Intelligent monitoring system and method for insulation performance of new energy high-voltage components

Abstract

The present application belongs to the technical field of high-voltage electrical safety monitoring, and discloses a new energy high-voltage component insulation performance intelligent monitoring system and method, which comprises: establishing an independent full-life-cycle baseline file and generating an initial readiness signal; performing time slot multiple access scheduling and anti-interference sampling, insulation measurement and self-state self-checking, and generating insulation state data and sampling node health state data; performing transient abnormality detection on the insulation resistance, and if there is transient abnormality, triggering active diagnosis of the corresponding sampling node in the next available time slot to generate an abnormal event determination result; generating a fault diagnosis request signal, synchronously verifying the insulation abnormal state, triggering high-voltage safety cut-off through two physically independent signals, locating the fault position based on a pre-set hard binding mapping table and reporting; periodically triggering all sampling nodes to successively perform full-range self-calibration and function verification, tracking the hardware degradation trend, optimizing the sampling node parameters, and forming a full-life-cycle closed-loop management link.

Description

Technical Field

[0001] This invention relates to the field of high-voltage electrical safety monitoring technology, and more specifically, to an intelligent monitoring system and method for the insulation performance of high-voltage components in new energy systems. Background Technology

[0002] With the rapid development of industries such as new energy vehicles and energy storage power stations, the insulation performance of high-voltage systems is directly related to the safety of equipment and personnel. Traditional insulation monitoring technologies are unable to cope with complex operating conditions, resulting in frequent false alarms and missed alarms, which can easily lead to safety accidents.

[0003] Most existing insulation monitoring solutions cannot distinguish between capacitive disturbances caused by the charging and discharging of Y capacitors and actual resistive insulation faults based on physical principles. This results in a high false alarm rate, and frequent false alarms significantly reduce the reliability of the system. Furthermore, the safety protection mechanisms rely heavily on the single software decision of the central processing unit, which poses a single point of failure risk. Any software failure or hardware malfunction may cause the safety protection function to fail, thus failing to meet the ASIL-D level functional safety requirements. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: an intelligent monitoring system for the insulation performance of high-voltage components in new energy systems, comprising: Initial calibration module: wakes up each sampling node sequentially according to the global synchronization clock to perform initial self-calibration, establishes an independent full life cycle baseline file, and generates an initial ready signal; Real-time monitoring module: Based on the full life cycle baseline file and initial ready signal, it performs time slot multiple access scheduling and anti-interference sampling according to the global synchronization clock. Each sampling node performs insulation measurement and self-check of its own status in parallel, dynamically adapts to the working conditions to perform insulation measurement compensation, and generates insulation status data and sampling node health status data. Anomaly Diagnosis Module: Based on real-time insulation status data, it performs transient anomaly detection on insulation resistance. If a transient anomaly is found, it triggers the corresponding sampling node to perform active diagnosis in the next available time slot. It distinguishes the anomaly type through preset hard logic rules and generates anomaly event judgment results. Safety protection module: Based on the abnormal event judgment result, it generates a fault diagnosis request signal, synchronously verifies the abnormal insulation state, triggers high voltage safety disconnection through two physically independent signals, locates the fault location based on the preset hard-binding mapping table and reports it. Health Management Module: Based on the health status data and fault records of sampling nodes collected throughout the entire operation cycle, it periodically triggers all sampling nodes to perform full-range self-calibration and functional verification, tracks hardware degradation trends, optimizes sampling node parameters, and forms a closed-loop management link for the entire life cycle.

[0005] Furthermore, the establishment of the full lifecycle baseline profile includes the following methods: After the device is powered on, it performs an internal end-to-end hardware self-test and generates and broadcasts a global synchronization clock. Based on a global synchronization clock, each sampling node is woken up sequentially according to its physical address. Each sampling node activates multiple dynamic reference sources within its dedicated measurement time slot to perform full-range self-calibration and injects a non-characteristic frequency single pulse for the first coupled network diagnosis. Based on the self-calibration and coupled network diagnostic results, an independent full life cycle baseline profile is established for each sampling node, and an initial ready signal is generated synchronously.

[0006] Furthermore, the methods for performing time-slot multiple access scheduling and anti-interference sampling include: Based on the received initial ready signal and global synchronization clock, time is divided into fixed dedicated time slots, and time slot tokens are broadcast cyclically to authorize a single sampling node to exclusively work, while the other sampling nodes maintain high-impedance isolation. Simultaneously, the operating status of all converters is acquired in real time, and non-integer multiple injection signal frequency commands are dynamically sent to the currently working sampling nodes. After receiving the frequency command, the sampling nodes sample within the quiet window where the interference of the converter switching cycle is weakest, and obtain raw measurement data with high anti-interference.

