Remote monitoring and alarm linkage system for power distribution cabinet

Abstract

The invention belongs to the technical field of power distribution cabinet monitoring, and discloses a power distribution cabinet remote monitoring and alarm linkage system, a sensing module can collect environment, electrical and component multi-dimensional physical parameters, and combines Kalman filtering and wavelet packet decomposition technologies to remove random noise and extract transient characteristics to generate multi-field associated data; the diagnosis and pre-judgment module can trace multi-stage fault inducements based on a fault causal chain map and a Bayesian reasoning algorithm, and establishes clear association between the fault inducements and phenomena through cross validation of a 1: 1 digital twinborn body and an LSTM attention mechanism algorithm, thereby improving the fault positioning precision; pre-judgment, linkage, verification and optimization of a whole-process closed loop are constructed, and a risk alarm module dynamically adjusts a fault level based on a three-dimensional system; the execution module is embedded into a multi-objective optimization algorithm, a balance is found among fault disposal effectiveness, load loss minimization and energy consumption optimization, an optimal disposal parameter combination is generated, and real-time calibration deviation is verified through twin comparison.

Description

Technical Field

[0001] This invention belongs to the field of power distribution cabinet monitoring technology, specifically a remote monitoring and alarm linkage system for power distribution cabinets. Background Technology

[0002] In the context of the coordinated development of power generation, grid, load, and storage in the new power system, the operating status of the distribution cabinet, as a key node in the flow of electrical energy, directly affects the stability and reliability of the power system. Although the existing remote monitoring and alarm systems for distribution cabinets have achieved basic data acquisition and alarm functions, the following technical problems exist at the level of technical mechanism and system architecture: Existing technologies only achieve independent acquisition and threshold judgment of electrical parameters, environmental parameters, and component status, without considering the coupling relationship between multiple physical field parameters. Due to the lack of analysis on the coupling mechanism of environmental, electrical, and component multi-field parameters, fault diagnosis relies only on single parameter threshold triggering, which cannot establish the causal relationship between fault causes and fault phenomena, resulting in missed detection of latent faults and inaccurate fault location.

[0003] The existing system's linkage logic is based on preset fixed rules, and the linkage strategy is not dynamically adjusted in conjunction with the fault prediction results; the linkage is limited to alarms and single action execution, and a closed loop of linkage effect verification and model reverse optimization has not been established; this leads to delayed linkage response and waste of resources, such as faults that do not require tripping being mistakenly triggered to trip, and the system has long relied on manual adjustment of strategies. Summary of the Invention

[0004] The purpose of this invention is to provide a remote monitoring and alarm linkage system for power distribution cabinets to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a remote monitoring and alarm linkage system for power distribution cabinets, the system comprising: The sensing module collects physical field parameters of the environment, electrical system, and components. It dynamically adjusts the collection priority of each parameter according to the real-time operating scenario, and realizes the adaptive parameter weight allocation to meet the different requirements of parameter collection under different working conditions. The processing method combines Kalman filtering and wavelet packet decomposition. Kalman filtering removes random noise, and wavelet packet decomposition extracts transient features of multi-field correlation parameters to generate fine-grained raw data. At the same time, a sensor type self-identification interface is configured to automatically identify various connected sensors. Finally, the processed multi-field correlation data is output.

[0006] Preferably, the diagnosis and prediction module incorporates multiple fault causal chain maps and combines them with Bayesian inference algorithms to achieve the tracing and location of multi-level fault causes; A digital twin of the power distribution cabinet corresponding to the physical equipment is established in a 1:1 ratio. Multi-field correlation data of the sensing module is synchronized in real time. By simulating the fault evolution process under different working conditions, the prediction results of the algorithm with LSTM combined with the attention mechanism are cross-validated to improve the reliability of long-term fault prediction. At the same time, an operation and maintenance experience feedback interface is configured to collect experience data transmitted by the operation and maintenance module. The diagnostic results are pushed to the risk alarm module and the execution module, and the prediction model parameters and the fault causal chain diagram are synchronously archived to the protection module.

