine learning model. The machine learning model uses the prediction of anomaly score value labels for each set of comprehensive feature vectors as the prediction objective and minimizes the sum of prediction errors for all anomaly score value labels as the training objective. The machine learning model is trained until the sum of prediction errors converges, at which point the model training stops. The anomaly score value is determined based on the model output. The machine learning model is a multinomial regression model.

[0025] Preferably, the accuracy of the current protocol parser's identification is evaluated by comparing the parser's frame legality judgment results with the anomaly score based on feature parameters, specifically including:

[0026] For each received data frame F i The main station system executes the following two judgment mechanisms in parallel:

[0027] Path A: The protocol parser determines the frame validity label P based on its judgment. i ∈{0,1}, where 1 indicates that the frame is judged to be valid, and 0 indicates that the frame is invalid or abnormal;

[0028] Path B: Input the comprehensive feature vector of the time period to which the current frame belongs into the trained anomaly scoring model, and output the anomaly score value y. i ∈[0,1];

[0029] The frame risk level R is determined based on the set threshold θy. i ∈{0,1}, where 1 represents high risk and 0 represents normal;

[0030] For each frame, compare the protocol parser label P i With model output label R i Are they consistent?

[0031] If P i =R i Recorded as a consistent determination;

[0032] If P i ≠R i These are recorded as potential misjudgments and categorized as follows:

[0033] False positive type A: P i =1,R i =1 indicates that the protocol parser mistakenly judged a high-risk frame as legitimate;

[0034] False positive type B: P i =0,R i =0 indicates that the protocol parser misjudged a normal frame as abnormal;

[0035] Based on the frame determination results from the sliding time window, the accuracy of the protocol parser's frame legitimacy recognition is calculated using the following expression: Where: N is the total number of data frames received within the current time window; if the accuracy rate continues to be lower than the set threshold, the protocol parser will be triggered to make adaptive adjustments.

[0036] Preferably, the feature vector X of high-risk scoring frames is extracted from the scoring model; for each misjudged sample, its verification failure features and communication parameters are recorded to construct an abnormal frame feature set Ferr.

[0037] Feature sensitivity learning is performed using locally weighted logistic regression, with samples from Ferr used as the training set. Local fitting is performed on the error type and communication state of the protocol fields to model the misjudgment probability P(misjudgment|X) under a specific scenario, expressed as: in It is a vector inner product, which is each feature multiplied by its corresponding weight and then summed. v is the weight vector, which are the parameters learned during the training of the logistic regression model.

[0038] Based on the feature segments with high misclassification probability output by the local weighted logistic regression model, select the adjustable parameters in the corresponding verification logic;

[0039] The optimal adjustment direction is obtained by minimizing the misjudgment prediction loss;

[0040] The adjusted parser configuration is dynamically distributed to the main station protocol stack.

[0041] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0042] 1. The IoT-based intelligent monitoring method for pole-mounted circuit breakers integrates primary and secondary components. Addressing the issues of data misjudgment caused by differences in protocol implementation among different manufacturers and imperfect fault tolerance mechanisms at the master station, this invention constructs a complete closed-loop system from field intelligent terminal data acquisition, edge computing, and structured communication to master station parsing, feature extraction, and anomaly scoring. This system can accurately identify malformed frames and incorrectly spliced ​​frames in the communication link, significantly improving the accuracy of data frame legitimacy identification and the ability to handle communication anomalies in power distribution automation systems.

[0043] 2. By introducing an anomaly scoring model, comparing the protocol parsing judgment results with the model output, and automatically triggering a verification logic optimization mechanism based on machine learning algorithms when the accuracy is insufficient, this invention constructs the adaptive adjustment and continuous learning capability of the protocol parser, realizes dynamic optimization of parsing rules, continuous monitoring of anomaly risks and intelligent enhancement of system robustness, and improves the stable operation and safety monitoring level of pole-mounted circuit breakers in complex power grid environments. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0045] Figure 1 This is a mind map of the method of the present invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of 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, 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 scope of protection of the present invention.

[0047] For examples, please refer to Figure 1 As shown in this embodiment, the IoT-based pole-mounted circuit breaker primary and secondary integrated intelligent monitoring method includes:

[0048] Integrate primary and secondary equipment on the pole-mounted circuit breaker;

[0049] Various sensors are used to collect the operating parameters of pole-mounted circuit breakers;

[0050] The raw data is initially processed by an embedded edge computing module to generate structured monitoring data.