[0007] Furthermore, the methods for generating insulation status data and sampling node health status data include: Based on the original measurement data, the original value of insulation resistance is generated; online diagnosis of the status of the coupling network is performed in time slots, and if an abnormality is detected, it is determined to be a coupling network fault and reported. Based on the full life cycle baseline archive, the initial equivalent resistance of the coupling network is extracted as the baseline value. The current equivalent resistance of the coupling network is compared with the original value of the insulation resistance that exceeds the preset deviation range, and dynamic compensation is performed to obtain the final insulation status data. Throughout the main measurement channel operation, the mirror self-test channel synchronously and in parallel performs hardware comparison and self-testing, monitors the health status of the measurement link in real time, and generates health status data of the sampling nodes.

[0008] Furthermore, the method for transient anomaly detection of insulation resistance includes: It continuously receives real-time insulation status data uploaded by each sampling node and monitors the maximum transient change in insulation resistance through a preset sliding time window; If the detected change exceeds the preset transient anomaly threshold, it is determined that a transient anomaly event has occurred. The event is sorted by occurrence time and the next available dedicated measurement time slot is allocated. An active diagnostic command is then sent to the corresponding sampling node.

[0009] Furthermore, the method for generating the abnormal event determination result includes: After receiving the active diagnostic command, the sampling node transmits a diagnostic pulse with specified parameters in the allocated dedicated diagnostic time slot and collects the complete current response waveform; then it extracts the response duration and peak stability as core features, and distinguishes four types of anomalies, namely Y capacitor disturbance, real insulation fault, external interference and measurement circuit fault, through preset hard logic rules. For current response waveforms with unclear core characteristics, multi-pulse consistency verification is triggered to generate abnormal event judgment results.

[0010] Furthermore, the method of triggering high-voltage safety disconnection through two physically independent signals includes: When the abnormal event is determined to be a real insulation fault, a fault diagnosis request signal is generated; the common mode voltage of the corresponding high-voltage branch is monitored synchronously, and after verifying and determining the insulation abnormality, an insulation crisis signal is generated. The fault diagnosis request signal and the insulation failure signal are sent simultaneously to the non-bypassable hardware AND gate of the high-voltage safety circuit to jointly trigger the high-voltage safety disconnection; based on the preset time slot, sampling node and high-voltage branch hard-binding mapping table, the faulty high-voltage branch is located, fault information is generated and reported.

[0011] Furthermore, the method of periodically triggering all sampling nodes to sequentially perform full-range self-calibration and functional verification includes: The system continuously tracks the cumulative running time, historical calibration records, and health status data of all sampling nodes, and evaluates the monitoring accuracy by combining historical fault information. When a preset calibration trigger threshold is triggered, idle time slots are allocated to each sampling node for periodic full-range self-calibration, while simultaneously performing coupled network diagnostics and measurement link function verification. If the verification passes, the corresponding sampling node parameters are updated and the calibration results are recorded. If the verification fails, an alarm is reported and the node is marked as pending maintenance.

[0012] Furthermore, the methods for tracking hardware degradation trends and optimizing sampling node parameters include: Based on historical calibration data, calibration results, and full-cycle operation fault records, the long-term change trend of calibration coefficients at each sampling node is analyzed through degradation analysis logic to identify hardware degradation trends, thereby identifying hardware degradation types and classifying degradation levels. Based on the degradation level and monitoring accuracy, the calibration cycle and sampling frequency of the sampling nodes are dynamically adjusted to verify the optimization effect and provide feedback to correct the degradation analysis logic, forming a closed-loop management link for the entire life cycle.

[0013] Furthermore, the intelligent monitoring method for the insulation performance of high-voltage components in new energy sources is characterized by including: S1: Wake up each sampling node sequentially according to the global synchronization clock to perform initial self-calibration, establish an independent full life cycle baseline file, and generate an initial ready signal; S2: Based on the full life cycle baseline file and initial ready signal, time slot multiple access scheduling and anti-interference sampling are performed according to the global synchronization clock. Each sampling node performs insulation measurement and self-check of its own status in parallel, dynamically adapts to the working conditions to perform insulation measurement compensation, and generates insulation status data and sampling node health status data. S3: Based on real-time insulation status data, transient anomaly detection is performed on insulation resistance. If a transient anomaly exists, the corresponding sampling node is triggered to perform active diagnosis in the next available time slot. The anomaly type is distinguished by preset hard logic rules, and an anomaly event judgment result is generated. S4: Based on the abnormal event judgment result, generate a fault diagnosis request signal, simultaneously verify the abnormal insulation state, trigger high voltage safety disconnection through two physically independent signals, locate the fault location based on the preset hard-binding mapping table and report it. S5: Based on the health status data and fault records of the sampling nodes collected throughout the entire operation cycle, all sampling nodes are periodically triggered to perform full-range self-calibration and functional verification in sequence, track hardware degradation trends, optimize sampling node parameters, and form a closed-loop management link for the entire life cycle.