[0007] Preferably, the risk alarm module establishes a three-dimensional risk classification system based on the fault diagnosis results and evolution trend of the diagnosis and prediction module, which includes the scope of influence, evolution speed, and causal correlation weight; for the same fault phenomenon, the risk level is dynamically adjusted according to the cause attribute. When caused by environmental causes that are prone to triggering chain faults, the risk level is increased accordingly; when caused by electrical causes with controllable scope of influence, the original level is maintained. It has a dual verification mechanism of cross-verification of fault mechanism and self-learning of false alarm. It verifies the authenticity of the abnormal parameters collected by the sensing module through fault mechanism, and automatically records false alarm cases to continuously optimize alarm triggering rules and timing, thereby reducing the probability of false alarm. The hierarchical alarm information is linked to the execution module and the operation and maintenance module in two directions: emergency fault information is pushed to the execution module to trigger rapid handling, and at the same time it is pushed to the operation and maintenance module to generate an emergency work order; general hidden danger information is synchronized to the operation and maintenance module to generate regular operation and maintenance tasks, and false alarm cases are fed back to the diagnosis and prediction module to assist in model optimization.

[0008] Preferably, the execution module embeds a multi-objective optimization algorithm based on the hierarchical alarm information of the risk alarm module and the fault diagnosis results of the diagnosis and prediction module. When making linkage decisions, it simultaneously considers the three major objectives of fault handling effectiveness, load loss minimization, and energy consumption optimization. It calculates the optimal handling parameter combination for different levels of faults to balance the fault resolution effect and power supply stability. Establish a twin comparison and verification mechanism. After the linkage is executed, compare the actual multi-field correlation parameters of the physical device with the expected optimization effect of the virtual twin simulation. When the deviation exceeds the reasonable range, the strategy adjustment operation is automatically triggered. A standardized interface is set up, and an integrated protocol adaptive conversion unit is used to automatically identify the protocol version differences of the connected systems. Cross-system data interaction can be completed without manual configuration, and the execution results and parameter deviation data are output.

[0009] Preferably, the operation and maintenance module integrates alarm information from the risk alarm module, execution results from the execution module, and real-time multi-field correlation data from the perception module to establish a full-process operation and maintenance system for work order management, remote collaboration, and data traceability; the work order system has a built-in fault handling knowledge base, and the generated operation and maintenance work orders are accompanied by historical successful cases of similar faults; Equipped with an AR remote guidance unit, on-site maintenance personnel can scan the power distribution cabinet using AR devices, and remote experts can directly mark the fault points on the AR screen and overlay animations of the handling steps in real time. Establish a full lifecycle data traceability system to automatically store multi-stage related data of the power distribution cabinet from installation and commissioning, daily operation and maintenance, fault handling to scrapping; and output experience data generated during operation and maintenance.

[0010] Preferably, the scenario adaptation and upgrade module automatically determines the application scenario based on the scenario association data of the perception module, the diagnostic model operation data of the diagnosis and prediction module, and the operation and maintenance feedback data of the operation and maintenance module, and matches the corresponding monitoring weights and diagnostic strategies through the multi-scenario association data initially collected by the perception module; when encountering unfamiliar scenarios, it automatically generates temporary adaptation strategies to increase the collection frequency of key parameters of the perception module. An incremental upgrade mode is adopted to upgrade the changed modules, including the fault mechanism knowledge base and the parameters of the new scenario prediction model. The upgrade process does not interrupt the normal monitoring and control functions of the equipment. At the same time, an extreme scenario simulation and verification function is built in. After the upgrade is completed, extreme working conditions are automatically simulated to verify key indicators, including parameter acquisition accuracy and diagnostic model stability, and output the upgraded adaptation strategy and model parameters.

[0011] Preferably, the protection module establishes a full-link security system with hierarchical protection, enhanced authentication, and early warning. It adopts a data-sensitive encryption mechanism, lightly encrypting the ordinary environmental temperature and humidity data collected by the sensing module using the SM4 algorithm, and using a combination of the SM4 and SM9 algorithms for the core data of the diagnosis and prediction module. A five-level authentication system is established, consisting of username, password, dynamic verification code, device fingerprint, and biometrics. When administrators perform operations such as modifying the fault mechanism knowledge base of the diagnostic and prediction module or adjusting the key operations of the execution module, fingerprint or facial biometrics are used, and the biometric information is bound to the hardware information of the operation and maintenance terminal. An abnormal behavior model is established based on the operation logs of each module to issue early warnings for risky operations.

