Intelligent comprehensive low-voltage power distribution cabinet with fault self-diagnosis

Abstract

The application provides an intelligent comprehensive low-voltage power distribution cabinet with fault self-diagnosis, and relates to the technical field of low-voltage power distribution cabinets. The intelligent comprehensive low-voltage power distribution cabinet with fault self-diagnosis comprises a cabinet body, a control combination table is arranged in the cabinet body, main loop power distribution elements, a five-dimensional multi-source heterogeneous sensing unit, an edge intelligent self-diagnosis core unit, a hierarchical protection execution and linkage unit, a communication and cloud collaborative unit and a man-machine interaction and emergency operation unit are respectively arranged in the control combination table, heat dissipation shells are arranged on the left and right sides of the cabinet body, and fans for heat dissipation and dehumidification are arranged in the heat dissipation shells. Through five-dimensional multi-source heterogeneous sensing, multi-source data fusion calculation, double-mode hierarchical diagnosis and hierarchical protection, health degree evaluation and component-level tracing, panoramic sensing, global judgment, differentiated treatment and precise operation and maintenance are realized, and the operation reliability and intelligent level of the power distribution cabinet are improved.

Description

Technical Field

[0001] This invention relates to the field of low-voltage distribution cabinet technology, specifically to an intelligent integrated low-voltage distribution cabinet with self-diagnosis of faults. Background Technology

[0002] Low-voltage switchgear is a key piece of equipment in power systems for power distribution, control, and protection, widely used in various power distribution scenarios including industrial, commercial, and residential applications. As the core hub connecting the power system and end users, the operational reliability of low-voltage switchgear directly affects the safety and economy of the entire power distribution network. With the continuous improvement of industrial automation and the in-depth advancement of the construction of new power systems, low-voltage switchgear is undergoing a profound transformation from traditional mechanical types to intelligent and digital ones. Intelligent switchgear integrates various sensing elements, communication modules, and data processing units within the cabinet to achieve real-time perception of equipment operating status, anomaly warning, and remote control.

[0003] However, existing low-voltage distribution cabinets still have many technical shortcomings in fault diagnosis and intelligent operation and maintenance, mainly in the following aspects.

[0004] First, the status monitoring methods are limited and the sensing dimensions are incomplete. Current monitoring of distribution cabinets primarily relies on manual inspections, with maintenance personnel judging equipment health status by "listening to sounds, feeling temperatures, and checking meters." This experience-dependent inspection model suffers from drawbacks such as long inspection cycles (typically 1 to 7 days), strong subjectivity, and non-quantifiable data, resulting in many hidden faults going undetected. While some distribution cabinets have been equipped with sensors for online monitoring, these are often limited to collecting single physical quantities, such as monitoring only electrical parameters like current and voltage, or only environmental parameters like temperature and humidity, lacking comprehensive sensing capabilities across multiple dimensions including electrical, thermal, mechanical, and insulation aspects. Due to this incomplete monitoring, many early signs of faults cannot be effectively captured.

[0005] Secondly, there is a lack of effective fusion and correlation analysis of multi-source heterogeneous data. The operating status of a distribution cabinet involves multiple physical dimensions, including contact temperature, ambient temperature and humidity, mechanical operating characteristics, and insulation status. In existing technologies, this data is often collected and stored independently by different instruments, creating severe data silos. Insufficient standardization of communication interfaces between different manufacturers' equipment makes multi-source data fusion difficult, hindering the comprehensive perception of the power distribution system. More critically, the traditional model lacks the ability to correlate data across dimensions, failing to link the isolated pieces of information such as "slow rise in circuit breaker contact temperature" and "slight extension of mechanism opening time" into an early diagnostic conclusion of "contact spring fatigue." This underutilization of data prevents existing systems from comprehensively assessing the true health status of equipment from a holistic perspective.

[0006] Third, fault diagnosis primarily relies on threshold triggering, lacking tiered diagnosis and predictive capabilities. Currently, most distribution cabinet fault judgments are still based on threshold comparisons of a single parameter; that is, an alarm or protection action is triggered when a parameter exceeds a preset upper or lower limit. This "exceeding the limit triggers an alarm" model is too crude. On the one hand, it cannot distinguish between the different handling requirements of hard, sudden faults and gradual degradation faults; on the other hand, it lacks sensitivity to early signs of faults. While some existing intelligent diagnostic systems can achieve preliminary anomaly detection, they are mostly in a "sensing-alarm" mode, lacking a deep understanding of the causes of faults and precise location capabilities. For gradual faults such as insulation aging and mechanical wear, existing technologies generally lack effective predictive methods.

[0007] Fourth, there is a lack of systematic health assessment and component-level fault tracing mechanisms. Existing technologies mostly focus on detecting and alarming single fault points, lacking a quantitative assessment of the overall health status of the distribution cabinet. When equipment malfunctions, maintenance personnel often need to check each component individually to pinpoint the fault location, which is inefficient and reliant on experience. Especially in scenarios such as insulation faults, existing monitoring methods are mostly coarse-grained monitoring of the entire distribution cabinet. After an alarm occurs, all equipment must be checked one by one while the power is off, making it impossible to achieve real-time, accurate location of individual equipment or components. This lack of systematic health assessment and accurate fault tracing capabilities severely restricts the improvement of distribution cabinet maintenance efficiency. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides an intelligent integrated low-voltage distribution cabinet with self-diagnosis of faults, which solves the technical shortcomings of existing low-voltage distribution cabinets, such as single monitoring dimensions, data silos, crude diagnosis, lack of quantitative health assessment and precise traceability at the component level.

[0009] To achieve the above objectives, the present invention provides the following technical solution: an intelligent integrated low-voltage distribution cabinet with fault self-diagnosis, comprising a cabinet body, wherein a control panel is provided inside the cabinet body, and the control panel integrates main circuit power distribution components, a five-dimensional multi-source heterogeneous sensing unit, an edge intelligent self-diagnosis core unit, a hierarchical protection execution and linkage unit, a communication and cloud collaboration unit, and a human-machine interaction and emergency operation unit. Heat dissipation shells are provided on both the left and right sides of the cabinet body, and fans for heat dissipation and dehumidification are provided inside the heat dissipation shells. The signal output terminal of the five-dimensional multi-source heterogeneous sensing unit is connected to the signal input terminal of the edge intelligent self-diagnosis core unit. It is used to synchronously collect multi-source heterogeneous operating status data of the power distribution cabinet in the electrical dimension, contact characteristic dimension, thermal state dimension, mechanical state dimension and insulation environment dimension. After performing time alignment and dimensional normalization preprocessing on the multi-source heterogeneous operating status data, it is transmitted to the edge intelligent self-diagnosis core unit. The first control output terminal of the edge intelligent self-diagnosis core unit is connected to the control input terminal of the hierarchical protection execution and linkage unit. It is used to perform hierarchical diagnosis based on preprocessed multi-source heterogeneous data, including millisecond-level hard fault discrimination and progressive fault prediction. The hierarchical diagnosis results are then integrated to generate a comprehensive health score and corresponding hierarchical protection instructions, which are then output to the hierarchical protection execution and linkage unit. The overall health score H is calculated as follows: First, construct a five-dimensional state deviation vector, expressed as: ; in, Represents the deviation in electrical dimensions. Represents the deviation in contact characteristic dimension. Represents the deviation of the thermal state dimension. Represents the deviation of the mechanical state dimension. Represents the deviation of the insulation environment dimension; The formulas for calculating the deviation of each dimension are as follows: ; in, Representing the Real-time detection feature values ​​of dimensions, Representing the The nominal reference value of the dimension, Representing the Fault threshold in dimensions; Secondly, the covariance matrix is ​​constructed based on the five-dimensional deviation vector, and its expression is: ; Solving the covariance matrix Maximum eigenvalue and the corresponding feature vectors ; Finally, the overall health score is calculated using the following formula: ; In the formula: This represents the preset scale compression factor. Represents the hyperbolic tangent function; The status feedback terminal of the hierarchical protection execution and linkage unit is connected to the status feedback input terminal of the edge intelligent self-diagnosis core unit, which is used to execute hierarchical protection actions and auxiliary linkage actions and transmit the execution status back in real time. The communication and cloud collaboration unit is bidirectionally connected to the edge intelligent self-diagnosis core unit, and the human-computer interaction and emergency operation unit is bidirectionally connected to the edge intelligent self-diagnosis core unit. These are used to realize data collaborative interaction between local and cloud, display of operating status, and priority input of emergency operation commands, respectively.

[0010] Preferably, the five-dimensional multi-source heterogeneous sensing unit includes an electrical parameter acquisition module, a contact characteristic detection module, a thermal state sensing module, a mechanical state sensing module, and an insulation and environmental sensing module. The electrical parameter acquisition module is used to acquire the three-phase voltage, three-phase current, active power, reactive power, harmonic content and zero-sequence current of each circuit at the first sampling frequency, and automatically perform fault recording when any electrical parameter exceeds the preset recording trigger threshold to generate fault recording data with time stamp. The contact characteristic detection module is used to inject a frequency of [frequency value missing] into the busbar joints and circuit breaker contacts using a low-frequency weak current injection method. Amplitude Low-frequency detection current, synchronously measuring contact voltage drop Calculate the contact resistance using the following formula. : ; To achieve micro-ohm-level online detection of contact resistance, and to transmit the contact resistance... The timestamp of the detection time is uploaded to the edge intelligent self-diagnosis core unit; The thermal state sensing module includes a passive wireless surface acoustic wave temperature sensor deployed on the busbar lap surface, circuit breaker contacts and cable terminals, as well as a cabinet temperature and humidity sensor and a condensation sensor, used to synchronously collect real-time temperature data of each heat point and the cabinet average temperature, relative humidity and condensation state signals. The mechanical state sensing module includes a circuit breaker travel displacement sensor, a closing coil current acquisition unit, and a micro vibration sensor. The circuit breaker travel displacement sensor is used to acquire the travel-time curve of the circuit breaker operating mechanism during the opening and closing process. The closing coil current acquisition unit is used to acquire the current-time waveform of the closing coil. The micro vibration sensor is used to acquire the vibration acceleration signal of the cabinet during the circuit breaker operation. The insulation and environmental sensing module includes a partial discharge ultrasonic sensor, a smoke concentration sensor and a water immersion sensor, which are used to collect the amplitude of the partial discharge ultrasonic signal, the smoke concentration value and the switching quantity of the water immersion status in the cabinet, respectively. The output of each module is connected to the edge intelligent self-diagnosis core unit. The edge intelligent self-diagnosis core unit performs unified timestamp labeling, outlier removal and missing value interpolation on the received data from each module to form a time-aligned five-dimensional perception data matrix.