[0014] The technical effects and advantages of the intelligent monitoring system and method for insulation performance of new energy high-voltage components of this invention are as follows: This invention focuses on insulation monitoring in new energy high-voltage systems, and constructs a complete technical system from initial calibration, real-time monitoring, anomaly diagnosis, safety protection to full life cycle management. It specifically addresses the core pain points of existing technologies, such as high false alarm rate and low safety level, and achieves high-precision, high-reliability, and high-safety insulation performance monitoring. First, an independent full-lifecycle baseline profile is established for each sampling node through initial calibration, providing a unified benchmark for subsequent measurement compensation and degradation analysis, and ensuring long-term stability of measurement accuracy; Secondly, a fixed time slot multiple access scheduling mechanism is adopted to eliminate signal interference between nodes from the physical source. Combined with the non-integer multiple injection signal with converter switching frequency adaptive and silent window sampling technology, the anti-interference capability in complex electromagnetic environments is greatly improved. Then, by detecting transient changes in insulation resistance in real time through a sliding time window, active diagnostic pulses are triggered. Based on the current response characteristics, hard logic four-classification judgment accurately distinguishes Y capacitor disturbances, real faults, external interference and measurement circuit faults, which greatly reduces the false alarm rate. Next, a dual-source independent adjudication mechanism of central diagnostic unit and pure hardware shadow monitoring unit is adopted. The two physically independent signals jointly trigger the non-bypassable high voltage safety cut-off, and the safety level reaches ASIL-D, ensuring that a single fault will not cause the safety function to fail. Finally, based on full-cycle health data tracking of hardware degradation trends, refined dynamic optimization of operating parameters and predictive maintenance are achieved, forming a complete closed-loop management chain throughout the entire lifecycle. This invention, through modular design and integration of multiple technologies, comprehensively improves the accuracy, safety and reliability of high-voltage insulation monitoring systems. It is applicable to various high-voltage electrical scenarios such as new energy vehicles, energy storage power stations, and photovoltaic power stations, and significantly reduces the risk of safety accidents and operation and maintenance costs. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the intelligent monitoring system for insulation performance of new energy high-voltage components according to the present invention; Figure 2 This is a schematic diagram of the active diagnosis and anomaly type determination process in the intelligent monitoring system for insulation performance of new energy high-voltage components of the present invention; Figure 3 This is a schematic diagram of the intelligent monitoring method for insulation performance of new energy high-voltage components according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1: Please refer to Figure 1 and Figure 2 As shown in this embodiment, the intelligent monitoring system for the insulation performance of new energy high-voltage components includes: Initial calibration module: wakes up each sampling node sequentially according to the global synchronization clock to perform initial self-calibration, establishes an independent full life cycle baseline file, and generates an initial ready signal; Real-time monitoring module: Based on the full life cycle baseline file and initial ready signal, it performs time slot multiple access scheduling and anti-interference sampling according to the global synchronization clock. Each sampling node performs insulation measurement and self-check of its own status in parallel, dynamically adapts to the working conditions to perform insulation measurement compensation, and generates insulation status data and sampling node health status data. Anomaly Diagnosis Module: Based on real-time insulation status data, it performs transient anomaly detection on insulation resistance. If a transient anomaly is found, it triggers the corresponding sampling node to perform active diagnosis in the next available time slot. It distinguishes the anomaly type through preset hard logic rules and generates anomaly event judgment results. Safety protection module: Based on the abnormal event judgment result, it generates a fault diagnosis request signal, synchronously verifies the abnormal insulation state, triggers high voltage safety disconnection through two physically independent signals, locates the fault location based on the preset hard-binding mapping table and reports it. Health Management Module: Based on the health status data and fault records of sampling nodes collected throughout the entire operation cycle, it periodically triggers all sampling nodes to perform full-range self-calibration and functional verification, tracks hardware degradation trends, optimizes sampling node parameters, and forms a closed-loop management link for the entire life cycle.