[0012] The beneficial effects of this invention are as follows: 1. The sensing module of this invention can collect multi-dimensional physical parameters of environment, electrical system, and components. It combines Kalman filtering and wavelet packet decomposition techniques to remove random noise and extract transient features to generate multi-field correlation data. The diagnosis and prediction module relies on the fault causal chain graph and Bayesian inference algorithm to trace the causes of multi-level faults. It also establishes a clear correlation between fault causes and phenomena through cross-validation with 1:1 digital twins and LSTM attention mechanism algorithm, solves the problem of missed detection of hidden faults, improves fault location accuracy, and ensures the stable operation of the power system.

[0013] 2. This invention constructs a closed-loop process encompassing prediction, linkage, verification, and optimization. The risk alarm module dynamically adjusts the fault level based on a three-dimensional system of impact scope, evolution speed, and causal correlation weights. The execution module embeds a multi-objective optimization algorithm to find a balance between fault handling effectiveness, load loss minimization, and energy consumption optimization, generating the optimal handling parameter combination. It also verifies and calibrates deviations in real time through twin comparison. Simultaneously, the hierarchical alarm bidirectional linkage execution and maintenance module enables rapid handling of emergency faults and routine maintenance of general hidden dangers, avoiding unnecessary misoperations, reducing resource waste, and improving the speed of linkage response and the rationality of handling.

[0014] 3. The scenario adaptation and upgrade module of this invention can automatically identify application scenarios based on multi-module data, dynamically match monitoring weights and diagnostic strategies, generate temporary adaptation solutions for unfamiliar scenarios, and the incremental upgrade mode will not interrupt equipment operation. After the upgrade, it can also simulate extreme working conditions to verify indicators. The operation and maintenance module integrates full-process data, has a built-in fault handling knowledge base, and reduces the operation and maintenance threshold with AR remote guidance. Full life cycle data traceability facilitates experience accumulation. The protection module adopts hierarchical encryption, and core data is doubly encrypted with SM4 and SM9. It establishes a five-level identity authentication system, binds biometric information to operation and maintenance terminal hardware, and issues early warnings for abnormal operations. This not only improves the system's adaptability to different working conditions, but also ensures data and operation security. Attached Figure Description

[0015] Figure 1 This is a flowchart of the remote monitoring and alarm linkage system for the power distribution cabinet of the present invention. Detailed Implementation

[0016] 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, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] like Figure 1 As shown in the figure, this embodiment of the invention provides a remote monitoring and alarm linkage system for power distribution cabinets, the system comprising: The sensing module collects physical field parameters of the environment, electrical system, and components. It dynamically adjusts the collection priority of each parameter according to the real-time operating scenario, and realizes the parameter weight allocation that is adaptive to the scenario. In high humidity environment, it focuses on capturing the correlation characteristics between temperature, humidity and insulation resistance. In high altitude scenario, it strengthens the monitoring intensity of air pressure and partial discharge parameters, thereby adapting to the different parameter collection needs of different working conditions. A processing method combining Kalman filtering and wavelet packet decomposition is adopted. Kalman filtering removes random noise, and wavelet packet decomposition extracts transient features of multi-field correlation parameters such as partial discharge transient pulses to generate fine-grained raw data. At the same time, a sensor type self-identification interface is configured, which can automatically identify various access sensors such as infrared thermometers and ultrasonic thermometers without manual protocol configuration, simplifying the on-site deployment process and improving the flexibility of sensor adaptation. Finally, the processed multi-field correlation data is output.

[0018] The sensing module employs a two-stage processing approach: Kalman filtering and wavelet packet decomposition. Kalman filter parameter initialization: The observation noise covariance matrix R is based on the sensor's factory accuracy preset (e.g., R=0.5 for an infrared temperature sensor). 2 The prediction error covariance matrix P is initially set to the identity matrix and is updated once every 10 acquisition cycles based on the residuals. Wavelet packet decomposition configuration: Select sym8 wavelet basis function, set the number of decomposition layers to 4, adapt to the frequency range of transient signals of power distribution cabinet 50Hz-1kHz, use thresholding to denoise the high-frequency coefficients after decomposition, and take the threshold as 3 times the standard deviation. Transient feature screening: Three types of features were extracted from the partial discharge pulse: peak value, rise time, and energy value. Only feature data with a signal-to-noise ratio ≥15dB were retained for subsequent correlation analysis.