[0011] Preferably, the edge intelligent self-diagnosis core unit uses an industrial-grade multi-core main control chip as its hardware carrier, and is equipped with a real-time operating system and a local data storage unit. Internally, it integrates a millisecond-level hard fault diagnosis engine, a progressive fault prediction and health assessment model, a component-level fault tracing and location algorithm, and a diagnostic system self-verification and channel self-test module. The millisecond-level hard fault diagnosis engine is used to monitor current surges in real time. and partial discharge ultrasonic amplitude ,when Exceeding the preset current surge threshold or Exceeding the preset partial discharge threshold When a hard fault occurs, it is determined within 5ms, and the fault type is identified according to the preset fault feature library, and the highest level protection trigger command is output synchronously. The progressive fault prediction and health assessment model is used to calculate the single-loop health score and the overall cabinet health score based on multi-dimensional historical data and current real-time data within a preset assessment period, without triggering a hard fault determination. ,when Below the preset warning threshold It generates progressive fault warning signals and outputs corresponding level protection instructions in real time; The aforementioned component-level fault tracing and localization algorithm is used to, after determining a hard fault or generating a progressive fault warning, convert the five-dimensional deviation vector... Input a pre-built Bayesian inference network, use deviation features in each dimension as evidence nodes and candidate faulty components as hypothesis nodes, calculate the posterior fault probability of each candidate faulty component, and select the component and location corresponding to the maximum posterior probability as the fault tracing result output. The diagnostic system's self-verification and channel self-test module is used to inject amplitude values ​​into each sensing channel of the five-dimensional multi-source heterogeneous sensing unit according to a preset self-test cycle. The test excitation signal is obtained, and the amplitude of the response signal of each sensing channel is monitored. ,when When the preset allowable deviation is exceeded, the corresponding sensing channel is determined to be abnormal, and a channel fault prompt message is generated. At the same time, the reliability of the diagnostic results is marked. The millisecond-level hard fault diagnosis engine, progressive fault prediction and health assessment model, component-level fault tracing and location algorithm, and diagnostic system self-verification and channel self-test module sequentially form a closed-loop processing link of data input - dual-mode hierarchical diagnosis - fault tracing output - diagnostic credibility verification.

[0012] Preferably, the graded protection execution and linkage unit includes a main circuit execution element and an auxiliary linkage mechanism; The main circuit execution element includes a circuit breaker tripping control module for each circuit, a load graded unloading control unit, and a circuit interlocking mechanism. The circuit breaker tripping control module is used to receive the tripping control command output by the edge intelligent self-diagnosis core unit and drive the tripping trip of the corresponding circuit breaker. The load graded unloading control unit is used to receive the load unloading command and cut off non-critical load circuits in sequence according to the preset load priority order. The circuit interlocking mechanism is used to apply mechanical interlocking to the operating mechanism of the fault circuit breaker after the tripping action is executed to prevent the fault circuit from being accidentally reclosed. The auxiliary linkage mechanism includes an intelligent heat dissipation and dehumidification control unit, a dual power supply switching linkage module, and a graded audible and visual alarm module. The intelligent heat dissipation and dehumidification control unit is used to receive heat dissipation and dehumidification control commands output by the edge intelligent self-diagnosis core unit, and dynamically adjust the start / stop status and speed of the fan according to the real-time temperature and condensation status inside the cabinet. The dual power supply switching linkage module is used to receive switching commands when the incoming power supply fails, and drive the dual power supply automatic switching unit to complete the main and backup power supply switching. The graded audible and visual alarm module is used to output differentiated audible and visual alarm signals with corresponding colors, flashing frequencies, and buzzer volumes according to the level of graded protection commands. The graded protection instructions output by the edge intelligent self-diagnosis core unit include a three-level instruction structure: the first-level instruction corresponds to the hard fault judgment result, triggering the fault circuit to trip, the circuit to block and the highest level of audible and visual alarm; the second-level instruction corresponds to the state where the health level is lower than the warning threshold, triggering the load to be unloaded in stages, the heat dissipation to be enhanced and the intermediate level of audible and visual alarm; the third-level instruction corresponds to the normal operation state, triggering the fan to adaptively adjust and the normal operation indicator light to be lit. After executing the graded protection command, the main circuit actuator and auxiliary linkage mechanism transmit the circuit breaker opening / closing status, load unloading result, interlocking mechanism status, fan speed, dual power supply circuit number, and alarm trigger status as execution status data back to the edge intelligent self-diagnosis core unit in real time.

[0013] Preferably, the main circuit power distribution components include an incoming circuit breaker, a busbar system, a multi-circuit drawer-type outgoing unit, a reactive power compensation unit, and a dual-power automatic switching unit. The multi-circuit drawer-type outgoing unit is equipped with a drawer insertion / removal counting and positioning detection structure to record the cumulative number of insertions / removals and detect the current positioning status. The operating mechanisms of the incoming circuit breaker and each outgoing circuit breaker have reserved sensor mounting positions for installing the travel displacement sensors of the mechanical state sensing module. The busbar system has reserved modular electromechanical interfaces for standardized and rapid access of the reactive power compensation unit and the dual-power automatic switching unit. The communication and cloud collaboration unit includes an Ethernet communication module, an RS485 communication module, a LoRa communication module, a 5G communication module, and a cloud collaboration module. The Ethernet and RS485 communication modules support the Modbus-RTU protocol, and the 5G communication module supports the IEC61850 protocol. The cloud collaboration module is used to upload five-dimensional sensing data, hierarchical diagnostic results, comprehensive health scores, and graded protection action records to the cloud platform through the communication module, and to receive multi-cabinet cluster load scheduling instructions and maintenance work order data issued by the cloud. The cloud collaboration module has a built-in breakpoint resume unit, which is used to write the data to be uploaded into the cache partition of the local data storage unit in timestamp order when network communication is interrupted, and resume the transmission from the breakpoint after communication is restored, ensuring the integrity of cloud data. The human-machine interaction and emergency operation unit includes an embedded industrial touch screen in the cabinet door and an emergency operation component. The industrial touch screen is used to graphically display in real time the topology diagram of the power distribution system, the three-phase electrical parameter values ​​of each circuit, the open / closed status of the circuit breaker, the health score of each circuit, the overall health score trend curve of the entire cabinet, and standardized fault handling operation guidance steps. The emergency operation component includes a physical emergency trip button, a physical reset button, a graded audible and visual alarm indicator light, and a buzzer. The contact signals of the physical emergency trip button and the physical reset button are directly connected to the interrupt input pin of the edge intelligent self-diagnosis core unit in a hard-wired manner. When pressed at any time, the corresponding operation is triggered with the highest interrupt priority.

[0014] A control method for an intelligent integrated low-voltage switchgear with self-diagnosis of faults includes the following steps: S1. Five-dimensional multi-source data synchronous acquisition and preprocessing: Through the five-dimensional multi-source heterogeneous sensing unit, each sensing module is triggered by a unified synchronous clock signal to synchronously acquire five types of operating status data of the power distribution cabinet in the electrical dimension, contact characteristic dimension, thermal state dimension, mechanical state dimension and insulation environment dimension. After outlier removal and missing value interpolation of the acquired data, the time sequence alignment and dimensional normalization preprocessing are completed according to the unified timestamp, a five-dimensional sensing data matrix is ​​constructed and transmitted to the edge intelligent self-diagnosis core unit. S2. Dual-mode hierarchical fault diagnosis and health assessment: After receiving the five-dimensional sensing data matrix, the edge intelligent self-diagnosis core unit performs dual-mode hierarchical diagnosis, including millisecond-level hard fault identification and progressive fault prediction. First, the millisecond-level hard fault diagnosis engine performs rapid hard fault identification based on current mutation rate and partial discharge ultrasonic amplitude. If a hard fault is detected, the highest-level protection command is immediately generated and the process jumps to the next step. If no hard fault is detected, the progressive fault prediction and health assessment model uses the five-dimensional deviation vector... Constructing the covariance matrix Solve for the covariance matrix Maximum eigenvalue and corresponding feature vectors ,according to Calculate the overall health score, where As a preset scale compression factor, when Below the preset warning threshold At the same time, the faulty component and its location are determined by the component-level fault tracing and location algorithm, and the corresponding level of graded protection instructions are generated. During the diagnosis process, the sensor channel self-test and the reliability verification of the diagnosis results are performed simultaneously. S3. Three-level protection and linkage execution: The edge intelligent self-diagnosis core unit outputs the graded protection command generated in step S2 to the graded protection execution and linkage unit. The graded protection execution and linkage unit performs differentiated actions according to the command level. When the first-level command is received, the fault circuit breaker is immediately tripped, the circuit is mechanically locked, and the highest level audible and visual alarm is triggered. When the second-level command is received, the non-critical load is unloaded in stages, the cooling fan speed is increased, and the intermediate level audible and visual warning is triggered. When the third-level normal operation command is in effect, the fan start / stop and speed are adaptively adjusted according to the real-time temperature and condensation status in the cabinet. When the incoming power supply fails, the dual power supply switching linkage is automatically triggered to ensure the power supply continuity of critical loads. After all actions are executed, the execution results are sent back to the edge intelligent self-diagnosis core unit. S4. Cloud Collaboration and Human-Machine Interaction Management: The communication and cloud collaboration unit uploads the sensing data collected in step S1, the diagnostic results and comprehensive health score generated in step S2, and the execution status data in step S3 to the cloud platform. It also receives multi-cabinet cluster load scheduling instructions and maintenance work orders issued by the cloud. The human-machine interaction and emergency operation unit displays the system topology, loop status, health score, and fault handling guidance in real time through the industrial touch screen. It also monitors the input status of the physical emergency trip button and reset button in real time. When button action is detected, the corresponding emergency operation instruction is injected into the edge intelligent self-diagnosis core unit with the highest interrupt priority, taking precedence over all currently executing automatic control processes.

[0015] Preferably, in step S1, the construction process of the five-dimensional sensing data matrix is ​​as follows: The electrical parameter acquisition module acquires the instantaneous values ​​of the three-phase voltage of each circuit at a first sampling frequency. Instantaneous values ​​of three-phase current The amplitude and phase of each harmonic are calculated in real time using Fast Fourier Transform, and the zero-sequence current is calculated simultaneously. The calculation formula is as follows: ; The contact characteristic detection module injects a frequency into the busbar joint and circuit breaker contacts according to the second sampling period. The low-frequency detection current is used to measure the contact voltage drop and calculate the contact resistance value sequence. ; The thermal state sensing module collects the temperature of each heat point using a passive wireless surface acoustic wave temperature sensor at a third sampling period. At the same time, the average temperature inside the cabinet was collected. relative humidity Condensation status switch quantity ; When the mechanical state sensing module detects a circuit breaker action event trigger signal, it synchronously acquires the travel-time curve at a fourth sampling frequency. Current-time waveform of opening and closing coils and cabinet vibration acceleration ; The insulation and environment sensing module collects partial discharge ultrasonic amplitude, smoke concentration and water immersion status switching quantity in the fifth sampling period. After performing outlier removal and linear interpolation on the raw data collected in each dimension, all data were aligned to a unified timestamp sequence using GPS / BDS timing signals as a unified clock source. Then, the dimensions of each dimension were normalized to construct a five-dimensional sensing data matrix with row vectors representing time sampling points and column vectors representing five-dimensional feature parameters. .