[0051] Select the communication method to transmit the processed structured monitoring data to the distribution automation master station;

[0052] On the main station side, the received data frames are parsed according to the protocol, and the legality of the data frames is judged based on the standard communication protocol;

[0053] Extract physical layer communication quality characteristic parameters and protocol layer behavioral characteristic parameters from the received communication data;

[0054] An anomaly scoring model is established based on the extracted feature parameters. A legality risk assessment is performed on each data frame, and an anomaly score is output.

[0055] The judgment results of the main station protocol parser are compared with the abnormal score value to evaluate its accuracy in identifying malformed or spliced ​​data frames;

[0056] When the recognition accuracy is lower than the set threshold, the verification logic in the protocol parser is adaptively adjusted based on the output of the scoring model.

[0057] The adjusted protocol parsing rules are fed back to the main station system and continuously optimized in conjunction with subsequent communication data.

[0058] In this embodiment of the invention, the pole-mounted circuit breaker adopts a primary and secondary integrated structure design, that is, based on the primary equipment (high-voltage side disconnection actuator) of the circuit breaker, a secondary equipment functional module is integrated to realize the integrated integration of sampling, control and communication functions, and to construct an intelligent monitoring terminal, specifically including:

[0059] The primary equipment is the pole-mounted high-voltage circuit breaker body, including:

[0060] Insulating support;

[0061] Breaking mechanisms (such as vacuum interrupters, magnetic blowout mechanisms, or SF6 switches);

[0062] Operating mechanism (electric, spring or permanent magnet mechanism);

[0063] Primary-side measuring elements such as current transformers and voltage transformers;

[0064] Its functions include fault interruption, load switching, and overcurrent protection.

[0065] Secondary equipment integration module

[0066] The secondary equipment module is encapsulated in an integrated terminal and installed on a circuit breaker control box or insulating support, and includes at least the following units:

[0067] a. Signal acquisition unit:

[0068] It collects data such as current, voltage, temperature, humidity, contact resistance, displacement, and vibration.

[0069] Supports multi-channel analog-to-digital conversion (ADC);

[0070] It can expand the fiber optic or radio temperature measurement channel to realize contact temperature rise monitoring.

[0071] b. Control Execution Unit:

[0072] Embedded microcontrollers (such as STM32, ARM Cortex-M series);

[0073] Execute remote / local opening and closing commands;

[0074] It has logic control capabilities to realize trip linkage control and time delay protection logic.

[0075] c. Communication and Protocol Processing Unit:

[0076] Supports communication methods such as RS485, Ethernet, LoRa, NB-IoT, and 4G;

[0077] It can run communication protocols such as DL / T 645, IEC 60870-5-104, and Modbus;

[0078] Built-in communication failure retry mechanism and frame integrity verification function.

[0079] d. Local edge computing unit:

[0080] It has edge computing capabilities and uses an embedded system to run device status algorithms;

[0081] Real-time processing of measurement data and prediction of equipment malfunctions;

[0082] Reduce communication load and optimize data transmission through compression.

[0083] e. Power Management and Backup Power Module:

[0084] Power is supplied by a pole-mounted PT power supply, combined with supercapacitors or lithium batteries for backup power.

[0085] To enable data retention and emergency communication during power outages.

[0086] All secondary equipment modules are integrated with the primary equipment housing or control box through a modular structural design; shielded cables are used, and electromagnetic compatibility design and protection level reach IP65 or above to adapt to outdoor high voltage environments.

[0087] In this embodiment of the invention, the intelligent monitoring terminal integrates multiple types of industrial-grade sensors to collect the operating parameters of the pole-mounted circuit breaker in real time. The collected data serves as the basis for status assessment, anomaly detection, and remote decision-making.

[0088] The current sensor uses an open-type Hall current transformer or a low-power shunt sensor and is installed at the input or output of the circuit breaker.

[0089] Voltage acquisition uses a high-impedance voltage divider module or a capacitive voltage sensor;

[0090] The collected data is used to determine the load operating status, short circuit faults, and undervoltage / overvoltage anomalies.

[0091] Thermistors (NTC / PT100) or digital temperature sensors are respectively installed in the contacts, conductive clamps, insulating supports and housing;

[0092] Used to monitor equipment temperature rise in real time and determine poor contact, overload, or abnormal environment;

[0093] It supports setting temperature threshold alarms to trigger remote prompts or tripping commands.

[0094] A capacitive relative humidity sensor is installed inside the housing;

[0095] Used to detect whether there is moisture or condensation risk in the circuit breaker cavity, and to judge the condensation trend in combination with temperature;

[0096] Abnormal humidity levels may indicate a seal failure or potential environmental corrosion.