[0018] The methods for establishing a full lifecycle baseline profile include: After the insulation testing equipment cluster is powered on, its core central diagnostic unit starts and performs an internal full-link hardware self-test: checks the timeout reset function of the internal independent watchdog circuit; checks the status of the power monitoring chip (a dedicated chip for monitoring abnormal power supply voltage) to confirm that the overvoltage and undervoltage protection thresholds are set correctly; initializes the CAN FD communication controller and performs an internal loopback test to verify that the communication function is normal. If any self-test fails, the central diagnostic unit immediately enters fault mode, illuminates the fault indicator light, and prohibits subsequent operations; if all self-tests pass, the central diagnostic unit generates a global synchronization clock signal based on its own clock and continuously broadcasts it to all sampling nodes. Based on the global synchronization clock, the central diagnostic unit starts timing scheduling according to the preset 100ms supercycle structure. Each supercycle is divided into a 1ms broadcast time slot and a 99ms measurement time slot. Within the first super-cycle broadcast time slot, wake-up commands are sent one by one according to the preset physical addresses of the sampling nodes in ascending order; After receiving a wake-up command matching its own address, each sampling node performs internal power-on and measurement circuit initialization operations and sends a wake-up response signal to the central diagnostic unit. After receiving the response signal, the central diagnostic unit allocates the next available dedicated measurement time slot to the sampling node. The awakened sampling node activates its internally integrated multi-segment dynamic reference source (including 6 precision voltage references) within its allocated dedicated measurement time slot to perform full-range self-calibration: The reference source injects multiple standard reference voltages covering the full input range of the ADC into the front end of the main measurement channel and the mirror self-test channel of the ADC (analog-to-digital converter) through an electronic switch; the two channels simultaneously sample each reference voltage multiple times and take the average value. Based on the sampling results, the sampling node uses internal hardware multipliers and adders to fit the transmission characteristic curves of the two measurement links (digital curves describing the relationship between the ADC input voltage and the output digital quantity, whose standard expression is: ADC output digital quantity = gain coefficient × input reference voltage + zero bias coefficient) and calculates the gain coefficient (ADC output digital quantity ÷ input voltage) and the zero bias coefficient (output digital quantity when the ADC input is 0V). The obtained gain coefficient and zero bias coefficient are written as calibration coefficients into the ADC hardware calibration register, and the calibration is automatically applied during subsequent ADC sampling. Then, the fitting error of each voltage range is calculated (fitting error = |(measured value - true value) ÷ true value| × 100%, where the measured value is the voltage value obtained by the ADC acquisition and conversion, and the true value is the standard reference voltage value output by the reference source). Ranges with errors exceeding 2% are marked as accuracy abnormalities and stored in the non-volatile memory inside the sampling node. If the fitting error of any range exceeds 5%, it is judged as a self-calibration failure. After completing self-calibration, in the last 1ms of the same dedicated measurement time slot, the sampling node activates the non-characteristic frequency pulse generator to inject a single pulse signal with an amplitude of 5V and a pulse width of 5μs into the coupling network; the ADC acquires the complete pulse response waveform at a sampling frequency of 1MHz, covering the rise, peak and decay processes of the pulse response; The hardware peak detection circuit locates the peak voltage of the response waveform. Starting from the peak point, it searches backward to find the sampling point where the voltage decays to 0.5 times the peak voltage, and records the half-peak voltage and corresponding time at that point. Based on the peak voltage, half-peak voltage, and corresponding time point, the equivalent parameters of the coupling network (including equivalent capacitance and equivalent resistance) are calculated according to the zero-input response formula of the RC circuit. The product of the equivalent resistance and equivalent capacitance of the coupling network is calculated to obtain the time constant of the coupling network. If the time constant of the coupled network exceeds the preset range threshold (default is [1μs, 100μs]), it is determined to be a coupled network fault, and the sampling node sends a fault code to the central diagnostic unit. Based on the self-calibration and coupling network diagnostic results, each sampling node packages its own physical address and serial number, main measurement channel calibration coefficient, mirror self-test channel calibration coefficient, accuracy abnormality interval marker, initial equivalent resistance of the coupling network, and initial equivalent capacitance of the coupling network into a data packet, and uploads it to the central diagnostic unit through the hard-wired synchronization bus. The central diagnostic unit assigns a unique file number to each sampling node, uses the physical address of the sampling node as an index, writes all initial data into non-volatile memory, and establishes an independent full lifecycle baseline file for each sampling node. After all sampling nodes have completed initial calibration and baseline establishment, the central diagnostic unit synchronously generates an initial ready signal.

[0019] Methods for performing time-slot multiple access scheduling and anti-interference sampling include: Upon receiving the initial ready signal, the central diagnostic unit immediately initiates a fixed-slot multiple access scheduling mechanism based on a global synchronization clock to ensure that only one device sends a signal at a time, thereby eliminating mutual interference at its physical source. The time is divided into fixed dedicated time slots: the scheduling cycle adopts a preset 100ms supercycle structure, and each supercycle is strictly divided into a 1ms broadcast time slot and a 99ms measurement time slot, with a time division error of ≤1μs; In the first 1ms broadcast time slot of each supercycle, the central diagnostic unit broadcasts a time slot token (an instruction frame used to authorize a specified sampling node to exclusively occupy the working rights in the corresponding measurement time slot) through the hard-wired synchronization bus. The tokens are sent in a cyclical manner according to the sampling node's physical address from smallest to largest, ensuring that all sampling nodes obtain working rights in a fixed order. All sampling nodes continuously listen to the time slot token on the hard-wired synchronization bus, and then compare the target address in the token with their own physical address. Only sampling nodes whose addresses match exactly will obtain exclusive working rights for the current measurement time slot, immediately close their internal excitation switch, and activate the measurement circuit. The measurement circuits of all other sampling nodes are in a high-impedance isolation state and are prohibited from transmitting any signals. When the measurement time slot ends, the currently working sampling node will disconnect the excitation switch, return to the listening state, and wait for the next token authorization. Real-time acquisition of operating status information for all converters, including: current switching frequency, modulation phase, and switching cycle of the converter; If a change in the converter switching frequency is detected, the injection signal frequency adjustment process is immediately triggered, and a non-integer multiple injection signal frequency command is dynamically issued to the currently operating sampling node: The central diagnostic unit calculates the optimal injection signal frequency based on the acquired converter switching frequency (optimal injection signal frequency = current switching frequency × (n + 0.618), where n is a positive integer and its value must ensure that the optimal injection signal frequency is within the operating frequency band of the sampling node injection circuit (usually 10Hz~1kHz)). The calculated optimal injection frequency value is sent to the sampling node that has obtained working authority via the hard-wired synchronization bus; After receiving the frequency command, the sampling node immediately adjusts the output frequency of the low-frequency signal injection circuit, and then calculates the start and end times of the silent window based on the converter switching cycle (start time = switching cycle × 0.2, end time = switching cycle × 0.8) to obtain the time interval of the silent window. The silent window is the period of time when electromagnetic interference is weakest after the converter switches on, which is usually the middle interval of the switching cycle. Then, within the silent window, the ADC samples the voltage and current signals of the measurement circuit. It samples multiple times within each injected signal cycle and takes the arithmetic mean as the effective sample value of that cycle to further suppress random noise and obtain raw measurement data with high anti-interference capability.