[0016] Preferably, in step S2, the specific judgment process of the millisecond-level hard fault diagnosis engine is as follows: Real-time calculation of three-phase current abrupt changes using a sliding window method ,in For one power frequency cycle duration, the ultrasonic amplitude of partial discharge is monitored simultaneously. The instantaneous value; Preset current mutation rate threshold Short-circuit current absolute value threshold Zero-sequence current threshold and partial ultrasound threshold ; When any phase current changes abruptly And the instantaneous value of the phase current When a phase-to-phase short-circuit fault is detected, a first-level trip command is immediately generated, and the fault type is marked as a short-circuit fault. When zero-sequence current And the amplitude of partial discharge ultrasound When a serious leakage fault accompanied by partial discharge is detected, a first-level trip command is immediately generated, and the fault type is marked as a combined leakage and partial discharge fault. When only satisfying Not satisfied When a general leakage fault is detected, a second-level warning command is generated. The total processing time for the above fault diagnosis process, from data input to instruction output, does not exceed 5ms.

[0017] Preferably, in step S2, the specific execution process of the component-level fault tracing and location algorithm is as follows: Pre-build a set of candidate faulty components, including circuit breaker contacts, busbar lap joints, cable terminals, insulation supports, and operating mechanisms. And for each candidate faulty component Establish corresponding fault feature fingerprint vectors ,in to These represent the typical deviation values ​​of the component when it fails in the electrical, contact characteristics, thermal, mechanical, and insulation environment dimensions, respectively. When a hard fault determination or a progressive fault warning triggers a source tracing requirement, the five-dimensional deviation vector at the current moment will be used. Fault feature fingerprint vectors of each candidate faulty component The weighted cosine similarity is calculated using the following formula: ; In the formula: Representing the The diagnostic weight coefficients for each dimension are determined by the information gain ratio algorithm from the historical fault case database. Select Candidate faulty components corresponding to the maximum value As a result of fault tracing, the corresponding component name, installation location number, and similarity score are also output.

[0018] Preferably, in step S3, the specific execution strategy for load grading and unloading is as follows: The load circuits connected to the distribution cabinet are pre-classified into first-level important loads, second-level important loads, and third-level non-important loads according to their importance level, and stored in the local data storage unit of the edge intelligent self-diagnosis core unit; When the graded protection command is at level two, the load graded unloading control unit obtains the real-time health score of each circuit. The order is as follows: first, disconnect the third-level non-critical loads; if the unloading requirements are still not met after disconnecting all third-level non-critical loads, then disconnect the second-level critical loads. Within each criticality level, the following steps are taken: The load is removed sequentially from low to high until the preset load unloading target power is met. ; The target power for load unloading is calculated using the following formula: ; In the formula: Represents the current total load power. This represents the upper limit of safe power supply determined based on power supply capacity and thermal stability limits. In a dual-power supply switching scenario, when the main incoming power supply voltage is detected to be lower than 85% of the rated voltage and the duration exceeds the switching delay... When the dual power supply switching linkage module first sends a switching request to the edge intelligent self-diagnosis core unit, the edge intelligent self-diagnosis core unit first executes the above-mentioned load grading and unloading strategy after receiving the request, reduces the total load power to below the preset ratio of the backup power supply's rated capacity, and then sends a permission to switch to the dual power supply switching linkage module to complete the main and backup power supply switching.

[0019] This invention provides an intelligent integrated low-voltage distribution cabinet with self-diagnosis of faults. It has the following beneficial effects: 1. This invention achieves a panoramic perception of the power distribution cabinet's full-dimensional operating status by setting up a five-dimensional multi-source heterogeneous sensing unit covering electrical, contact characteristics, thermal state, mechanical state, and insulation environment, and by combining it with a time-series alignment and dimensional normalization preprocessing mechanism. This effectively makes up for the shortcomings of traditional equipment monitoring with its single dimension and can accurately capture early characteristic signals of various progressive faults.

[0020] 2. This invention constructs a five-dimensional state deviation vector and covariance matrix fusion calculation model, and in conjunction with the unified data processing link of the edge intelligent self-diagnosis core unit, completes the deep correlation and fusion analysis of multi-source heterogeneous data, breaks the data silo problem of different sensing dimensions, and can comprehensively judge the true health status of the device from a global perspective.

[0021] 3. This invention establishes a dual-mode hierarchical diagnostic architecture that combines millisecond-level hard fault diagnosis with progressive fault prediction, and coordinates with a three-level graded protection execution mechanism to achieve differentiated fault identification and precise handling. It breaks through the limitations of the traditional threshold-triggered diagnosis and takes into account both rapid response to sudden faults and early warning of degradation faults.

[0022] 4. This invention uses a comprehensive health quantification calculation method based on the maximum eigenvalue and hyperbolic tangent function, combined with a component-level fault tracing and location algorithm, to achieve quantitative assessment of the overall cabinet health status and precise location of fault points. This effectively improves the problem of low efficiency in troubleshooting and locating traditional equipment, and significantly reduces maintenance costs and reliance on personnel experience. Attached Figure Description

[0023] Figure 1 This is a perspective view of the present invention; Figure 2 This is a cross-sectional view of the cabinet of the present invention; Figure 3This is a schematic diagram of the control console structure of the present invention; Figure 4 This is a modular framework diagram of the present invention; Figure 5 This is a flowchart of the steps of the present invention; Figure 6 This is a data comparison diagram of the present invention.

[0024] The components include: 1. Cabinet; 2. Control panel; 3. Main circuit power distribution components; 4. Five-dimensional multi-source heterogeneous sensing unit; 5. Edge intelligent self-diagnosis core unit; 6. Hierarchical protection execution and linkage unit; 7. Communication and cloud collaboration unit; 8. Human-machine interaction and emergency operation unit; 9. Heat dissipation shell; 10. Fan. Detailed Implementation

[0025] 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.