[0097] Install a magnetic encoder or Hall displacement sensor on the shaft end of the operating mechanism;

[0098] Real-time monitoring of opening and closing stroke, speed, and position changes;

[0099] It is used to diagnose mechanical faults such as jamming, slow movement, overshoot, or incomplete positioning.

[0100] All sensor sampling data is accessed to the edge computing module via a multi-channel analog-to-digital converter (ADC) or digital I2C / SPI bus; different sampling periods are set (e.g., 50-100ms for electrical parameters, 15s for temperature and humidity), and noise filtering is performed; the data is packaged according to the set protocol and uploaded to the main station system via the wireless module.

[0101] Through the aforementioned multi-dimensional sensor layout and data acquisition mechanism, this invention achieves holographic perception of the operating status of pole-mounted circuit breakers, providing comprehensive raw data support for remote monitoring, anomaly diagnosis, and autonomous control.

[0102] In this embodiment of the invention, the pole-mounted circuit breaker intelligent monitoring terminal integrates an embedded edge computing module for real-time preprocessing and logical analysis of the collected raw sensor data to generate structured monitoring data, specifically including:

[0103] The module includes a low-power embedded processor (such as ARM Cortex-M4 / M7, STM32 series, RISC-V architecture);

[0104] It is equipped with a lightweight real-time operating system (such as FreeRTOS or uC / OS);

[0105] It includes a matching memory, an analog-to-digital converter (ADC), a data buffer, and a clock management unit.

[0106] Raw sensor data is sampled in real time using an ADC;

[0107] High-frequency interference is eliminated by using algorithms such as moving average and median filtering for high-frequency signals such as voltage and current.

[0108] Low-frequency signals such as temperature, humidity, and displacement are smoothed using a windowing process to ensure data stability.

[0109] The collected data is grouped and coded according to parameter type (such as current, voltage, temperature);

[0110] Each set of data includes a timestamp, device number, unit of measurement, and accuracy description.

[0111] The output is a standardized structure or a JSON format data packet, which is convenient for communication protocol parsing.

[0112] The built-in rule base can perform preliminary fault identification at the edge (such as: temperature exceeding the set threshold, abnormal current imbalance rate, etc.);

[0113] It supports setting an event triggering mechanism. When an abnormal status is detected, an event tag is generated and marked in the data frame for the main station to quickly judge and handle.

[0114] Structured data is lightly compressed (such as LZ4 or differential encoding) to reduce communication load; when communication is interrupted, data is cached in local memory (such as FRAM or external Flash) in time sequence and automatically sent back after communication is restored to ensure data integrity.

[0115] All structured data frames are accompanied by CRC checksums and timestamps; a data synchronization mechanism is supported, and the master station can request retransmission or data verification based on the timestamp to ensure consistency between the upper and lower level machines.

[0116] In this embodiment of the invention, to achieve data transmission between the intelligent monitoring terminal of the pole-mounted circuit breaker and the distribution automation master station, the system integrates multiple communication interfaces and transmission standards to adapt to different power grid deployment environments and bandwidth requirements. Specific communication methods include, but are not limited to, RS485, Ethernet, and wireless radio frequency.

[0117] The intelligent monitoring terminal has multiple pre-installed communication ports, including an RS485 serial port, an RJ45 Ethernet interface, and a LoRa / NB-IoT wireless module slot;

[0118] One of the communication methods can be dynamically enabled based on on-site configuration or remote parameters.

[0119] The system supports automatic protocol detection and link self-recovery mechanisms to improve communication stability.

[0120] It achieves strong anti-interference and flexible wiring based on differential signaling;

[0121] Master-slave communication is performed using the Modbus RTU protocol or the DL / T 645 protocol.

[0122] Each device has a unique address code, and the main station polls and reads structured monitoring data frames.

[0123] Data frames consist of a frame header, function code, data field, and CRC checksum, and support an abnormal retransmission mechanism.

[0124] Suitable for fixed wiring environments in urban substations or power distribution rooms;

[0125] It supports the TCP / IP protocol stack, and structured monitoring data is sent via Socket using JSON or binary encoding.

[0126] It can run standard protocols such as IEC 60870-5-104 and IEC 61850 to realize real-time reception and status mapping of the master station;

[0127] A watchdog mechanism is provided to ensure uninterrupted communication links.