[0020] The methods for generating insulation status data and sampling node health status data include: Based on the original measurement data, within each sampling node, the original value of insulation resistance is calculated according to the unbalanced bridge method (the insulation resistance measurement principle used in this embodiment is to inject a low-frequency AC signal into the high-voltage bus and measure the leakage current to calculate the insulation resistance) using the calculation formula (original value of insulation resistance = actual amplitude of the injected signal ÷ effective value of the leakage circuit of the measurement loop - known internal resistance of the sampling loop (nominal value 10kΩ)). In the last 1ms of the current dedicated measurement time slot, the sampling node pauses insulation measurement and collects the equivalent parameters of the current coupled network; If the time constant of the current coupled network exceeds the preset range threshold (default is [1μs, 100μs]), it indicates a diagnostic anomaly and is determined to be a coupled network fault. The sampling node immediately sends a fault code to the central diagnostic unit. Based on the full lifecycle baseline archive, the initial equivalent resistance of the coupled network is extracted as the baseline value. The current equivalent resistance of the coupled network is compared with the baseline value to calculate the parameter drift rate (first calculate the difference between the current equivalent resistance and the baseline value, then calculate the ratio of the difference to the baseline value as the drift rate). If the drift rate exceeds the preset deviation range (default ±10%), it indicates that the parameters of the coupling network have changed significantly. In this case, the original insulation resistance value is dynamically compensated (by multiplying the ratio of the baseline value to the current equivalent resistance with the original insulation resistance value) to obtain the final insulation resistance value, which is used as the final insulation status data. If the drift rate does not exceed the preset deviation range, the original measured value is directly used as the final insulation resistance value. Throughout the entire measurement time slot, during the entire process of insulation measurement in the main measurement channel, the mirror self-test channel synchronously and in parallel generates the same injection signal as the main channel to perform hardware comparison self-test and monitor the health status of the measurement link in real time: real-time acquisition of the voltage of the intermediate node of the bridge in the main channel and the voltage of the intermediate node of the bridge in the mirror channel to obtain the differential voltage between the two. When the differential voltage exceeds the preset differential threshold (default 50mV) and the duration exceeds 10ms, a node hardware fault signal is generated to mark the main measurement link as abnormal. If the differential voltage does not exceed the preset differential threshold, the main measurement link is determined to be working normally. The differential voltage data and the node hardware fault signal (if any) are integrated to generate the health status data of the sampled node.

[0021] Methods for transient anomaly detection of insulation resistance include: The central diagnostic unit continuously receives real-time insulation status data and health status data of all sampling nodes via a hard-wired synchronous bus. Then, it establishes an independent sliding time window (default 10ms) buffer for each sampling node and stores all insulation resistance values ​​within the most recent time window in timestamp order. Whenever a new measurement value is received, the oldest data point in the buffer is removed. Real-time analysis is performed on the sliding time window data of each sampling node to extract the maximum and minimum values ​​of insulation resistance within the window. The absolute difference between the maximum and minimum values ​​is then used as the maximum transient change in insulation resistance within the sliding time window. The calculated maximum transient change is compared with the preset transient anomaly threshold (default 1MΩ). If the maximum transient change is greater than the transient anomaly threshold, a transient anomaly event is determined to have occurred; otherwise, normal monitoring continues. Upon detecting a transient anomaly, immediately record the key information of the anomaly, including: the physical address of the anomaly sampling node, the precise timestamp of the anomaly occurrence, the change in insulation resistance within the anomaly time window, and the current insulation resistance value. If multiple sampling nodes are detected to be abnormal at the same time, they are sorted by priority according to the time of occurrence, with the earliest abnormality receiving the highest processing priority; the sorted abnormal events are stored in the queue of abnormal events to be processed, waiting for the allocation of diagnostic time slots; Query the current time-series scheduling table, find the next unoccupied dedicated measurement time slot, and allocate the time slot to the highest priority anomaly sampling node in the pending anomaly queue without interrupting the currently ongoing measurement task, thus ensuring the timing integrity of the time slot multiple access scheduling. If all time slots are currently occupied, the allocation will be made in the first measurement time slot of the next supercycle after the current supercycle ends. After allocation, an active diagnostic command is generated and sent to the corresponding sampling node. The active diagnostic command includes: the physical address of the target sampling node, the active diagnostic mode command code, the diagnostic pulse amplitude (default 5V), and the diagnostic pulse width (in milliseconds ms, which refers to the high-level duration of the diagnostic pulse from the rising edge to the falling edge, and is set by default to more than 3 times the time constant of the total Y capacitance of the high-voltage components). The time constant of the total Y capacitance of the high-voltage components is equal to the product of the known internal resistance of the sampling circuit and the sum of the parasitic capacitance of the high-voltage components to ground. Among them, the total Y capacitance of high-voltage components refers to the sum of the parasitic capacitances of all high-voltage components to ground (offline calibration before leaving the factory), and its charging and discharging process will generate transient signals similar to real insulation faults. The diagnostic pulse width setting is based on the fact that a time constant of 3 times the total Y capacitance of the high-voltage component can ensure that the total Y capacitance of the high-voltage component is fully charged and discharged, thus distinguishing in principle from the actual resistive insulation fault and the capacitive disturbance caused by the charging and discharging of the Y capacitance. One ms before the start of the allocated diagnostic time slot, an active diagnostic command is broadcast to the target sampling node via a hard-wired synchronization bus.