[0026] Example: like Figure 1-6 As shown in the figure, this embodiment of the invention provides an intelligent integrated low-voltage distribution cabinet with fault self-diagnosis, including a cabinet body 1. Inside the cabinet body 1 is a control console 2, which integrates main circuit power distribution components 3, a five-dimensional multi-source heterogeneous sensing unit 4, an edge intelligent self-diagnosis core unit 5, a hierarchical protection execution and linkage unit 6, a communication and cloud collaboration unit 7, and a human-machine interaction and emergency operation unit 8. Heat dissipation shells 9 are provided on both the left and right sides of the cabinet body 1, and fans 10 for heat dissipation and dehumidification are installed inside each heat dissipation shell 9. The signal output terminal of the five-dimensional multi-source heterogeneous sensing unit 4 is connected to the signal input terminal of the edge intelligent self-diagnosis core unit 5, for synchronously collecting data on the distribution cabinet's electrical, contact characteristic, thermal, mechanical, and insulation dimensions. Multi-source heterogeneous operational status data in the environmental dimension is processed and preprocessed by time alignment and dimensional normalization before being transmitted to the edge intelligent self-diagnosis core unit 5. The five-dimensional multi-source heterogeneous sensing unit 4 includes an electrical parameter acquisition module, a contact characteristic detection module, a thermal state sensing module, a mechanical state sensing module, and an insulation and environmental sensing module. The electrical parameter acquisition module is used to acquire the three-phase voltage, three-phase current, active power, reactive power, harmonic content, and zero-sequence current of each circuit at a first sampling frequency, and automatically executes fault recording when any electrical parameter exceeds the preset recording trigger threshold, generating time-stamped fault recording data. The contact characteristic detection module is used to inject a frequency of [missing information] into the busbar joints and circuit breaker contacts using a low-frequency weak current injection method. Amplitude Low-frequency detection current, synchronously measuring contact voltage drop Calculate the contact resistance using the following formula. : To achieve micro-ohm-level online detection of contact resistance, and to transmit the contact resistance... The timestamp of the detection time is uploaded to the edge intelligent self-diagnosis core unit 5; the thermal state sensing module includes passive wireless surface acoustic wave temperature sensors deployed on the busbar contact surface, circuit breaker contacts, and cable terminals, as well as cabinet temperature and humidity sensors and condensation sensors, used to synchronously collect real-time temperature data of each heat point and cabinet average temperature, relative humidity, and condensation status signals; the mechanical state sensing module includes a circuit breaker travel displacement sensor, a current acquisition unit for the opening and closing coils, and a micro vibration sensor. The circuit breaker travel displacement sensor is used to collect the travel-time curve of the circuit breaker operating mechanism during the opening and closing process, and the current acquisition unit for the opening and closing coils... The current acquisition unit is used to acquire the current-time waveform of the opening and closing coils, and the miniature vibration sensor is used to acquire the vibration acceleration signal of the cabinet during the operation of the circuit breaker; the insulation and environmental sensing module includes a partial discharge ultrasonic sensor, a smoke concentration sensor and a water immersion sensor, which are used to acquire the amplitude of the partial discharge ultrasonic signal, the smoke concentration value and the switching quantity of the water immersion status in the cabinet, respectively; the output of each module is connected to the edge intelligent self-diagnosis core unit 5. The edge intelligent self-diagnosis core unit 5 performs unified timestamp labeling, outlier removal and missing value interpolation on the received data from each module to form a time-aligned five-dimensional sensing data matrix; The first control output terminal of the edge intelligent self-diagnosis core unit 5 is connected to the control input terminal of the hierarchical protection execution and linkage unit 6. This connection is used to perform hierarchical diagnosis based on preprocessed multi-source heterogeneous data, including millisecond-level hard fault identification and progressive fault prediction. The edge intelligent self-diagnosis core unit 5 uses an industrial-grade multi-core main control chip as its hardware carrier, equipped with a real-time operating system and a local data storage unit. Internally, it integrates a millisecond-level hard fault diagnosis engine, a progressive fault prediction and health assessment model, a component-level fault tracing and location algorithm, and a diagnostic system self-verification and channel self-test module. The millisecond-level hard fault diagnosis engine is used to monitor current surges in real time. and partial discharge ultrasonic amplitude ,when Exceeding the preset current surge threshold or Exceeding the preset partial discharge threshold Within 5ms, a hard fault is detected, and the fault type is identified based on a preset fault feature library. Simultaneously, the highest-level protection trigger command is output. A progressive fault prediction and health assessment model is used to calculate the single-loop health score and the overall cabinet health score based on multi-dimensional historical data and current real-time data within a preset assessment cycle, provided that a hard fault detection has not been triggered. ,when Below the preset warning threshold It generates progressive fault warning signals and outputs corresponding graded protection instructions; a component-level fault tracing and location algorithm is used to locate the five-dimensional deviation vector after hard fault determination or progressive fault warning generation. The pre-constructed Bayesian inference network is input, using deviation features in each dimension as evidence nodes and candidate faulty components as hypothesis nodes. The posterior fault probability of each candidate faulty component is calculated, and the component and location corresponding to the maximum posterior probability are selected as the fault tracing result output. The diagnostic system self-verification and channel self-testing module is used to inject amplitude values ​​into each sensing channel of the five-dimensional multi-source heterogeneous sensing unit 4 according to a preset self-testing cycle. The test excitation signal is obtained, and the amplitude of the response signal of each sensing channel is monitored. ,when When the preset allowable deviation is exceeded, the corresponding sensing channel is determined to be abnormal and a channel fault prompt message is generated. At the same time, the credibility of the diagnostic results is marked. The millisecond-level hard fault diagnosis engine, the progressive fault prediction and health assessment model, the component-level fault tracing and location algorithm, and the diagnostic system self-verification and channel self-test module form a closed-loop processing link of data input-dual-mode hierarchical diagnosis-fault tracing output-diagnostic credibility verification in sequence. The stratified diagnostic results are integrated to generate a comprehensive health score and corresponding graded protection instructions, which are then output to the graded protection execution and linkage unit 6. The comprehensive health score H is calculated as follows: First, a five-dimensional state deviation vector is constructed, with the expression: ;in, Represents the deviation in electrical dimensions. Represents the deviation in contact characteristic dimension. Represents the deviation of the thermal state dimension. Represents the deviation of the mechanical state dimension. This represents the deviation in the insulation environment dimension; the formulas for calculating the deviation in each dimension are: ;in, Representing the Real-time detection feature values ​​of dimensions, Representing the The nominal reference value of the dimension, Representing the The fault threshold of the dimension; secondly, the covariance matrix is ​​constructed based on the five-dimensional deviation vector, and the expression is: Solve for the covariance matrix. Maximum eigenvalue and the corresponding feature vectors Finally, calculate the overall health score using the following formula: In the formula: This represents the preset scale compression factor. Representing the hyperbolic tangent function; the status feedback terminal of the graded protection execution and linkage unit 6 is connected to the status feedback input terminal of the edge intelligent self-diagnosis core unit 5, used to execute graded protection actions and auxiliary linkage actions and transmit the execution status back in real time. The graded protection execution and linkage unit 6 includes a main circuit execution element and an auxiliary linkage mechanism; the main circuit execution element includes a circuit breaker tripping control module for each circuit, a load graded unloading control unit and a circuit locking mechanism. The circuit breaker tripping control module for each circuit is used to receive the tripping control command output by the edge intelligent self-diagnosis core unit 5 and drive the tripping trip of the corresponding circuit breaker. The load graded unloading control unit is used to receive the load unloading command and cut off non-critical load circuits in sequence according to the preset load priority order. The circuit locking mechanism is used to apply mechanical locking to the operating mechanism of the faulty circuit breaker after the tripping action is executed to prevent the faulty circuit from being reclosed by mistake; the auxiliary linkage mechanism includes an intelligent heat dissipation and dehumidification control unit, a dual power supply switching linkage module and a graded sound and light alarm module. The intelligent heat dissipation and dehumidification control unit is used to receive the heat dissipation and dehumidification control output by the edge intelligent self-diagnosis core unit 5. The instructions dynamically adjust the start / stop status and speed of the fan 10 based on the real-time temperature and condensation status inside the cabinet. The dual power supply switching linkage module receives switching instructions when the incoming power supply fails, and drives the dual power supply automatic switching unit to complete the main and backup power supply switching. The graded audible and visual alarm module outputs differentiated audible and visual alarm signals with corresponding colors, flashing frequencies, and buzzer volumes according to the graded protection instructions. The graded protection instructions output by the edge intelligent self-diagnosis core unit 5 include a three-level instruction structure: the first-level instruction corresponds to the hard fault judgment result, triggering the fault circuit to open, the circuit to lock, and the highest level audible and visual alarm; the second-level instruction corresponds to the state where the health level is lower than the warning threshold, triggering the load to be unloaded in a graded manner, enhancing heat dissipation, and providing a medium-level audible and visual warning; the third-level instruction corresponds to the normal operation state, triggering the fan to self-adaptive adjustment and the normal operation indicator light to illuminate. After executing the graded protection instructions, the main circuit actuator and auxiliary linkage mechanism transmit the circuit breaker opening and closing status, load unloading result, lockout mechanism status, fan speed, dual power supply circuit number, and alarm trigger status as execution status data back to the edge intelligent self-diagnosis core unit 5 in real time. The communication and cloud collaboration unit 7 and the edge intelligent self-diagnosis core unit 5 are bidirectionally connected, as are the human-machine interaction and emergency operation unit 8 and the edge intelligent self-diagnosis core unit 5. These are used to achieve data collaboration and interaction between local and cloud environments, display of operational status, and priority input of emergency operation commands, respectively. The main circuit power distribution component 3 includes an incoming circuit breaker, a busbar system, a multi-circuit drawer-type outgoing unit, a reactive power compensation unit, and a dual-power automatic switching unit. The multi-circuit drawer-type outgoing unit is equipped with a drawer insertion / removal counting and positioning detection structure to record the cumulative number of insertions / removals and detect the current positioning status. Sensor mounting positions are reserved at the operating mechanisms of the incoming and outgoing circuit breakers for installing travel displacement sensors of the mechanical status sensing module. The busbar system has a reserved modular electromechanical interface for standardized and rapid access to the reactive power compensation unit and the dual-power automatic switching unit. The communication and cloud collaboration unit 7 includes an Ethernet communication module, an RS485 communication module, a LoRa communication module, a 5G communication module, and a cloud collaboration module. The Ethernet and RS485 communication modules support the Modbus-RTU protocol, and the 5G communication module supports the IEC61850 protocol. The cloud collaboration module is used to upload five-dimensional sensing data, hierarchical diagnostic results, comprehensive health scores, and graded protection action records to the cloud platform through the communication module, and to receive multi-cabinet cluster load scheduling instructions and maintenance work order data issued by the cloud. The cloud collaboration module has a built-in breakpoint resume unit, which is used to write the data to be uploaded into the cache partition of the local data storage unit in the order of timestamps when the network communication is interrupted, and resume the transmission from the breakpoint after the communication is restored, so as to ensure the integrity of the cloud data. The human-machine interaction and emergency operation unit 8 includes an embedded industrial touch screen in the cabinet door and an emergency operation component. The industrial touch screen is used to graphically display the power distribution system topology diagram, the three-phase electrical parameter values ​​of each circuit, the circuit breaker opening and closing status, the health score of each circuit, the overall health score trend curve of the cabinet, and the standardized fault handling operation guidance steps in real time. The emergency operation component includes a physical emergency trip button, a physical reset button, graded audible and visual alarm indicator lights and a buzzer. The contact signals of the physical emergency trip button and the physical reset button are directly connected to the interrupt input pin of the edge intelligent self-diagnosis core unit 5 in a hard-wired manner. When pressed at any time, the corresponding operation is triggered with the highest interrupt priority.

[0027] Example 1: Cabinet Structure and Modular Integration Cabinet 1 is constructed entirely of high-quality cold-rolled steel sheet, bent and welded, with the surface treated by phosphating and then electrostatically sprayed with epoxy resin powder. Inside cabinet 1 is a control console 2, which employs a modular installation architecture. Partitions divide the cabinet space into multiple independent installation compartments, effectively preventing the spread of faults between different functional units. Control console 2 integrates main circuit power distribution components 3, a five-dimensional multi-source heterogeneous sensing unit 4, an edge intelligent self-diagnosis core unit 5, a hierarchical protection execution and linkage unit 6, a communication and cloud collaboration unit 7, and a human-machine interaction and emergency operation unit 8. Both sides of cabinet 1 are equipped with heat dissipation shells 9, which have a louvered structure, combining ventilation, heat dissipation, and protection. Each heat dissipation shell 9 contains a fan 10 for heat dissipation and dehumidification. The fan 10 is an axial flow fan, and its start / stop and speed are adaptively controlled by the hierarchical protection execution and linkage unit 6 based on real-time temperature and humidity conditions inside the cabinet. The top of the cabinet 1 is also equipped with a dust filter, which works with the heat dissipation shell 9 to form a circulating air duct, and fireproof sealing mud is installed at the bottom cable inlet and outlet holes.

[0028] Example 2: Five-dimensional multi-source heterogeneous sensing and data fusion The five-dimensional multi-source heterogeneous sensing unit 4 is installed in the central mounting compartment of the control console 2, and its signal output terminal is connected to the signal input terminal of the edge intelligent self-diagnosis core unit 5. The five-dimensional multi-source heterogeneous sensing unit 4 includes an electrical parameter acquisition module, a contact characteristic detection module, a thermal state sensing module, a mechanical state sensing module, and an insulation and environmental sensing module.

[0029] The electrical parameter acquisition module converts primary-side signals into low-amplitude analog signals using miniature precision voltage and current transformers. After anti-aliasing filtering and analog-to-digital conversion, the signals are sent to a digital signal processor. The sampling rate for three-phase voltage and current in each circuit is 256 points per cycle. When any electrical parameter exceeds the preset recording trigger threshold, fault recording is automatically executed, and time-stamped recording data is generated. The contact characteristic detection module injects low-frequency detection current into the busbar joints and circuit breaker contacts through isolated coupling, simultaneously measuring the contact voltage drop. The edge intelligent self-diagnostic core unit 5 completes the contact resistance calculation and temperature compensation, with a detection accuracy of ±0.5μΩ. In the thermal state sensing module, a passive wireless surface acoustic wave temperature sensor is deployed at the busbar joint surface, circuit breaker contacts, and cable terminals. It communicates with the reader via radio frequency, requires no battery power, and has a measurement range of -40℃ to +200℃ with an accuracy of ±0.5℃. The cabinet temperature and humidity sensor and condensation sensor are located at the center height inside the cabinet. In the mechanical condition sensing module, the circuit breaker travel displacement sensor uses a linear variable differential transformer type displacement sensor, whose iron core is linked with the moving contact through a connecting rod to collect travel-time curves; the opening and closing coil current acquisition unit uses a Hall effect current sensor with a sampling rate of 200kS / s to fully capture the peaks and oscillation details of the coil current; the miniature vibration sensor uses a piezoelectric accelerometer installed on the inner wall of the cabinet side panel. In the insulation and environmental sensing module, two partial discharge ultrasonic sensors are arranged diagonally inside the cabinet to ensure full cabinet coverage; the smoke concentration sensor is a photoelectric scattering type smoke detector; and the water immersion sensor is located at the cable trench entrance at the bottom of the cabinet.