[0128] Supports low-power wide-area wireless communication protocols such as LoRa, NB-IoT, and 4G;

[0129] Widely deployed in remote mountainous areas, hilly regions, or distribution areas;

[0130] Data is encapsulated using MQTT, CoAP, or a custom protocol and then uploaded to the cloud middleware layer or edge master station.

[0131] The sampled structured data includes a device ID and timestamp, supports breakpoint resume and periodic heartbeat mechanism, and improves transmission integrity.

[0132] The main station is equipped with a unified protocol adaptation module to automatically parse the data frame structure of different communication methods; it performs decoding and legality judgment based on device address, frame number, and parameter fields; abnormal data frames can trigger feedback requests or be automatically ignored, and the validity of the frames is further evaluated in conjunction with the edge scoring system.

[0133] In this embodiment of the invention, the distribution automation master station system is equipped with a protocol parsing module for parsing and legality assessment of structured monitoring data frames uploaded from the pole-mounted circuit breaker intelligent monitoring terminal. This parsing process is based on industrial communication protocol standards, combined with equipment communication behavior characteristics and anomaly tolerance mechanisms, and specifically includes the following processing flow:

[0134] The main station receives data from different monitoring terminals via serial port, Ethernet, or wireless communication interface;

[0135] For the received data frames, preliminary protocol identification is performed according to the preset protocol type (such as Modbus RTU, DL / T 645, IEC 104, etc.);

[0136] Frame boundaries are defined using the start flag, function code, and frame length fields of the data frame.

[0137] Perform field-level decomposition on the data frame and extract:

[0138] Device address code;

[0139] Timestamp (if any);

[0140] Data load (such as measured values ​​of voltage, current, temperature, etc.);

[0141] Verification fields (such as CRC, BCC, etc.);

[0142] The frame is parsed and verified using the field structure and verification rules defined in the protocol.

[0143] The main station determines the validity of each data frame based on the following rules:

[0144] Field integrity check: Determine if there are any missing fields, misaligned fields, or abnormal lengths;

[0145] Data consistency verification: Check whether the data falls within a reasonable range (e.g., whether the voltage exceeds the equipment's rated value);

[0146] Checksum verification: Frame integrity is verified using algorithms such as CRC and cumulative summation.

[0147] Communication behavior rule matching: Compare the current frame with the device's historical behavior (such as sending cycle, data change trend), and mark it as abnormal if the deviation is too large.

[0148] For data frames that fail to validate, have abnormal fields, or exhibit sudden changes in behavior, the main site records them as "abnormal frames".

[0149] If the proportion of abnormal frames continues to rise or is concentrated on a certain device, the master station can send a return request or trigger a device self-test command.

[0150] If parsing errors occur frequently, the system will automatically switch to a low-frequency polling mode to reduce communication conflicts.

[0151] The results of the data frame legality judgment are compared with the results of the edge-side scoring model; if a frame is legal but the score is high-risk, it is considered a potential misjudgment, triggering adaptive optimization of the protocol parsing parameters.

[0152] In this embodiment of the invention, to achieve accurate evaluation of data frame legitimacy identification and adaptive adjustment of the master station protocol parser, the master station system further extracts physical layer communication quality characteristic parameters and protocol layer behavioral characteristic parameters after receiving communication data, for use in subsequent anomaly scoring and dynamic optimization analysis. Specifically, this includes the following:

[0153] Physical layer parameters primarily reflect the stability and anti-interference capability of the communication channel. Key features extracted include:

[0154] RSSI (Received Signal Strength Indicator) indicates the strength of the wireless signal received by the master station, measured in dBm. It is read in real time by the wireless module. Weak signals may increase the probability of frame distortion.

[0155] SNR (Signal-to-Noise Ratio) represents the ratio of signal power to noise power, reflecting the clarity of the communication link; a low SNR is usually accompanied by frame splicing abnormalities or misjudgments.

[0156] Bit error rate (BER) is the bit-level error rate estimated by comparing received data with expected data; it corresponds to common interference situations in serial ports or wireless communication links.

[0157] Retransmission frequency, which records the number of times the master station requests data retransmission or the connection failure rate, can serve as a historical trend indicator of changes in communication stability.

[0158] Specifically, after analyzing the failure rate of data retransmission requests from the main site within a fixed time period, abnormal values ​​for main site retransmission requests were obtained, including:

[0159] Within a fixed time period T, the master station monitors N pole-mounted circuit breaker devices, and records the retransmission failure rate (or a time series statistical value, such as the mean) for each device.