[0022] The methods for generating abnormal event determination results include: Upon receiving the active diagnostic command, the sampling node immediately pauses the routine insulation measurement task and switches to the active diagnostic working mode. At the start of the allocated dedicated diagnostic time slot, it transmits a diagnostic pulse with the specified parameters (amplitude and width) in the active diagnostic command to the high-voltage DC bus. Then, the high-speed current sampling circuit synchronously acquires the complete current response waveform during the diagnostic pulse injection period (by default, 2000 sampling points need to be continuously acquired to completely record the entire process from the pulse leading edge trigger to the decay of the response current to the steady state). The hardware feature extraction circuit built into the sampling node preprocesses the original current response waveform to filter out high-frequency noise; then the peak current and corresponding peak time of the response waveform are located by the hardware peak detector. Search backward from the peak current point to find the sampling point where the current decays to 0.1 times the peak current, record the corresponding decay time, and then calculate the difference between the decay time and the peak time as the response duration; Then, the fluctuation coefficient of the peak current during the response duration is calculated (the difference between the maximum and minimum peak current during the response duration is taken and the ratio is calculated with the peak current of the response waveform to obtain the fluctuation coefficient), which is used as the peak stability. Using response duration and peak stability as core features, four types of anomalies are distinguished through preset hard logic rules: Y-capacitor disturbance, actual insulation fault, external interference, and measurement circuit fault. Specifically: The hard logic rules are a pre-defined hard logic four-category judgment table, which includes the judgment thresholds for the response duration characteristics and peak stability characteristics of the four types of anomalies. The default thresholds for identifying the four types of anomalies are as follows: Y capacitance disturbance: response duration ≤ 3 times the time constant of the total Y capacitance of the high-voltage component (i.e., the duration of the diagnostic pulse width); peak stability is not required. True insulation fault: response duration > 3 times the time constant of the total Y capacitance of the high-voltage components; peak stability ≤ 5%; External interference: No requirement for response duration; peak stability > 20% or high-frequency oscillations in the waveform; Measurement loop failure: No detectable current response; peak stability is not required. If the core characteristics cannot be clearly classified into these four categories (such as a response duration close to 3 times the time constant of the total Y capacitance of the high-voltage component, or a fluctuation coefficient between 5% and 20%), multi-pulse consistency verification is triggered. The sampling node continuously transmits three diagnostic pulses with the same parameters as the initial diagnosis. The core features are extracted from the response waveform of each pulse, and the same four-class classification judgment is performed. The diagnosis result is accepted only if the events in the three diagnoses are completely consistent; otherwise, it is considered random electromagnetic interference. If the feature parameters completely match the judgment conditions of a certain category, it is directly judged as the corresponding abnormal event type; the final abnormal event judgment result is generated, including: physical address of sampling node, precise timestamp of abnormal occurrence, event and abnormal type, response duration and peak stability, and multi-pulse verification execution status (if any).