[0030] Each module's output is connected to the edge intelligent self-diagnosis core unit 5. The edge intelligent self-diagnosis core unit 5 performs 3σ criterion outlier removal and linear interpolation missing value filling on the received data from each module, and then performs timing alignment and dimension normalization using GPS / BDS timing signal as a unified clock source to form a five-dimensional sensing data matrix.

[0031] Example 3: Hardware Platform and Diagnostic Logic of Edge Intelligent Self-Diagnosis Core like Figure 4 As shown, the edge intelligent self-diagnostic core unit 5 uses an industrial-grade multi-core main control chip as its hardware carrier, and is equipped with a real-time operating system and a local data storage unit. The multi-core main control chip adopts a heterogeneous multi-core architecture of ARM Cortex-A and Cortex-M. The Cortex-A core runs the Linux operating system to handle complex algorithm calculations and network communication, while the Cortex-M core runs the FreeRTOS real-time operating system to handle data acquisition and fast interrupt response. The two cores exchange data through shared memory. The edge intelligent self-diagnostic core unit 5 integrates a millisecond-level hard fault diagnosis engine, a progressive fault prediction and health assessment model, a component-level fault tracing and location algorithm, and a diagnostic system self-verification and channel self-test module.

[0032] The millisecond-level hard fault diagnosis engine runs on the Cortex-M core, reading the analog-to-digital converter results via direct memory access, eliminating operating system scheduling latency, and achieving microsecond-level real-time monitoring of current surges ΔI / Δt and partial discharge ultrasonic amplitude Upd. When ΔI / Δt exceeds the preset current surge threshold Kset or Upd exceeds the preset partial discharge threshold Uth, a hard fault is determined to have occurred within 5ms. The fault type is identified based on a preset fault feature library, and the highest-level protection trigger command is output simultaneously.

[0033] The progressive fault prediction and health assessment model calculates a comprehensive health score H based on multi-dimensional historical data and current real-time data within a preset assessment period, provided that no hard fault determination is triggered. When H is lower than the preset warning threshold Halert, a progressive fault warning signal is generated and a corresponding level of graded protection instruction is output. The scale compression factor β ranges from 0.1 to 0.5, and is preferably 0.3 in this embodiment.

[0034] The component-level fault tracing and localization algorithm calculates the weighted cosine similarity between the current five-dimensional deviation vector D and the fault feature fingerprint vectors of each candidate faulty component after a fault occurs. The component and its location corresponding to the maximum similarity are selected as the fault tracing result output. The diagnostic weight coefficients ωk for each dimension are determined from the historical fault case database using the information gain ratio algorithm.

[0035] The diagnostic system's self-calibration and channel self-test module injects test excitation signals into each sensor channel according to a preset self-test cycle. When the amplitude of the response signal deviates from the theoretically calculated value by more than a preset allowable deviation, the corresponding sensor channel is determined to be abnormal, and a channel fault prompt message is generated. At the same time, the reliability of the diagnostic results is marked. The above four modules form a closed-loop processing link of data input - dual-mode hierarchical diagnosis - fault tracing output - diagnostic reliability verification.

[0036] Example 4: Hierarchical Protection Execution and Linkage Mechanism The graded protection execution and linkage unit 6 includes main circuit execution elements and auxiliary linkage mechanisms. The main circuit execution elements include circuit breaker tripping control modules for each circuit, load graded unloading control units, and circuit interlocking mechanisms. Each circuit breaker tripping control module is connected to the circuit breaker tripping trip control coil via opto-isolation, ensuring electrical isolation between the control signal and the primary circuit. The load graded unloading control unit reads the current power value and health score of each load circuit via a communication bus and sequentially cuts off non-critical load circuits according to preset load priorities. The circuit interlocking mechanism adopts an electromagnet-driven mechanical locking pin structure; after tripping, the locking pin pops out and engages in the linkage slot of the circuit breaker operating mechanism, preventing the faulty circuit from being accidentally reclosed.

[0037] The auxiliary linkage mechanism includes an intelligent heat dissipation and dehumidification control unit, a dual-power switching linkage module, and a graded audible and visual alarm module. The intelligent heat dissipation and dehumidification control unit dynamically adjusts the start / stop and speed of the fan 10 based on the real-time temperature and condensation status inside the cabinet: it runs at full speed when the temperature exceeds 45℃, reduces the speed to 50% of the rated speed when the temperature is below 35℃, and forces full-speed operation and activates the heating dehumidifier when condensation is detected. The dual-power switching linkage module executes a "disconnect first, then reconnect" safety switching logic in the event of an incoming power failure. The graded audible and visual alarm module outputs differentiated alarm signals according to the command level.

[0038] The edge intelligent self-diagnostic core unit 5 outputs a three-level protection command structure: the first-level command triggers fault circuit tripping, mechanical interlocking, and a red constant-on highest-level audible and visual alarm; the second-level command triggers graded load unloading, enhanced heat dissipation, and a yellow 1Hz flashing intermediate-level warning; the third-level command triggers fan adaptive adjustment and illuminates the green normal operation indicator. After execution by the main circuit actuators and auxiliary linkage mechanisms, the circuit breaker opening / closing status, load unloading results, fan speed, and other status data are transmitted back to the edge intelligent self-diagnostic core unit 5 in real time.

[0039] Example 5: Power Distribution Components, Communication Collaboration, and Human-Computer Interaction The main circuit power distribution component 3 includes an incoming circuit breaker, a busbar system, a multi-circuit drawer-type outgoing unit, a reactive power compensation unit, and a dual-power automatic switching unit. The incoming circuit breaker is a frame-type intelligent circuit breaker. The busbar system uses rectangular copper busbars with silver-plated overlapping surfaces. The multi-circuit drawer-type outgoing unit is equipped with a drawer insertion / removal counting and positioning detection structure. Sensor mounting positions are reserved at the operating mechanisms of the incoming circuit breaker and each outgoing circuit breaker. The busbar system has a reserved modular electromechanical interface for standardized and rapid access of the reactive power compensation and dual-power switching units.

[0040] The communication and cloud collaboration unit 7 includes an Ethernet communication module, an RS485 communication module, a LoRa communication module, a 5G communication module, and a cloud collaboration module. The Ethernet and RS485 communication modules support the Modbus-RTU protocol, and the 5G communication module supports the IEC61850 protocol. The cloud collaboration module uploads five-dimensional sensing data, diagnostic results, health scores, and protection action records to the cloud platform through the aforementioned communication modules. It also receives multi-cabinet cluster load scheduling instructions and maintenance work order data from the cloud. A built-in breakpoint resume unit ensures that data is not lost during network interruptions.

[0041] The human-machine interaction and emergency operation unit 8 includes an embedded industrial touch screen display and emergency operation components. The industrial touch screen displays the power distribution system topology diagram, three-phase electrical parameters of each circuit, circuit breaker opening and closing status, health scores of each circuit, overall cabinet health score trend curve, and standardized fault handling guidelines in a graphical and real-time manner. The emergency operation components include a physical emergency trip button, a physical reset button, tiered audible and visual alarm indicators, and a buzzer. The contact signals of the physical emergency trip button and the reset button are directly connected to the interrupt input pin of the edge intelligent self-diagnosis core unit 5 via hard wiring. Pressing the button at any time triggers the corresponding operation with the highest interrupt priority, unaffected by the software running status.