[0160] For each device i∈{1,2,...,N}, extract the retransmission failure feature x. i The expression is: in: This represents the number of retransmission failures of device i within the time period. x represents the total number of retransmission requests for device i within the time period. i ∈[0,1]: represents the retransmission failure rate of the device. Construct a similarity graph G=(V,E), where each node v i ∈V represents a device; similarity edge weights between nodes are defined using distance functions, such as the Gaussian kernel function: Where: w ij Let be the edge weight (similarity) between nodes i and j; σ is the scaling factor that controls the decay of similarity (an empirical value can be set to half the mean distance between all nodes). Define the degree matrix D∈R. N×N :D ii =∑ j w ij R is the set of real numbers; the Laplace matrix is: L = DW; where W is the edge weight matrix [w ij The expression for calculating the retransmission exception value requested by the main station is: in: WD represents the k nearest neighbors of node i; WD indicates that the master station requests to retransmit abnormal values, and k is the number of nearest neighbors, which can generally be set to 5 to 10, or can be adaptively selected according to the total number of nodes.

[0161] The larger the retransmission anomaly value requested by the master station, the higher the retransmission failure rate of the device per unit time is compared with that of similar devices. This indicates that its communication stability is poor and there may be frame loss, frame errors, or splicing anomalies. At this time, the risk of data frame legitimacy increases significantly, and the master station parser is more likely to misjudge malformed frames as normal frames or miss real anomalies.

[0162] Conversely, when the retransmission anomaly value is significantly smaller, meaning that the device's retransmission failure rate is much lower than the overall average, it may mean that the data frames uploaded by the device are abnormally stable, or even "too regular." In some cases, this may be a false normal phenomenon caused by communication module freezing, buffer blockage, or pseudo-response. It also carries the hidden risk of missing frame content authenticity, and therefore should also be monitored as a low-frequency, high-risk hidden danger.

[0163] Protocol layer parameters are primarily used to identify whether device communication behavior conforms to normal patterns. The key features extracted include:

[0164] Inter-frame time interval change rate: The magnitude of change in the time difference between consecutively received legal frames; normal equipment has a stable period, and a sudden change in the change rate indicates abnormal behavior.

[0165] Field stability: Compare the frequency of changes in specific protocol fields (such as frame header, function code, and data segment length) in consecutive frames; large field fluctuations may indicate a soft fault in the device or a frame splicing error signal.

[0166] Frame sequence integrity marker: If the protocol contains a frame number or serial number field, the frame structure can be determined by detecting behaviors such as sequence jumps, rollbacks, and repetitions.

[0167] Abnormal frame ratio: The proportion of frames that fail to be verified, conflict in fields, or fail to be parsed within a unit time window, serving as a dynamic indicator of protocol behavior.

[0168] After analyzing the changes in the ratio of abnormal frames within a unit time window, abnormal values ​​for the increase in the ratio of abnormal frames are generated, specifically:

[0169] The monitoring period is divided into n equal-length windows T1, T2, ..., T. n ; in each time window T n Within, record the abnormal frame rate 'a' of the device. n : in: For window T n The number of data frames that failed to resolve internally or were deemed illegal; For window T n The total number of data frames received.

[0170] Calculate the adjacent window increment Δa for the abnormal frame ratio i The expression is: Δa i =a i -a i-1 ,i=2,3,...,n; indicates that from window T i To T i-1 The increase in the ratio of abnormal frames between them. The mean μ of the historical increase sequence. ΔaWith standard deviation σ Δa The abnormal frame rate increase is calculated using the following expression: SD represents the abnormal increase in the abnormal frame rate, indicating the degree of deviation of the abnormal increase in this window from the historical level.

[0171] The larger the abnormal frame rate increase, the more significant the increase in the number of abnormal frames in the current time window compared to the previous window. This reflects a sharp deterioration in the device's communication quality or protocol compatibility. At this time, the data frames received by the master station are more likely to have field misalignment, verification failure, or frame structure damage. The risk of data frame legality increases significantly, which can easily lead to misinterpretation or missed fault reports.

[0172] Conversely, when the abnormal frame rate increase is significantly negative and below the statistical deviation range, it may mean that the device suddenly drops from a highly abnormal state to "too normal" in a short period of time. This may be due to abnormal masking phenomena such as device communication pause, cache retransmission, or the main station parser mistakenly identifying abnormal frames as normal frames. Therefore, the risk of data frame legitimacy should not be ignored at this stage.