[0019] Kalman filter state update formula:

[0020]

[0021] In the formula: Indicates the first The optimal state estimation vector at time t is the pure parameter vector output after filtering, which includes the denoised result of multiple physical field parameters such as temperature and humidity, insulation resistance, air pressure, and partial discharge. Indicates the first The time is based on the previous time. The state prediction vector is derived from the optimal parameter estimate of the previous acquisition cycle and is used to predict the parameter trend at the current moment. Indicates the first The Kalman gain matrix at time t, dynamically balancing the reliability of predicted values ​​and the reliability of observed values, determines the weight allocation for noise removal; Indicates the first The sensor observation vector at any given time corresponds to the raw parameters (including random noise) directly collected by sensors such as infrared thermometry and ultrasonic thermometry. It represents the observation matrix, describes the mapping relationship between the multiphysics state vector and the sensor observation vector, and adapts to the signal conversion logic of different types of sensors in the sensing module; Indicates the first The prediction error covariance matrix at time step quantifies the uncertainty of the state prediction at the previous time step and is updated by the error statistics of historical data collected in the sensing module. The observation noise covariance matrix represents the inherent noise characteristics of the sensors in the sensing module, such as the measurement error of infrared thermometry and the random interference of ultrasonic sensors, which are preset through sensor calibration data. Represents the observation matrix The transpose of the matrix is ​​a necessary term for linear algebra operations, ensuring dimension matching in matrix multiplication and guaranteeing the effectiveness of filtering calculations.

[0022] (·) This represents the inverse operation of a matrix, used to solve the denominator term in the Kalman gain, thereby normalizing the weights of observation noise and prediction error.

[0023] The diagnosis and prediction module incorporates multiple typical "environment-electrical-component" fault causal chain diagrams, covering complete causal links such as "condensation → increased contact resistance → increased contact temperature → enhanced partial discharge"; combined with Bayesian inference algorithms, it enables the tracing and location of multi-level fault causes. A digital twin of the power distribution cabinet corresponding to the physical equipment is established in a 1:1 ratio. Multi-field correlation data of the sensing module is synchronized in real time. By simulating the fault evolution process under different working conditions, the prediction results of the algorithm with LSTM combined with the attention mechanism are cross-validated to improve the reliability of long-term fault prediction. At the same time, an operation and maintenance experience feedback interface is configured to collect experience data transmitted by the operation and maintenance module. The diagnostic results, including fault type, cause, and evolution trend, are pushed to the risk alarm module and execution module, and the prediction model parameters and fault causal chain graph are synchronously archived to the protection module.

[0024] Construction and validation of a 1:1 digital twin of the diagnostic module: Synchronized data scope: Includes 15 categories of data, including environmental parameters (temperature, humidity, air pressure), electrical parameters (current / voltage / partial discharge), and component parameters (contact temperature, insulation resistance, cabinet door status); Synchronization frequency: 1Hz under normal operating conditions, automatically increased to 10Hz when the sensing module detects abnormal parameters (such as temperature exceeding the threshold); Consistency verification: The root mean square error (RMSE) is used for evaluation. If the RMSE of all synchronization parameters is ≤5%, the twin is considered valid. If the RMSE is >5%, the sensor calibration process is triggered.

[0025] Details of the LSTM attention model for the diagnostic module: Network structure: 3-layer LSTM (64 hidden units) + 1-layer soft attention layer. The input is 12-dimensional multi-field correlation data (temperature, humidity, current, partial discharge, etc.) output by the perception module, and the output is the fault probability in the next 24 hours. Attention weight calculation: An additive attention mechanism is used, with weights...

[0026] in ， For LSTM Step into hidden state, For query vector, , , These are learnable parameters; Training data: 1000 sets of labeled fault data were used, including three categories of labels: normal, minor fault, and severe fault. Data preprocessing included normalization (to the [0,1] interval) and missing value imputation (linear interpolation). The training set and validation set were divided in an 8:2 ratio. Training was stopped when the validation set loss was ≤0.05.

[0027] Bayesian inference posterior probability formula:

[0028] In the formula: Indicates known fault phenomena When it occurs, the triggers The posterior probability of establishment corresponds to the core indicator for tracing and locating multi-level fault causes in the diagnosis and prediction module, such as known contact temperature rise faults ( When condensation occurs, the inducing factors ( The probability of ). Indicates known causes When it occurs, the fault symptoms are as follows: The likelihood probability of occurrence was determined by simulating the failure evolution under different inducing factors in the laboratory, such as simulating condensation in a 95% RH environment, measuring the probability of contact temperature rise failure, and repeating the test 30 times for each inducing factor, taking the average value as the probability. ,For example Contact temperature rise 0.82; Indicates the trigger The prior probability of occurrence is based on statistical data of distribution cabinet faults in GB / T 14598.30-2018, and corrected by combining 500 sets of operation and maintenance data collected by this system in the past two years. For example, the probability of condensation cause C1 is... =0.25, short-term current overload cause C2 =0.18; This represents the set of observed fault phenomena, such as enhanced partial discharge, increased contact resistance, and increased contact temperature; corresponding to the fault types diagnosed by the diagnostic and prediction modules. Indicates the first One potential cause of failure ( These are various triggers in the causal chain diagram of the diagnosis and prediction module, such as condensation, short-term current overload, abnormal air pressure, and insulation aging. Represents the set of all potential causes of failure. ， (Total number of causes), covering all types of environmental, electrical, and component-related causes built into the diagnostic and prediction module, ensuring no omissions in the reasoning; Indicates all The summation of the probabilities of each potential trigger is a normalization operation, which ensures that the sum of the posterior probabilities is 1, thus conforming to the rules of probability statistics.

[0029] The risk alarm module establishes a three-dimensional risk classification system based on the fault diagnosis results and evolution trends of the diagnosis and prediction module, which includes the scope of influence, evolution speed, and causal correlation weights. The risk level of the same fault phenomenon is dynamically adjusted according to the attribute of the cause. When it is caused by environmental causes such as condensation that are prone to triggering chain faults, the risk level is increased accordingly. When it is caused by electrical causes with a controllable scope of influence, such as short-term current overload, the original level is maintained, so as to achieve accurate adaptation of risk classification. It has a dual verification mechanism of cross-verification of fault mechanism and self-learning of false alarm. It verifies the authenticity of the abnormal parameters collected by the sensing module through fault mechanism, and automatically records false alarm cases such as single temperature rise without other related parameter abnormalities, and feeds them back to the diagnosis and prediction module to optimize alarm triggering rules and Bayesian inference model, continuously optimize alarm triggering rules and timing, and reduce the probability of false alarm. The hierarchical alarm information is linked to the execution module and the operation and maintenance module in two directions: emergency fault information is pushed to the execution module to trigger rapid handling, and at the same time it is pushed to the operation and maintenance module to generate an emergency work order; general hidden danger information is synchronized to the operation and maintenance module to generate regular operation and maintenance tasks, and false alarm cases are fed back to the diagnosis and prediction module to assist in model optimization.

[0030] The execution module focuses on the closed-loop management of the entire process of prediction, linkage, verification, and fine-tuning, based on the hierarchical alarm information of the risk alarm module and the fault diagnosis results of the diagnosis and prediction module. It embeds a multi-objective optimization algorithm, which simultaneously considers the three objectives of fault handling effectiveness, load loss minimization, and energy consumption optimization when making linkage decisions. It calculates the optimal handling parameter combination for different levels of faults to balance the fault resolution effect and power supply stability. Establish a twin comparison and verification mechanism. After the linkage is executed, the actual multi-field correlation parameters of the physical equipment are compared with the expected optimization effect of the virtual twin simulation. When the deviation exceeds the reasonable range, the dehumidification power adjustment, cooling fan speed adjustment and other strategy fine-tuning operations are automatically triggered. The system features standardized interfaces for OPCUA / IEC61850 / MQTT, integrates an adaptive protocol conversion unit, automatically identifies protocol version differences between connected systems, and enables cross-system data interaction without manual configuration, outputting execution results and parameter deviation data.

[0031] The multi-objective optimization algorithm for the execution module employs an improved genetic algorithm: Algorithm parameter configuration: Population size set to 50, crossover probability 0.8, mutation probability 0.05, maximum number of iterations 100; Weight dynamic adjustment rules: Emergency fault (affecting ≥3 circuits): ; General fault (affecting 1-2 circuits): 、 、 ; Potential hazards (affecting ≤ 1 circuit): ; Convergence criterion: When the objective function value of the population fluctuates by ≤2% for 5 consecutive generations, output the current optimal combination of treatment parameters.