[0042] A control method for an intelligent integrated low-voltage switchgear with self-diagnosis of faults includes the following steps: S1. Synchronous Acquisition and Preprocessing of Five-Dimensional Multi-Source Data: The five-dimensional multi-source heterogeneous sensing unit 4 triggers each sensing module with a unified synchronous clock signal to synchronously acquire five types of operational status data of the distribution cabinet in the electrical, contact characteristic, thermal, mechanical, and insulation environment dimensions. After outlier removal and missing value interpolation of the acquired data, time alignment and dimensional normalization preprocessing are performed according to a unified timestamp. A five-dimensional sensing data matrix is ​​then constructed and transmitted to the edge intelligent self-diagnosis core unit 5. The specific process of constructing the five-dimensional sensing data matrix is ​​as follows: the electrical parameter acquisition module acquires the instantaneous values ​​of the three-phase voltage of each circuit at the first sampling frequency. Instantaneous values ​​of three-phase current The amplitude and phase of each harmonic are calculated in real time using Fast Fourier Transform, and the zero-sequence current is calculated simultaneously. The calculation formula is as follows: The contact characteristic detection module injects a frequency of [frequency value] into the busbar joints and circuit breaker contacts according to the second sampling cycle. The low-frequency detection current is used to measure the contact voltage drop and calculate the contact resistance value sequence. The thermal state sensing module collects the temperature of each heat point using a passive wireless surface acoustic wave temperature sensor at the third sampling period. At the same time, the average temperature inside the cabinet was collected. relative humidity Condensation status switch quantity When the mechanical condition sensing module detects a circuit breaker action event trigger signal, it synchronously acquires the travel-time curve at the fourth sampling frequency. Current-time waveform of opening and closing coils and cabinet vibration acceleration The insulation and environmental sensing module acquires partial discharge ultrasonic amplitude, smoke concentration, and water immersion status switching quantities during the fifth sampling period. After performing outlier removal and linear interpolation on the raw data for each dimension, all data are aligned to a unified timestamp sequence using GPS / BDS timing signals as a unified clock source. Then, the dimensions of each dimension are normalized to construct a five-dimensional sensing data matrix with row vectors representing time sampling points and column vectors representing five-dimensional feature parameters. ; S2. Dual-mode hierarchical fault diagnosis and health assessment: After receiving the five-dimensional sensing data matrix, the edge intelligent self-diagnosis core unit 5 performs dual-mode hierarchical diagnosis, which includes millisecond-level hard fault identification and progressive fault prediction. First, the millisecond-level hard fault diagnosis engine performs rapid hard fault identification based on current mutation rate and partial discharge ultrasonic amplitude. If a hard fault is detected, the highest-level protection command is immediately generated and the process jumps to the next step. If no hard fault is detected, the progressive fault prediction and health assessment model uses the five-dimensional deviation vector... Constructing the covariance matrix Solve for the covariance matrix Maximum eigenvalue and corresponding feature vectors ,according to Calculate the overall health score, where As a preset scale compression factor, when Below the preset warning threshold At that time, the faulty component and its location are determined through a component-level fault tracing and location algorithm, and a corresponding level of graded protection command is generated. During the diagnosis process, sensor channel self-test and diagnostic result reliability verification are performed simultaneously. The specific judgment process of the millisecond-level hard fault diagnosis engine is as follows: the sudden change in three-phase current is calculated in real time using a sliding window method. ,in For one power frequency cycle duration, the ultrasonic amplitude of partial discharge is monitored simultaneously. Instantaneous value; preset current mutation rate threshold Short-circuit current absolute value threshold Zero-sequence current threshold and partial ultrasound threshold When any phase current changes abruptly And the instantaneous value of the phase current When a phase-to-phase short-circuit fault is detected, a first-level tripping command is immediately generated, and the fault type is marked as a short-circuit fault; when the zero-sequence current... And the amplitude of partial discharge ultrasound When a severe leakage current accompanied by partial discharge fault is detected, a first-level tripping command is immediately generated, and the fault type is marked as a combined leakage current and partial discharge fault; when only the following conditions are met... Not satisfied When a general leakage fault is detected, a second-level warning command is generated; the total processing time of the above fault detection process from data input to command output does not exceed 5ms. The specific execution process of the component-level fault tracing and localization algorithm is as follows: a set of candidate faulty components, including circuit breaker contacts, busbar lap surfaces, cable terminals, insulation supports, and operating mechanisms, is pre-constructed. And for each candidate faulty component Establish corresponding fault feature fingerprint vectors ,in to These represent the typical deviation values ​​of the component when it fails in the electrical, contact characteristic, thermal, mechanical, and insulation environment dimensions, respectively. When a hard fault determination or progressive fault warning triggers a source tracing requirement, the five-dimensional deviation vector at the current moment is used. Fault feature fingerprint vectors of each candidate faulty component The weighted cosine similarity is calculated using the following formula: In the formula: Representing the The diagnostic weight coefficients for each dimension are determined from the historical fault case database using an information gain ratio algorithm; [The text abruptly ends here, likely due to an incomplete sentence or missing information.] Candidate faulty components corresponding to the maximum value As a result of fault tracing, the corresponding component name, installation location number, and similarity score are also output; S3. Three-level protection and linkage execution: The edge intelligent self-diagnosis core unit 5 outputs the graded protection command generated in step S2 to the graded protection execution and linkage unit 6. The graded protection execution and linkage unit 6 executes differentiated actions according to the command level. When it receives the first-level command, it immediately executes the circuit breaker tripping of the fault circuit, mechanical interlocking of the circuit, and the highest level audible and visual alarm. When it receives the second-level command, it executes the graded unloading of non-critical loads, increases the speed of the cooling fan, and provides a medium-level audible and visual warning. When it is under the third-level normal operation command, it adaptively adjusts the start and stop of the fan and its speed according to the real-time temperature and condensation status in the cabinet. When the incoming power supply fails, it automatically triggers the dual power supply switching linkage to ensure the continuity of power supply to critical loads. After all actions are executed, the execution results are sent back to the edge intelligent self-diagnosis core unit 5. The specific execution strategy for load grading and offloading is as follows: The load circuits connected to the distribution cabinet are pre-classified into Level 1 critical loads, Level 2 critical loads, and Level 3 non-critical loads according to their importance level, and stored in the local data storage unit of the edge intelligent self-diagnosis core unit 5; when the graded protection command is Level 2, the load graded unloading control unit obtains the real-time health score of each circuit. The order is as follows: first, disconnect the third-level non-critical loads; if the unloading requirements are still not met after disconnecting all third-level non-critical loads, then disconnect the second-level critical loads. Within each criticality level, the following steps are taken: The load is removed sequentially from low to high until the preset load unloading target power is met. The target power for load unloading is calculated using the following formula: In the formula: Represents the current total load power. This represents the upper limit of safe power supply determined based on power supply capacity and thermal stability limits; in dual-power supply switching scenarios, when the main incoming power supply voltage is detected to be lower than 85% of the rated voltage and the duration exceeds the switching delay. When the dual power supply switching linkage module first sends a switching request to the edge intelligent self-diagnosis core unit 5, the edge intelligent self-diagnosis core unit 5 first executes the above-mentioned load grading and unloading strategy after receiving the request, reduces the total load power to below the preset ratio of the rated capacity of the backup power supply, and then sends a permission switching command to the dual power supply switching linkage module to complete the main and backup power supply switching.

[0043] S4. Cloud Collaboration and Human-Machine Interaction Management: The communication and cloud collaboration unit 7 uploads the sensing data collected in step S1, the diagnostic results and comprehensive health score generated in step S2, and the execution status data in step S3 to the cloud platform. It also receives multi-cabinet cluster load scheduling instructions and maintenance work orders issued by the cloud. The human-machine interaction and emergency operation unit 8 displays the system topology, loop status, health score, and fault handling guidance in real time through the industrial touch screen. It also monitors the input status of the physical emergency trip button and reset button in real time. When a button action is detected, the corresponding emergency operation instruction is injected into the edge intelligent self-diagnosis core unit 5 with the highest interrupt priority, taking precedence over all currently executing automatic control processes.

[0044] Example 6: Synchronous Construction of a Five-Dimensional Sensing Data Matrix In step S1, the electrical parameter acquisition module collects instantaneous values ​​of three-phase voltage and current at a sampling rate of 12.8kHz, obtains harmonic content through FFT calculation, and simultaneously calculates zero-sequence current; the contact characteristic detection module injects low-frequency detection current into the busbar joints and circuit breaker contacts at a 1-minute cycle and calculates the contact resistance sequence; the thermal state sensing module collects the temperature of each heat point and the temperature and humidity inside the cabinet at a 10-second cycle through a surface acoustic wave temperature sensor; the mechanical state sensing module simultaneously collects the travel-time curve, coil current waveform, and vibration acceleration at a sampling rate of 50kHz when the circuit breaker is detected to have operated; the insulation and environmental sensing module collects the partial discharge ultrasonic amplitude, smoke concentration, and water immersion status at a 1-second cycle. After outlier removal and linear interpolation completion using the 3σ criterion, the data from each module are time-aligned using the GPS / BDS second pulse as the synchronization reference to ensure that the time deviation of each dimension does not exceed ±1ms. After dimensional normalization, a five-dimensional sensing data matrix is ​​constructed and transmitted to the edge intelligent self-diagnosis core unit 5. When the power distribution cabinet is under heavy load, the sampling period of the thermal state sensing module can be automatically increased to 5 seconds to improve the sensitivity of capturing temperature rise trends.

[0045] In step S2, the millisecond-level hard fault diagnosis engine calculates the three-phase current surges using a sliding window of one power frequency cycle. The current surge rate threshold Kset is set to 8 times the rated current per millisecond, the short-circuit current absolute value threshold I_scmax is set to 10 times the rated current, the zero-sequence current threshold I0th is set to 5% of the rated current, and the partial discharge ultrasonic threshold Uth is set to 3 times the standard deviation of the background noise. All of these thresholds can be adjusted on-site via an industrial touchscreen display. When the current surge of any phase exceeds the threshold and the instantaneous value exceeds the short-circuit threshold, it is determined to be a phase-to-phase short-circuit fault, and a first-level trip command is generated. When the zero-sequence current exceeds the standard and the partial discharge ultrasonic amplitude exceeds the standard, it is determined to be a combined leakage and partial discharge fault, and a first-level trip command is generated. When only the zero-sequence current exceeds the standard, it is determined to be a general leakage fault, and a second-level warning command is generated. The entire process is executed in the Cortex-M core using hardware interrupts, and the total processing delay from data input to instruction output does not exceed 5ms. The diagnostic engine also has built-in reclosing logic. If the fault characteristics do not reappear within 200ms after the first-stage trip, automatic reclosing is allowed; otherwise, it is permanently blocked.

[0046] In step S2, the progressive fault prediction and health assessment model calculates the comprehensive health score H every 5 minutes when no hard fault is triggered. This period is automatically shortened to 2 minutes during periods of severe load fluctuation and extended to 15 minutes during stable nighttime loads. The warning threshold is dynamically adjusted as the equipment ages, increasing by 0.02 for each additional year of operation and by 0.01 for every additional 1000 cumulative operations. The scale compression factor β ranges from 0.1 to 0.5, with 0.2 preferred for newly commissioned equipment and 0.4 preferred for equipment that has been in operation for more than 5 years. The edge intelligent self-diagnosis core unit 5 automatically fits the optimal β value based on historical health change curves. The warning release adopts a false alarm prevention mechanism: a progressive fault warning is officially released and component-level fault tracing is triggered only when the H value is low for three consecutive calculation cycles.

[0047] In step S2, the component-level fault tracing and localization algorithm calculates a weighted cosine similarity between the five-dimensional deviation vector D and the fault feature fingerprint vector of the candidate faulty component set after a fault occurs. Candidate components include circuit breaker contacts, busbar lap joints, cable terminals, insulating supports, and operating mechanisms. The diagnostic weight coefficient ωk for each dimension is determined from the historical fault case database using an information gain ratio algorithm, and the sum of the weights for each dimension is normalized to 1. When the maximum similarity exceeds 0.6, the corresponding component name, installation location number, and similarity score are output; if the maximum value is below 0.6, the tracing result is simultaneously marked with a "low credibility" warning, suggesting confirmation through on-site inspection.

[0048] In step S3, the graded protection execution and linkage unit 6 executes differentiated actions according to the command level. The first-level command triggers the fault circuit to trip, mechanically lock, and the highest-level alarm with a continuous red light and a sound pressure level of not less than 90dB; the second-level command triggers the load to be unloaded in stages, enhances heat dissipation, and provides a medium-level warning with a yellow 1Hz flashing light and intermittent audible ringing; the third-level command triggers the fan to adaptively adjust and the green indicator light to remain on.

[0049] During load grading and unloading, the edge intelligent self-diagnostic core unit 5 executes the process according to a preset order of importance: first, it cuts off level 3 non-critical loads; if the unloading target is still not met after all loads are cut off, then it cuts off level 2 critical loads; within the same level, loads are cut off sequentially from low to high based on their health score Hk, until the target unloading power is reached. In a dual-power supply switching scenario, when the incoming power supply voltage is lower than 85% of its rated value for more than 3 seconds, load grading and unloading is first performed to reduce the total power to below a preset proportion of the backup power supply's rated capacity, and then a "disconnect first, then reconnect" switch is performed. The entire switching process is completed within 100ms.

[0050] In step S4, the communication and cloud collaboration unit 7 uploads the sensing data, diagnostic results, health scores, and protection action records to the cloud via a 5G module. The data is encapsulated in JSON format and encrypted using AES-256. The ring buffer capacity of the breakpoint resume unit is 2GB, and it automatically overwrites the oldest historical data when the cache utilization reaches 80%. The industrial touch screen displays the system's single-line topology, real-time electrical parameters, health score bar charts for each circuit, and 90-day historical trend curves in a paginated interface. When a fault occurs, a standardized handling flowchart automatically pops up. The physical emergency trip button adopts an anti-accidental touch design with a transparent flip-top protective cover. The contact signals of the button and the physical reset button are directly connected to the GPIO interrupt input pin of the edge intelligent self-diagnosis core unit 5 via hard wiring. When pressed, the corresponding operation is triggered with the highest priority. No condition judgment or delay processing is performed in the interrupt service routine; it directly drives the tripping or reset action. The graded audible and visual alarm indicator lights and buzzers are located in prominent positions on the outside of the cabinet door, and the light colors and sound modes of each alarm level meet the relevant requirements of GB / T38113-2019.