[0173] In this embodiment of the invention, to quantitatively assess the legality risk of each received data frame, the system constructs an anomaly scoring model based on extracted feature parameters, and scores the risk of each data frame. This model employs a multinomial regression algorithm, using the main station's communication behavior features as the input feature vector, and outputs continuous anomaly scores (ranging from 0 to 1), with higher scores indicating higher frame anomaly risk. The model establishment and training process includes:

[0174] The abnormal values ​​of the main station's request for retransmission and the abnormal frame rate increase are converted into comprehensive feature vectors. These comprehensive feature vectors are used as input to the machine learning model. The machine learning model uses the prediction of abnormal score labels for each set of comprehensive feature vectors as the prediction objective and minimizes the sum of prediction errors for all abnormal score labels as the training objective. The machine learning model is trained until the sum of prediction errors converges, at which point the model training stops. The abnormal score is determined based on the model output. The machine learning model is a multinomial regression model.

[0175] In this embodiment of the invention, to improve the main station protocol parser's ability to identify malformed frames or frames with splicing errors, the system introduces an anomaly scoring mechanism on the main station side. This mechanism dynamically evaluates the accuracy of the current protocol parser's identification by comparing the parser's frame legitimacy judgment results with the anomaly score based on feature parameters. Specifically, this includes:

[0176] For each received data frame F i The main station system executes the following two judgment mechanisms in parallel:

[0177] Path A: Protocol parser judgment result: The data frame is structurally parsed according to the communication protocol rules (such as frame header, length, checksum, etc.);

[0178] Resulting in frame validity label P i ∈{0,1}, where 1 indicates that the frame is considered valid, and 0 indicates that it is an illegal or abnormal frame.

[0179] Path B: Anomaly Scoring Model Results:

[0180] Input the comprehensive feature vector of the time period to which the current frame belongs into the trained anomaly scoring model, and output the anomaly score value y. i ∈[0,1];

[0181] The frame risk level R is determined based on the set threshold θy. i ∈{0,1}, where 1 represents high risk and 0 represents normal.

[0182] For each frame, compare the protocol parser label P i With model output label R i Are they consistent?

[0183] If P i =R i Recorded as a consistent determination;

[0184] If P i ≠R i These are recorded as potential misjudgments and categorized as follows:

[0185] False positive type A (falsely judged as valid): P i =1,R i =1 indicates that the protocol parser mistakenly judged a high-risk frame as legitimate;

[0186] False positive type B (false positive for anomaly): P i =0,R i =0 indicates that the protocol parser misjudged a normal frame as abnormal.

[0187] Based on the frame determination results from the sliding time window, the accuracy of the protocol parser's frame legitimacy recognition is calculated using the following expression: Where: N is the total number of data frames received within the current time window.

[0188] If the accuracy rate remains below the set threshold (e.g., 85%), the protocol parser adaptive adjustment module will be triggered; the specific error pattern (e.g., frame check deviation, field drift, etc.) can be located by combining the misjudgment type statistics.

[0189] When the accuracy of the main station protocol parser is lower than the set threshold, the process of adaptively adjusting the verification logic based on the scoring model results.

[0190] Continuously monitor the accuracy of the protocol parser; when the condition is met: Accuracy < θ; where θ is a set threshold (e.g., 85%), the parser verification logic optimization process is initiated.

[0191] Extract the feature vector X of high-risk scoring frames from the scoring model; for each misjudged sample (e.g., misjudged as legal but with a high score), record its verification failure characteristics (CRC error type, field misalignment pattern, etc.); communication parameters (RSSI, frame length fluctuation, etc.); construct an abnormal frame feature set Ferr.

[0192] Feature sensitivity learning is performed using locally weighted logistic regression, with samples from Ferr used as the training set. Local fitting is performed on the error type and communication state of the protocol fields to model the misjudgment probability P(misjudgment|X) under a specific scenario, expressed as: in It is a vector inner product (or dot product), which is each feature multiplied by its corresponding weight and then summed to output a real value. v is the weight vector, which is the parameter learned during the training of the logistic regression model, representing the degree of influence of each feature on the "probability of misclassification".

[0193] Based on the feature segments with high misclassification probability output by the local weighted logistic regression model, select the adjustable parameters in the corresponding verification logic:

[0194] CRC fault tolerance range (adjustable check window, code bit tolerance);

[0195] Field drift tolerance (e.g., allowable start character position offset ±n bytes);

[0196] The optimal adjustment direction is obtained by minimizing the misjudgment prediction loss.