[0032] Multi-objective optimization objective function:

[0033]

[0034]

[0035] In the formula: The comprehensive objective function representing multi-objective optimization needs to be minimized through an algorithm. It serves as the decision-making basis for the optimal combination of processing parameters in the execution module and is used to balance the three core objectives. The weight coefficients of each objective are represented by the following: =1, dynamically adjusted by the risk level of the risk alarm module, such as in case of an emergency failure. Increased capacity to accommodate different fault handling priorities; This indicates that the fault handling efficiency is high. The corresponding fault handling effectiveness indicators in the execution module, such as the condensation elimination rate after dehumidification operation and the temperature rise and fall rate after heat dissipation adjustment, are verified by the sensing module by real-time data collection. This indicates the amount of load loss caused by fault handling, corresponding to the load loss minimization target in the execution module, such as the load reduction due to temporary power outages during the handling process; This indicates the rated load capacity of the equipment, which is a design rated parameter of the power distribution cabinet. It serves as a normalization benchmark for load loss to ensure the comparability of different equipment. This represents the total energy consumption during the fault handling process, corresponding to the optimal energy consumption target in the execution module, such as the operating energy consumption of dehumidifiers and cooling fans. This represents the reference energy consumption, such as the upper limit of energy consumption for a standard treatment plan. It is preset based on equipment operation and maintenance specifications and is used for normalized calculation of energy consumption to avoid evaluation bias caused by differences in equipment type. These represent key operating parameters during the treatment process, such as dehumidification power and fan speed. They are core variables in the combination of treatment parameters in the execution module and directly affect the balance of the three objectives. This indicates the allowable range of values ​​for key operating parameters, which are limited by the equipment's hardware performance and operational safety specifications, such as the upper limit of fan speed and the safety threshold of dehumidification power. This indicates the output power of the processing operation, such as the power of the dehumidifier or the power of the cooling system. It is an adjustable execution parameter in the execution module and needs to be optimized within a safe range. This indicates the maximum allowable output power for the treatment operation, which is the rated output limit of the equipment hardware. It is used to avoid overload damage and ensure equipment safety during the treatment process.

[0036] The operation and maintenance module integrates alarm information from the risk alarm module, execution results from the execution module, and real-time multi-field correlation data from the perception module to establish a full-process operation and maintenance system for work order management, remote collaboration, and data traceability. The work order system has a built-in fault handling knowledge base, and the generated operation and maintenance work orders are accompanied by historical successful cases of similar faults, including mature handling procedures and key points of operation, providing accurate reference for on-site operation and maintenance. Equipped with an AR remote guidance unit, on-site maintenance personnel can scan the power distribution cabinet using AR devices, and remote experts can directly mark the fault points on the AR screen and overlay real-time animations of the handling steps, solving the problems of information asymmetry and unintuitive guidance in traditional remote communication. Establish a full lifecycle data traceability system to automatically store multi-stage related data of the power distribution cabinet from installation and commissioning, daily operation and maintenance, fault handling to scrapping, including the data collected by the sensing module and the execution data of the execution module; output the experience data generated during operation and maintenance, including misjudgment cases and optimization suggestions.

[0037] AR remote guidance implementation in the operations and maintenance module: Device compatibility: Supports Microsoft HoloLens2 and Magic Leap2, system version ≥ Android 11 / iOS15; Fault location: Image recognition and IMU inertial navigation are used. The AR device scans the feature points of the distribution cabinet panel (such as nameplates and terminals) and combines them with the abnormal location of the sensing module parameters (such as 'contact A phase temperature overheating'). The location error is ≤5mm. Handling animations: 50 typical fault handling animation templates are pre-stored (such as replacing fuses and cleaning condensation), supporting remote experts to annotate in real time on the AR screen (such as marking fault points with red boxes) and add text annotations.

[0038] The scenario adaptation and upgrade module, based on scenario-related data from the perception module, diagnostic model operation data from the diagnosis and prediction module, and operation and maintenance feedback data from the operation and maintenance module, automatically determines the application scenario through the multi-scenario related data initially collected by the perception module, and matches the corresponding monitoring weights and diagnostic strategies. It then uses a data bus to reverse-synchronize the monitoring weights to the perception module, dynamically adjusting the parameter collection priority; reverse-synchronizes the diagnostic strategies to the diagnosis and prediction module, optimizing the fault analysis logic; and automatically generates temporary adaptation strategies when encountering unfamiliar scenarios such as high salt spray, increasing the frequency of key parameter collection by the perception module, and pushing the scenario data to the R&D end to optimize the scenario adaptation template library. The incremental upgrade mode is adopted to upgrade the changed modules, including the fault mechanism knowledge base and the parameters of the new scenario prediction model. The upgrade process does not interrupt the normal monitoring and control functions of the equipment, and significantly shortens the upgrade time. At the same time, the built-in extreme scenario simulation and verification function automatically simulates extreme working conditions such as low temperature and high humidity, high dust and strong electromagnetic interference after the upgrade is completed, verifies key indicators such as parameter acquisition accuracy and diagnostic model stability, and outputs the upgraded adaptation strategy and model parameters.