[0051] 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. An intelligent integrated low-voltage distribution cabinet with fault self-diagnosis, comprising a cabinet (1), wherein a control panel (2) is provided inside the cabinet (1), characterized in that: The control console (2) integrates a main circuit power distribution component (3), a five-dimensional multi-source heterogeneous sensing unit (4), an edge intelligent self-diagnosis core unit (5), a hierarchical protection execution and linkage unit (6), a communication and cloud collaboration unit (7), and a human-computer interaction and emergency operation unit (8). The cabinet (1) is equipped with heat dissipation shells (9) on both the left and right sides, and each heat dissipation shell (9) is equipped with a fan (10) for heat dissipation and dehumidification. The signal output terminal of the five-dimensional multi-source heterogeneous sensing unit (4) is connected to the signal input terminal of the edge intelligent self-diagnosis core unit (5) to synchronously collect multi-source heterogeneous operating status data of the power distribution cabinet in electrical dimension, contact characteristic dimension, thermal state dimension, mechanical state dimension and insulation environment dimension, and transmit the multi-source heterogeneous operating status data to the edge intelligent self-diagnosis core unit (5) after performing time alignment and dimensional normalization preprocessing. The first control output terminal of the edge intelligent self-diagnosis core unit (5) is connected to the control input terminal of the hierarchical protection execution and linkage unit (6), which is used to perform hierarchical diagnosis including millisecond-level hard fault discrimination and progressive fault prediction based on preprocessed multi-source heterogeneous data, and then output the hierarchical diagnosis results to the hierarchical protection execution and linkage unit (6) after merging the hierarchical diagnosis results to generate a comprehensive health score and corresponding hierarchical protection instructions. The overall health score H is calculated as follows: First, construct a five-dimensional state deviation vector, expressed as: ; in, Represents the deviation in electrical dimensions. Represents the deviation in contact characteristic dimension. Represents the deviation of the thermal state dimension. Represents the deviation of the mechanical state dimension. Represents the deviation of the insulation environment dimension; The formulas for calculating the deviation of each dimension are as follows: ; in, Representing the Real-time detection feature values ​​of dimensions Representing the The nominal reference value of the dimension, Representing the Fault threshold of dimensions; Secondly, the covariance matrix is ​​constructed based on the five-dimensional deviation vector, and its expression is: ； Solving the covariance matrix Maximum eigenvalue and the corresponding feature vectors ; Finally, the overall health score is calculated using the following formula: ; In the formula: This represents the preset scale compression factor. Represents the hyperbolic tangent function; The status feedback terminal of the graded protection execution and linkage unit (6) is connected to the status feedback input terminal of the edge intelligent self-diagnosis core unit (5) to execute graded protection actions and auxiliary linkage actions and transmit the execution status back in real time; The communication and cloud collaboration unit (7) is bidirectionally connected to the edge intelligent self-diagnosis core unit (5), and the human-computer interaction and emergency operation unit (8) is bidirectionally connected to the edge intelligent self-diagnosis core unit (5), respectively used to realize local and cloud data collaborative interaction, operation status display and priority input of emergency operation instructions.

2. The intelligent integrated low-voltage distribution cabinet with fault self-diagnosis according to claim 1, characterized in that: The five-dimensional multi-source heterogeneous sensing unit (4) includes an electrical parameter acquisition module, a contact characteristic detection module, a thermal state sensing module, a mechanical state sensing module, and an insulation and environment sensing module. The electrical parameter acquisition module is used to acquire the three-phase voltage, three-phase current, active power, reactive power, harmonic content and zero-sequence current of each circuit at the first sampling frequency, and automatically perform fault recording when any electrical parameter exceeds the preset recording trigger threshold to generate fault recording data with time stamp. The contact characteristic detection module is used to inject a frequency of [frequency value missing] into the busbar joints and circuit breaker contacts using a low-frequency weak current injection method. Amplitude Low-frequency detection current, synchronously measuring contact voltage drop Calculate the contact resistance using the following formula. : ; To achieve micro-ohm-level online detection of contact resistance, and to transmit the contact resistance... The timestamp of the detection time is uploaded to the edge intelligent self-diagnosis core unit (5); The thermal state sensing module includes a passive wireless surface acoustic wave temperature sensor deployed on the busbar lap surface, circuit breaker contacts and cable terminals, as well as a cabinet temperature and humidity sensor and a condensation sensor, used to synchronously collect real-time temperature data of each heat point and the cabinet average temperature, relative humidity and condensation state signals. The mechanical state sensing module includes a circuit breaker travel displacement sensor, a closing coil current acquisition unit, and a micro vibration sensor. The circuit breaker travel displacement sensor is used to acquire the travel-time curve of the circuit breaker operating mechanism during the opening and closing process. The closing coil current acquisition unit is used to acquire the current-time waveform of the closing coil. The micro vibration sensor is used to acquire the vibration acceleration signal of the cabinet during the circuit breaker operation. The insulation and environmental sensing module includes a partial discharge ultrasonic sensor, a smoke concentration sensor and a water immersion sensor, which are used to collect the amplitude of the partial discharge ultrasonic signal, the smoke concentration value and the switching quantity of the water immersion status in the cabinet, respectively. The output of each module is connected to the edge intelligent self-diagnosis core unit (5). The edge intelligent self-diagnosis core unit (5) performs unified timestamp labeling, outlier removal and missing value interpolation on the received data from each module to form a time-aligned five-dimensional perception data matrix.

3. The intelligent integrated low-voltage distribution cabinet with fault self-diagnosis according to claim 1, characterized in that: The edge intelligent self-diagnosis core unit (5) uses an industrial-grade multi-core main control chip as its hardware carrier, and is equipped with a real-time operating system and a local data storage unit. It integrates a millisecond-level hard fault diagnosis engine, a progressive fault prediction and health assessment model, a component-level fault tracing and location algorithm, and a diagnostic system self-verification and channel self-testing module. The millisecond-level hard fault diagnosis engine is used to monitor current surges in real time. and partial discharge ultrasonic amplitude ,when Exceeding the preset current surge threshold or Exceeding the preset partial discharge threshold When a hard fault occurs, it is determined within 5ms, and the fault type is identified according to the preset fault feature library, and the highest level protection trigger command is output synchronously. The progressive fault prediction and health assessment model is used to calculate the single-loop health score and the overall cabinet health score based on multi-dimensional historical data and current real-time data within a preset assessment period, without triggering a hard fault determination. ,when Below the preset warning threshold It generates progressive fault warning signals and outputs corresponding level protection instructions in real time; The aforementioned component-level fault tracing and localization algorithm is used to, after determining a hard fault or generating a progressive fault warning, convert the five-dimensional deviation vector... Input a pre-built Bayesian inference network, use deviation features in each dimension as evidence nodes and candidate faulty components as hypothesis nodes, calculate the posterior fault probability of each candidate faulty component, and select the component and location corresponding to the maximum posterior probability as the fault tracing result output. The diagnostic system self-verification and channel self-test module is used to inject amplitude values ​​into each sensing channel of the five-dimensional multi-source heterogeneous sensing unit (4) according to a preset self-test cycle. The test excitation signal is obtained, and the amplitude of the response signal of each sensing channel is monitored. ,when When the preset allowable deviation is exceeded, the corresponding sensing channel is determined to be abnormal, and a channel fault prompt message is generated. At the same time, the reliability of the diagnostic results is marked. The millisecond-level hard fault diagnosis engine, progressive fault prediction and health assessment model, component-level fault tracing and location algorithm, and diagnostic system self-verification and channel self-test module sequentially form a closed-loop processing link of data input - dual-mode hierarchical diagnosis - fault tracing output - diagnostic credibility verification.

4. The intelligent integrated low-voltage distribution cabinet with fault self-diagnosis according to claim 1, characterized in that: The graded protection execution and linkage unit (6) includes a main circuit execution element and an auxiliary linkage mechanism; The main circuit execution element includes a circuit breaker tripping control module for each circuit, a load graded unloading control unit and a circuit interlocking mechanism. The circuit breaker tripping control module is used to receive the tripping control command output by the edge intelligent self-diagnosis core unit (5) and drive the tripping tripping device of the corresponding circuit breaker to act. The load graded unloading control unit is used to receive the load unloading command and cut off non-critical load circuits in sequence according to the preset load priority order. The circuit interlocking mechanism is used to apply mechanical interlocking to the operating mechanism of the fault circuit breaker after the tripping action is executed to prevent the fault circuit from being accidentally reclosed. The auxiliary linkage mechanism includes an intelligent heat dissipation and dehumidification control unit, a dual power supply switching linkage module, and a graded sound and light alarm module. The intelligent heat dissipation and dehumidification control unit is used to receive the heat dissipation and dehumidification control command output by the edge intelligent self-diagnosis core unit (5) and dynamically adjust the start-stop status and speed of the fan (10) according to the real-time temperature and condensation status in the cabinet. The dual power supply switching linkage module is used to receive the switching command when the incoming power supply fails and drive the dual power supply automatic switching unit to complete the main and backup power supply switching. The graded sound and light alarm module is used to output differentiated sound and light alarm signals with corresponding colors, flashing frequencies, and buzzer volumes according to the level of graded protection command. The edge intelligent self-diagnosis core unit (5) outputs a graded protection instruction with a three-level instruction structure: the first level instruction corresponds to the hard fault judgment result, triggering the fault circuit to trip, the circuit to lock and the highest level of audible and visual alarm; the second level instruction corresponds to the state where the health level is lower than the warning threshold, triggering the load to be unloaded in stages, enhancing heat dissipation and the intermediate level of audible and visual alarm; the third level instruction corresponds to the normal operation state, triggering the fan to adaptively adjust and the normal operation indicator light to illuminate. After executing the graded protection command, the main circuit actuator and the auxiliary linkage mechanism transmit the circuit breaker opening and closing status, load unloading result, interlocking mechanism status, fan speed, dual power supply circuit number and alarm trigger status as execution status data back to the edge intelligent self-diagnosis core unit (5) in real time.