[0197] Dynamically distribute the adjusted parser configuration to the main station protocol stack;

[0198] Set up a rollback monitoring mechanism. If the false positive rate does not decrease after the adjustment or the frame validity accuracy deteriorates further, restore the original configuration.

[0199] In this embodiment of the invention, when the master station protocol parser triggers an adaptive parameter adjustment mechanism due to a decrease in recognition accuracy, the system feeds back the resulting adjustment to the master station protocol stack and scheduling control module. Based on real-time monitoring results of subsequent communication data, the system continuously optimizes and verifies the parsing effect, realizing a closed-loop rule optimization process, specifically including:

[0200] The adjusted protocol parser parameters (including CRC tolerance, field offset range, frame structure threshold, etc.) are uploaded to the main station system in the form of a structured configuration file. The main station maintains a protocol rule version control library, generates a version number and timestamp for each adjustment action, and records the corresponding triggering reason, adjustment parameters, and device identifier.

[0201] The main station continuously records the following indicators for specific devices or communication channels after the adjustment:

[0202] Accuracy of data frame validity determination;

[0203] False positive statistics (falsely judged as valid / falsely judged as abnormal);

[0204] Trends in retransmission failure rate and anomaly score;

[0205] The sampling period for indicators can be configured at the minute or hour level to ensure that the adjustment effect is statistically significant.

[0206] Every so often (e.g., after every 30 minutes of operation or after receiving 1000 frames), the system evaluates the parser optimization performance based on historical data, with metrics including:

[0207] Has the accuracy of legitimacy identification improved?

[0208] Has the consistency between the abnormal rating values ​​and the main site's judgment been improved?

[0209] After the adjustment, will new misjudgment types or communication errors occur?

[0210] If the optimization effect is found to be unsatisfactory, the system will automatically roll back to the previous parser rule version; or trigger a new round of feature extraction and scoring model updates to enter the relearning phase.

[0211] The main station can optionally enable the self-learning engine, using the adjusted communication data feedback as an expanded training data set; incrementally updating the original anomaly scoring model, multinomial regression, or locally weighted learning model; the updated model is used for the next round of adaptive judgment, thus forming a complete feedback optimization closed loop.

[0212] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0213] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0214] It should be understood that the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Additionally, the character " / " in this document generally indicates an "or" relationship between the preceding and following related objects, but it may also indicate an "and / or" relationship; please refer to the context for specific understanding. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0215] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for integrated primary and secondary intelligent monitoring of pole-mounted circuit breakers based on the Internet of Things, characterized in that: include: Integrate primary and secondary equipment on the pole-mounted circuit breaker; Various sensors are used to collect the operating parameters of pole-mounted circuit breakers; The raw data is initially processed by an embedded edge computing module to generate structured monitoring data. Select the communication method to transmit the processed structured monitoring data to the distribution automation master station; On the main station side, the received data frames are parsed according to the protocol, and the legality of the data frames is judged based on the standard communication protocol; Extracting physical layer communication quality characteristic parameters and protocol layer behavioral characteristic parameters from received communication data. Specifically, this includes analyzing the failure rate of master station data retransmission requests within a fixed time period to obtain master station retransmission anomalies. Specifically, within a fixed time period T, the master station monitors N pole-mounted circuit breaker devices, recording the retransmission failure rate for each device once. For each device i∈{1,2,...,N}, retransmission failure features are extracted. The expression is: ;in: This represents the number of retransmission failures of device i within the time period. This represents the total number of retransmission requests for device i within the specified time period. ∈[0,1]: represents the retransmission failure rate of the device; construct a similarity graph G=(V,E), where each node... ∈V represents a device; the similarity edge weights between nodes are defined using a distance function. ;in: Define the degree matrix for the edge weights between nodes i and j. : R is the set of real numbers; Laplace matrix for: Where W is the edge weight matrix. The expression for calculating the retransmission exception value requested by the main station is: ;in: WD represents the k nearest neighbors of node i; WD indicates that the master station requests a retransmission of the abnormal value, and k is the number of nearest neighbors; An anomaly scoring model is established based on the extracted feature parameters. A legality risk assessment is performed on each data frame, and an anomaly score is output. The judgment results of the main station protocol parser are compared with the abnormal score value to evaluate its accuracy in identifying malformed or spliced ​​data frames; When the recognition accuracy is lower than the set threshold, the verification logic in the protocol parser is adaptively adjusted based on the output of the scoring model. The adjusted protocol parsing rules are fed back to the main station system and continuously optimized in conjunction with subsequent communication data.