[0023] Methods that trigger high-voltage safety disconnection using two physically independent signals include: The central diagnostic unit receives the abnormal event judgment results. When the abnormal event is read as a real insulation fault, it immediately generates a high-level active fault diagnosis request signal and sends it directly to the high-voltage contactor drive terminal of the corresponding high-voltage branch through an independent hard line without going through any intermediate processor. If the abnormal event is of other types, it only records the event log and does not trigger any safety actions. The distributed pure hardware shadow monitoring unit (deployed next to each high-voltage branch, an independent safety hardware consisting only of analog circuits, physically isolated from the central diagnostic unit, not relying on any software, clock or communication, and holding the final safety decision-making authority) monitors the common-mode voltage to ground of the corresponding high-voltage branch in real time throughout the entire process, verifying and determining insulation abnormalities: when a real insulation fault occurs, the voltage balance to ground of the entire high-voltage system is broken, and the common-mode voltage will deviate from the balance value, exceeding the hardware's safe range; when the voltage exceeds the limit, a pure analog RC delay circuit is triggered, with a fixed delay time of 10ms (used to filter out false triggers caused by transient electromagnetic interference); if the over-limit state continues for more than 10ms, an insulation abnormality is determined to have occurred, and an insulation critical signal is generated; The fault diagnosis request signal and the insulation emergency signal are simultaneously sent to the non-bypassable hardware AND gate (a pure hardware logic gate encapsulated inside the high-voltage contactor, without any software control interface, and cannot be bypassed by any program or instruction) in the high-voltage safety circuit, which together triggers the high-voltage safety disconnection: the AND gate outputs a low level only when both physically independent signals are high at the same time, directly cutting off the power supply circuit of the main contacts of the high-voltage contactor and disconnecting the high-voltage bus; if either signal is low, the AND gate remains high, and the high-voltage contactor remains closed. This dual-source independent adjudication mechanism ensures that a single hardware failure, software malfunction, or malicious attack will not lead to false or missed disconnections. Based on the preset hard-binding mapping table of time slots, sampling nodes and high-voltage branches, the central diagnostic unit records the absolute timestamp of the fault occurrence time, matches the time slot ID corresponding to that time, queries the hard-binding mapping table through the time slot ID, uniquely locks the physical installation location of the high-voltage branch where the fault occurred, and thus locates the faulty high-voltage branch. Among them, the hard-binding mapping table is a fixed mapping relationship that is burned into the central diagnostic unit memory once at the factory. Its contents are a one-to-one correspondence between the time slot ID, the physical address of the sampling node and the physical installation location of the high-voltage branch. This positioning method is a pure hardware logic matching method, with a positioning accuracy of a single high-voltage branch. Since there is no software calculation process, the positioning time can be shortened to less than 1ms. After locating the faulty high-voltage branch, the fault occurrence timestamp, the physical installation location of the faulty high-voltage branch, the physical address of the corresponding sampling node, the fault type (actual insulation resistance fault), and the insulation resistance measurement value at the time of the fault occurrence are integrated to generate fault information and report it to the upper-level controller (such as the vehicle controller or battery management control center).

[0024] Methods for periodically triggering all sampling nodes to perform full-range self-calibration and functional verification sequentially include: The central diagnostic unit continuously tracks and collects core operational data from all sampling nodes, including cumulative operating time, the most recent calibration time, historical calibration records, and all fault records. Statistical analysis of historical fault records dynamically assesses monitoring accuracy: Statistical analysis of the false alarm rate, missed alarm rate, and fault location accuracy over the past 30 days; if the accuracy rate is below 95%, the subsequent calibration cycle will be shortened by 50%. Real-time monitoring of the overall operating status ensures no insulation abnormality alarms and no sampling nodes in fault mode, thereby creating safe conditions for starting the calibration process; When a preset calibration trigger threshold is triggered, the periodic calibration process is started. Examples of calibration trigger threshold settings include: the cumulative running time of the sampling node reaches 24 hours, more than 7 days have passed since the last successful calibration, the health status data of the sampling node shows abnormal fluctuations for 3 consecutive measurement cycles, and a non-fatal measurement loop failure occurs and is restored to normal. If there is an insulation abnormality alarm or a fault event is being processed, the calibration process will be automatically postponed and will be executed immediately after the system is restored to normal. After initiating the periodic calibration process, query the timing schedule table for the next 10 supercycles, identify unoccupied idle measurement time slots, and allocate an independent idle time slot to each node to be calibrated according to the order of the physical addresses of the sampling nodes from smallest to largest. Perform full-range self-calibration in the mode of calibrating one by one while the rest operate normally, to ensure that at least 90% of the sampling nodes maintain normal insulation monitoring status at any time, without affecting the overall safety function. Full-range self-calibration method: The sampling node to be calibrated pauses routine insulation measurement within the allocated time slot, activates internal multi-segment dynamic reference sources, and sequentially injects multiple standard reference voltages covering 0%~100% of the full scale. The main measurement channel and the mirror self-test channel sample simultaneously and take the average value. The transmission characteristic curves of the two measurement links are fitted by the least squares method to calculate the new gain coefficient and zero bias coefficient, which are used as the new calibration coefficients. Synchronous execution of coupled network diagnostics and measurement link function verification: Coupled network diagnosis: Inject a 5μs non-characteristic frequency single pulse, calculate the equivalent resistance and equivalent capacitance of the current coupled network, and use them as new coupled network parameters; Measurement link function verification: The same standard signal is injected synchronously into the main channel and the mirror channel, and the output differential voltage is compared; The criteria for passing the verification are: deviation of the new calibration coefficient from the baseline value ≤ ±5%, deviation of the coupled network parameters ≤ ±20%, and differential voltage of the mirror channel ≤ 50mV; If the verification is successful: the sampling node writes the new calibration coefficients into the ADC hardware calibration register, overwriting the old parameters, uploads the new parameters to the central diagnostic unit, updates the health status of the corresponding sampling node, and records the complete data of this calibration result, including calibration time, coefficient change and verification result; If verification fails: the sampling node immediately resumes using the last valid calibration parameters, sends a calibration failure alarm to the central diagnostic unit, the central diagnostic unit marks the node as pending maintenance and displays the corresponding alarm information; After calibration, the sampling nodes resumed their regular insulation measurement tasks.