5. The intelligent integrated low-voltage distribution cabinet with fault self-diagnosis according to claim 1, characterized in that: The main circuit power distribution component (3) includes an incoming circuit breaker, a busbar system, a multi-circuit drawer-type outgoing unit, a reactive power compensation unit, and a dual-power automatic switching unit. The multi-circuit drawer-type outgoing unit is equipped with a drawer insertion / removal counting and positioning detection structure to record the cumulative number of insertions / removals of the drawer-type outgoing unit and detect its current positioning status. The operating mechanisms of the incoming circuit breaker and each outgoing circuit breaker have reserved sensor mounting positions for installing the travel displacement sensors of the mechanical state sensing module. The busbar system has reserved modular electromechanical interfaces for standardized and rapid access of the reactive power compensation unit and the dual-power automatic switching unit. The communication and cloud collaboration unit (7) includes an Ethernet communication module, an RS485 communication module, a LoRa communication module, a 5G communication module, and a cloud collaboration module. The Ethernet communication module and the RS485 communication module support the Modbus-RTU protocol, and the 5G communication module supports the IEC61850 protocol. The cloud collaboration module is used to upload five-dimensional perception data, hierarchical diagnosis results, comprehensive health scores, and graded protection action records to the cloud platform through the communication module, and to receive multi-cabinet cluster load scheduling instructions and operation and maintenance work order data issued by the cloud. The cloud collaboration module has a built-in breakpoint resume transmission unit, which is used to write the data to be uploaded into the cache partition of the local data storage unit in the order of timestamps when the network communication is interrupted, and resume transmission from the breakpoint after the communication is restored to ensure the integrity of the cloud data. The human-machine interaction and emergency operation unit (8) includes an embedded industrial touch screen in the cabinet door and an emergency operation component. The industrial touch screen is used to display the topology diagram of the power distribution system, the three-phase electrical parameter values ​​of each circuit, the circuit breaker opening and closing status, the health score of each circuit, the overall health score trend curve of the cabinet, and the standardized fault handling operation guidance steps in a graphical manner in real time. The emergency operation component includes a physical emergency trip button, a physical reset button, a graded sound and light alarm indicator and a buzzer. The contact signals of the physical emergency trip button and the physical reset button are directly connected to the interrupt input pin of the edge intelligent self-diagnosis core unit (5) in a hard-wired manner. When pressed at any time, the corresponding operation is triggered with the highest interrupt priority.

6. A control method for an intelligent integrated low-voltage distribution cabinet with fault self-diagnosis, using the intelligent integrated low-voltage distribution cabinet with fault self-diagnosis as described in any one of claims 1 to 5, characterized in that, Includes the following steps: S1. Synchronous acquisition and preprocessing of five-dimensional multi-source data: The five-dimensional multi-source heterogeneous sensing unit (4) triggers each sensing module with a unified synchronous clock signal to synchronously acquire five types of operating status data of the power distribution cabinet in the electrical dimension, contact characteristic dimension, thermal state dimension, mechanical state dimension and insulation environment dimension. After removing outliers and interpolating missing values, the collected data is preprocessed by timing alignment and dimensional normalization according to a unified timestamp, and a five-dimensional sensing data matrix is ​​constructed and transmitted to the edge intelligent self-diagnosis core unit (5). S2. Dual-mode hierarchical fault diagnosis and health assessment: After receiving the five-dimensional sensing data matrix, the edge intelligent self-diagnosis core unit (5) performs dual-mode hierarchical diagnosis, which includes millisecond-level hard fault discrimination and progressive fault prediction. First, the millisecond-level hard fault diagnosis engine performs fast hard fault discrimination based on current mutation rate and partial discharge ultrasonic amplitude. If a hard fault is determined to occur, the highest level protection instruction is immediately generated and the next step is executed. If no hard fault is triggered, the progressive fault prediction and health assessment model is based on the five-dimensional deviation vector. Constructing the covariance matrix Solve for the covariance matrix Maximum eigenvalue and corresponding feature vectors ,according to Calculate the overall health score, where As a preset scale compression factor, when Below the preset warning threshold At the same time, the faulty component and its location are determined by the component-level fault tracing and location algorithm, and the corresponding level of graded protection instructions are generated. During the diagnosis process, the sensor channel self-test and the reliability verification of the diagnosis results are performed simultaneously. S3, Level 3 protection and linkage execution: The edge intelligent self-diagnosis core unit (5) outputs the level protection command generated in step S2 to the level protection execution and linkage unit (6). The level protection execution and linkage unit (6) performs differentiated actions according to the command level. When the first level command is received, the fault circuit breaker is immediately tripped, the circuit is mechanically locked, and the highest level of sound and light alarm is executed. When the second level command is received, the non-critical load is unloaded, the cooling fan speed is increased, and the intermediate sound and light alarm is executed. When the third level normal operation command is in effect, the fan start-stop and speed are adaptively adjusted according to the real-time temperature and condensation status in the cabinet. When the incoming power supply fails, the dual power supply switching linkage is automatically triggered to ensure the continuity of power supply for critical loads. After all actions are executed, the execution results are returned to the edge intelligent self-diagnosis core unit (5). S4. Cloud Collaboration and Human-Machine Interaction Management: The communication and cloud collaboration unit (7) uploads the sensing data collected in step S1, the diagnostic results and comprehensive health score generated in step S2, and the execution status data in step S3 to the cloud platform, and receives the multi-cabinet cluster load scheduling instructions and maintenance work orders issued by the cloud. The human-machine interaction and emergency operation unit (8) displays the system topology, loop status, health score and fault handling guidance in real time through the industrial touch screen, and monitors the input status of the physical emergency trip button and reset button in real time. When the button action is detected, the corresponding emergency operation instruction is injected into the edge intelligent self-diagnosis core unit (5) with the highest interrupt priority, which is higher than all currently executed automatic control processes.

7. The intelligent integrated low-voltage distribution cabinet control method with fault self-diagnosis according to claim 6, characterized in that: In step S1, the construction process of the five-dimensional sensing data matrix is ​​specifically as follows: The electrical parameter acquisition module acquires the instantaneous values ​​of the three-phase voltage of each circuit at a first sampling frequency. Instantaneous values ​​of three-phase current The amplitude and phase of each harmonic are calculated in real time using Fast Fourier Transform, and the zero-sequence current is calculated simultaneously. The calculation formula is as follows: ; The contact characteristic detection module injects a frequency into the busbar joint and circuit breaker contacts according to the second sampling period. The low-frequency detection current is used to measure the contact voltage drop and calculate the contact resistance value sequence. ; The thermal state sensing module collects the temperature of each heat point using a passive wireless surface acoustic wave temperature sensor at a third sampling period. At the same time, the average temperature inside the cabinet was collected. relative humidity Condensation status switch quantity ; When the mechanical state sensing module detects a circuit breaker action event trigger signal, it synchronously acquires the travel-time curve at a fourth sampling frequency. Current-time waveform of opening and closing coils and cabinet vibration acceleration ; The insulation and environment sensing module collects partial discharge ultrasonic amplitude, smoke concentration and water immersion status switching quantity in the fifth sampling period. After performing outlier removal and linear interpolation on the raw data collected in each dimension, all data are aligned to a unified timestamp sequence using GPS / BDS timing signals as a unified clock source. Then, the dimensions of each dimension are normalized to construct a five-dimensional sensing data matrix with row vectors representing time sampling points and column vectors representing five-dimensional feature parameters. .

8. The intelligent integrated low-voltage distribution cabinet control method with fault self-diagnosis according to claim 6, characterized in that: In step S2, the specific discrimination process of the millisecond-level hard fault diagnosis engine is as follows: Real-time calculation of three-phase current abrupt changes using a sliding window method ,in For one power frequency cycle duration, the ultrasonic amplitude of partial discharge is monitored simultaneously. The instantaneous value; Preset current mutation rate threshold Short-circuit current absolute value threshold Zero-sequence current threshold and partial ultrasound threshold ; When any phase current changes abruptly And the instantaneous value of the phase current When a phase-to-phase short-circuit fault is detected, a first-level trip command is immediately generated, and the fault type is marked as a short-circuit fault. When zero-sequence current And the amplitude of partial discharge ultrasound When a serious leakage fault accompanied by partial discharge is detected, a first-level trip command is immediately generated, and the fault type is marked as a combined leakage and partial discharge fault. When only satisfying Not satisfied When a general leakage fault is detected, a second-level warning command is generated. The total processing time for the above fault diagnosis process, from data input to instruction output, does not exceed 5ms.

9. A control method for an intelligent integrated low-voltage distribution cabinet with fault self-diagnosis as described in claim 6, characterized in that: In step S2, the specific execution process of the component-level fault tracing and location algorithm is as follows: Pre-build a set of candidate faulty components, including circuit breaker contacts, busbar lap joints, cable terminals, insulation supports, and operating mechanisms. And for each candidate faulty component Establish corresponding fault feature fingerprint vectors ,in to These represent the typical deviation values ​​of the component when it fails in the electrical, contact characteristics, thermal, mechanical, and insulation environment dimensions, respectively. When a hard fault determination or a progressive fault warning triggers a source tracing requirement, the five-dimensional deviation vector at the current moment will be used. Fault feature fingerprint vectors of each candidate faulty component The weighted cosine similarity is calculated using the following formula: ; In the formula: Representing the The diagnostic weight coefficients for each dimension are determined by the information gain ratio algorithm from the historical fault case database. Select Candidate faulty components corresponding to the maximum value As a result of fault tracing, the corresponding component name, installation location number, and similarity score are also output.

10. A control method for an intelligent integrated low-voltage distribution cabinet with fault self-diagnosis as described in claim 6, characterized in that: In step S3, the specific execution strategy for load grading and unloading is as follows: The load circuits connected to the distribution cabinet are pre-classified into first-level important loads, second-level important loads and third-level non-important loads according to their importance level, and stored in the local data storage unit of the edge intelligent self-diagnosis core unit (5); When the graded protection command is at level two, the load graded unloading control unit obtains the real-time health score of each circuit. The order is as follows: first, disconnect the third-level non-critical loads; if the unloading requirements are still not met after disconnecting all third-level non-critical loads, then disconnect the second-level critical loads. Within each criticality level, the order is as follows: The load is removed sequentially from low to high until the preset load unloading target power is met. ; The target power for load unloading is calculated using the following formula: ; In the formula: Represents the current total load power. This represents the upper limit of safe power supply determined based on power supply capacity and thermal stability limits. In a dual-power supply switching scenario, when the main incoming power supply voltage is detected to be lower than 85% of the rated voltage and the duration exceeds the switching delay... When the dual power supply switching linkage module first sends a switching request to the edge intelligent self-diagnosis core unit (5), the edge intelligent self-diagnosis core unit (5) first executes the above-mentioned load grading and unloading strategy after receiving the request, reduces the total load power to below the preset ratio of the rated capacity of the backup power supply, and then sends a permission switching command to the dual power supply switching linkage module to complete the main and backup power supply.

Images