[0039] Extreme condition verification of the scene adaptation module: Extreme operating parameters: Low temperature: -30℃, for 2 hours; High temperature: 60℃, lasting for 4 hours; Overcurrent: 2 times the rated current for 10 minutes; High humidity + high dust: RH 95% + PM2.5 500μg / m³ 3 It lasted for 8 hours; Verification pass criteria: Parameter acquisition accuracy: error ≤2% (e.g., the deviation between infrared temperature measurement value and standard thermometer value ≤1℃); Diagnostic model: Fault identification accuracy ≥90%, false alarm rate ≤5%; Execution module: The actual effect of the treatment parameters deviates from the twin simulation by ≤10%.

[0040] Unfamiliar scene handling in the scene adaptation module: Preset templates include high humidity (RH≥85%), high altitude (altitude≥3000m), and high dust (PM2.5≥150μg / m³). 3 The template includes four basic scenarios: strong electromagnetic interference (electric field strength ≥1000V / m), with each scenario containing 10 sets of typical parameter thresholds. Matching degree calculation: Cosine similarity is used to calculate the similarity between the current scene parameters and the template. Formula A value of 0.6 indicates an unfamiliar scene; Temporary strategy: In unfamiliar scenarios, the acquisition frequency of key parameters (partial discharge quantity, insulation resistance) is increased from 1Hz to 5Hz, and the fault threshold of the diagnostic model is relaxed by 10%.

[0041] The protection module establishes a full-link security system with hierarchical protection, enhanced authentication, and early warning. It adopts a data-sensitive encryption mechanism, lightly encrypting the ordinary environmental temperature and humidity data collected by the sensing module using the SM4 algorithm, and using a combination of the SM4 and SM9 algorithms to double-encrypt the core data of the diagnosis and prediction module, such as the partial discharge quantity, prediction model parameters, and fault causal chain diagram. A five-level authentication system is established, consisting of username, password, dynamic verification code, device fingerprint, and biometrics. When administrators perform operations such as modifying the fault mechanism knowledge base of the diagnostic and prediction module or adjusting key operations of the execution module, they must perform fingerprint or facial biometrics. The biometric information is bound to the hardware information of the operation and maintenance terminal to enhance the security of identity authentication. An abnormal behavior model is established based on the operation logs of each module to issue early warnings for risky operations such as multiple attempts to access the fault knowledge base and batch downloading of related data during non-working hours. The abnormal response is controlled within minutes, realizing the transformation from passive recording to proactive warning.

[0042] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0043] 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 power distribution cabinet remote monitoring and alarm linkage system, characterized in that: The system comprises: a perception module: collecting multi-dimensional physical parameters and dynamically adjusting the collection priority, filtering noise and extracting transient signal features; a diagnosis and prediction module: tracing and positioning fault causes, establishing a digital twin to simulate the fault process, predicting long-term trends, and outputting diagnosis results; a risk warning module: establishing a risk assessment system and dividing risks into grades; according to the risk grade, different information is pushed; an execution module: generating disposal parameter combinations for different grade faults, comparing and calibrating deviations through digital twins, and automatically adjusting strategies; an operation and maintenance module: built-in fault disposal knowledge base, supporting AR remote guidance, and outputting experience data generated by operation and maintenance; a scene adaptation and upgrade module: identifying application environments and dynamically adjusting monitoring weights and diagnosis strategies, using incremental upgrade mode to reduce downtime, simulating extreme working conditions after upgrading and verifying indicators; a protection module: hierarchical encryption of data, establishment of a five-level identity authentication system, binding of hardware and biological information, and early warning of abnormal operation behavior.

2. The power distribution cabinet remote monitoring and alarm linkage system according to claim 1, characterized in that: The perception module collects environmental, electrical, and component physical field parameters, dynamically adjusts the collection priority and weight distribution according to the real-time running scene, and adapts to different working condition requirements; Using a combination of Kalman filtering and wavelet packet decomposition, random noise is removed and the transient features of multi-field related parameters are extracted to generate fine-grained raw data; configure a sensor type self-identification interface to automatically identify the connected sensors and output processed multi-field related data.