2. The IoT-based intelligent monitoring method for pole-mounted circuit breakers, characterized in that: The secondary equipment includes: a signal acquisition unit, a control execution unit, a communication and protocol processing unit, an embedded edge computing unit, and a power management module. All units are encapsulated in the control box or housing of the pole-mounted circuit breaker, forming an intelligent monitoring terminal.

3. The IoT-based intelligent monitoring method for pole-mounted circuit breakers, characterized in that: The signal acquisition unit includes a current sensor, a voltage sensor, a thermal temperature sensor, a humidity sensor, a displacement sensor, and an acceleration sensor, used to acquire the circuit breaker's operating status, environmental parameters, and mechanical characteristic parameters.

4. The IoT-based intelligent monitoring method for pole-mounted circuit breakers, characterized in that: After analyzing the changes in the ratio of abnormal frames within a unit time window, abnormal values ​​for the increase in the ratio of abnormal frames are generated, specifically: The monitoring period is divided into n equal-length windows. ; in each time window Within, record the abnormal frame rate of the device. : in: For window The number of data frames that failed to resolve internally or were deemed illegal; For window The total number of data frames received internally; Calculate the adjacent window increment for the abnormal frame ratio The expression is: ; indicates from window arrive The increase in the ratio of abnormal frames between different periods, and the average of the historical increase sequence. with standard deviation The abnormal frame rate increase is calculated using the following expression: ; The abnormal frame rate increases the abnormal value.

5. The IoT-based intelligent monitoring method for pole-mounted circuit breakers, characterized in that: An anomaly scoring model is established based on the extracted feature parameters. A legality risk assessment is performed on each data frame, and an anomaly score value is output. Specifically, the main station request retransmission anomaly value and the anomaly frame ratio increase anomaly value are converted into a comprehensive feature vector. The comprehensive feature vector is used as the input of the machine learning model. The machine learning model uses the prediction of anomaly score value labels for each set of comprehensive feature vectors as the prediction objective and minimizes the sum of prediction errors for all anomaly score value labels as the training objective. The machine learning model is trained until the sum of prediction errors converges, at which point the model training stops. The anomaly score value is determined based on the model output. The machine learning model is a multinomial regression model.

6. The IoT-based intelligent monitoring method for pole-mounted circuit breakers, characterized in that: The accuracy of the current protocol parser's identification is evaluated by comparing the parser's frame legitimacy judgment results with the anomaly score based on feature parameters. Specifically, this includes: For each received data frame The main station system executes the following two judgment mechanisms in parallel: Path A: The protocol parser determines the frame validity label based on the result. ∈{0,1}, where 1 indicates that the frame is judged to be valid, and 0 indicates that the frame is invalid or abnormal; Path B: Input the comprehensive feature vector of the time period to which the current frame belongs into the trained anomaly scoring model, and output the anomaly score value. ∈[0,1]; The risk level of a frame is determined based on a set threshold θy. ∈{0,1}, where 1 represents high risk and 0 represents normal; For each frame, compare the protocol parser tags. With model output labels Are they consistent? like = Recorded as a consistent determination; like ≠ These are recorded as potential misjudgments and categorized as follows: False positive type A: =1, =1 indicates that the protocol parser mistakenly judged a high-risk frame as legitimate; False positive type B: =0, =0 indicates that the protocol parser misjudged a normal frame as abnormal; Based on the frame determination results from the sliding time window, the frame legitimacy recognition accuracy of the protocol parser is calculated. The expression is: Where: N is the total number of data frames received within the current time window; if the accuracy rate continues to be lower than the set threshold, the protocol parser will be triggered to make adaptive adjustments.

7. The IoT-based intelligent monitoring method for pole-mounted circuit breakers, characterized in that: Extract the feature vector X of high-risk scoring frames from the scoring model; for each misjudged sample, record its verification failure features and communication parameters, and construct an abnormal frame feature set Ferr; Feature sensitivity learning is performed using locally weighted logistic regression, with samples from Ferr as the training set; local fitting is performed on protocol field error types and communication states to model the probability of misjudgment in specific scenarios. The expression is: ;in It is a vector inner product, which is calculated by multiplying each feature by its corresponding weight and then summing the results. The weight vector consists of parameters learned during the training of the logistic regression model. Based on the feature segments with high misclassification probability output by the local weighted logistic regression model, select the adjustable parameters in the corresponding verification logic; The optimal adjustment direction is obtained by minimizing the misjudgment prediction loss; The adjusted parser configuration is dynamically distributed to the main station protocol stack.

Images