3. The power distribution cabinet remote monitoring and alarm linkage system according to claim 2, characterized in that: The diagnosis and prediction module has multiple fault cause and effect chain maps built-in, and uses Bayesian inference algorithm to realize multi-level fault cause tracing and positioning; A digital twin corresponding to the physical device 1:1 is established, real-time synchronization of multi-field related data is realized, the fault evolution process is simulated, and the prediction results are cross-verified with the LSTM combined attention mechanism algorithm to improve the prediction reliability; configure an operation and maintenance experience feedback interface to collect experience data transmitted by the operation and maintenance module; The diagnosis results are pushed to the risk warning module and the execution module, and the prediction model parameters and fault cause and effect chain maps are synchronized to the protection module.

4. The power distribution cabinet remote monitoring and alarm linkage system of claim 3, wherein: The risk warning module establishes a three-dimensional risk grading system based on fault diagnosis results and evolution trends, including influence range, evolution speed, and causal correlation weight; the same fault phenomenon dynamically adjusts the risk level according to the cause attribute, and the risk level is increased when it is caused by an environmental cause that can easily trigger a chain fault; the original level is maintained when it is caused by an electrical cause with controllable influence range; It has a double-checking mechanism of fault mechanism cross-checking and false alarm self-learning to reduce the false alarm probability; The graded alarm information is bidirectionally linked to the execution module and the operation and maintenance module: emergency faults trigger rapid disposal and generate emergency work orders; False alarm cases are fed back to the diagnosis and prediction module to assist model optimization.

5. The power distribution cabinet remote monitoring and alarm linkage system according to claim 4, characterized in that: The execution module is based on graded alarm information and fault diagnosis results, embedded with a multi-objective optimization algorithm, and simultaneously considers fault disposal effectiveness, load loss minimization, and energy optimization when making decisions, and calculates the optimal disposal parameter combination for different grade faults; A twin comparison verification mechanism is established to compare the actual parameters of the physical device with the expected optimization effect of the virtual twin, and when the deviation exceeds the reasonable range, the strategy adjustment operation is automatically triggered; A standardized interface and protocol adaptive conversion unit are set up to automatically identify the protocol version difference of the connected system, complete cross-system data interaction, and output the execution result and parameter deviation data.

6. The power distribution cabinet remote monitoring and alarm linkage system of claim 5, wherein: The operation and maintenance module integrates alarm information, execution results, and real-time multi-field correlation data to establish a full-process operation and maintenance system for work order management, remote collaboration, and data traceability. The work order system has a built-in fault handling knowledge base, and the operation and maintenance work order is accompanied by historical successful cases of similar faults. An AR remote guidance unit is installed, and after the on-site maintenance personnel scan through the AR device, the remote expert can mark the fault point and superimpose the disposal step animation in the AR picture. A full-life-cycle data traceability system is established to store multi-field correlation data of the power distribution cabinet at all stages and output experience data generated during the operation and maintenance process.

7. The power distribution cabinet remote monitoring and alarm linkage system of claim 6, wherein: The scene adaptation and upgrade module automatically judges the application scene based on the scene correlation data of the perception module, the diagnostic model running data of the diagnosis and prediction module, and the operation and maintenance feedback data of the operation and maintenance module, and matches the corresponding monitoring weight and diagnosis strategy. In unfamiliar scenes, a temporary adaptation strategy is automatically generated to improve the key parameter acquisition frequency. An incremental upgrade mode is used to upgrade the change module, and the upgrade process does not interrupt the normal monitoring and control functions of the device. At the same time, an extreme scene simulation and verification function is built in, which automatically simulates extreme working conditions after upgrading to verify key indicators including parameter acquisition accuracy and diagnostic model stability, and outputs the upgraded adaptation strategy and model parameters.

8. The power distribution cabinet remote monitoring and alarm linkage system of claim 7, wherein: The protection module establishes a full-link security system with hierarchical protection, enhanced authentication, and early warning. It uses a data sensitivity level encryption mechanism, with SM4 algorithm for light encryption of ordinary environment temperature and humidity data, and SM4 algorithm combined with SM9 algorithm for double encryption of core data. A five-level identity authentication system is established for username, password, dynamic verification code, device fingerprint, and biometric identification. Biometric identification is required for critical operations, and the biological information is bound to the hardware information of the operation and maintenance terminal. An abnormal behavior model is established based on the operation logs of each module to issue an early warning for risky operations.

Images