[0025] Methods for tracking hardware degradation trends and optimizing sampling node parameters include: Continuously collect and integrate the full lifecycle operation data of all sampling nodes, including periodic calibration data (gain coefficient and zero bias coefficient for each calibration, deviation rate from baseline value, calibration time and verification results), calibration results (calibration success and failure records, maintenance pending marks and alarm information), routine operation data (historical curves of equivalent parameters of coupling network, self-test differential voltage of mirror channel, cumulative running time and ambient temperature), and fault event data (all insulation abnormality alarms, false alarm and missed alarm records and fault location results). Then, using the physical address of the sampling node as a unique index, an independent health record is created for each node, and the full life cycle operation data is permanently stored. Based on historical calibration data, the long-term trend of calibration coefficients at each sampling node is analyzed using degradation analysis logic. Specifically: The degradation analysis logic uses a linear regression algorithm to fit the long-term trend of the calibration coefficients, thereby calculating the daily degradation rate of each sampling node, including the daily degradation rate of the gain coefficient and the daily degradation rate of the zero bias coefficient. Calculate the difference between the current gain coefficient obtained from the nth periodic calibration and the baseline gain coefficient when the full life cycle baseline profile was established. Then, calculate the daily degradation rate of the gain coefficient by dividing the difference by the cumulative number of operating days from the device's manufacturing date to the nth calibration. Calculate the difference between the current zero bias coefficient obtained from the nth periodic calibration and the baseline zero bias coefficient when the full life cycle baseline file was established. Then, calculate the ratio of the difference to the cumulative number of operating days from the equipment's manufacture to the nth calibration to obtain the daily degradation rate of the zero bias coefficient. For example: The baseline gain coefficient of a certain sampling node is 1000 digital units / V. After running for 100 days, the current gain coefficient is 998 digital units / V (where V stands for volt, representing the unit of voltage). Then: The daily degradation rate of the gain coefficient = (998-1000)÷100 = -0.02 (digital quantity / V) / day; indicating that the gain coefficient of this acquisition node decreases by an average of 0.02 units per day. It should be noted that the daily degradation rate is calculated using a simplified two-point method formula, which is mainly used to illustrate the basic principle. In actual scenarios, the sliding window multi-point linear regression method is usually used to calculate the daily degradation rate. The daily degradation rate of the integrated calibration coefficients (gain coefficient, zero bias coefficient) is used to obtain the calibration coefficient change curve of the sampling node, thereby forming the long-term change trend of the calibration coefficients as the hardware degradation trend. By combining the changing trends of coupling network parameters with the self-test data of the mirror channel, three typical degradation modes are identified: measurement link drift (the calibration coefficient deviation continues to increase), coupling network aging (the equivalent resistance or equivalent capacitance slowly deviates from the baseline value), and hardware performance degradation (the differential voltage of the mirror channel gradually increases). Degradation levels are categorized, and the calibration cycle and sampling frequency of sampling nodes are dynamically adjusted based on the degradation level and monitoring accuracy. Specifically: Based on the deviation between the current calibration coefficient and the baseline value, if the deviation between the gain coefficient and the zero bias coefficient is ≤±3%, the degradation level is judged to be normal, and the calibration cycle and sampling frequency remain at the default. If any deviation range is >±3% and ≤±5%, it is judged as mild degradation, the calibration cycle is shortened, and the self-test frequency of the mirror channel is increased; If the deviation range is > ±5% and ≤ ±8%, it is judged as moderate degradation, and the calibration cycle is further shortened and the sampling frequency is increased to 2kHz. If the deviation range is > ±8%, it is judged as severe degradation, an emergency maintenance alarm is generated, and it is recommended to replace the hardware; In the mild and moderate degradation levels, the optimization measures corresponding to the degradation level are refined based on the hardware degradation trend. In the case of mild degradation, if the daily degradation rate of any calibration coefficient is ≤0.01%, it is considered low mild degradation, the calibration cycle is shortened to 5 days, and the default sampling frequency is maintained; if it is >0.01% and ≤0.03%, it is considered medium mild degradation, the calibration cycle is shortened to 3 days, the self-test frequency of the mirror channel is increased to once per hour, and if it is >0.03%, it is considered high mild degradation, the calibration cycle is shortened to 2 days, and the sampling frequency is increased to 1.5kHz. In the case of moderate degradation, if the daily degradation rate of any calibration coefficient is ≤0.02%, it is considered low-to-moderate degradation, the calibration cycle is shortened to 48 hours, and the sampling frequency is increased to 1.5kHz; if it is >0.02% and ≤0.05%, it is considered medium-to-moderate degradation, the calibration cycle is shortened to 24 hours, and the sampling frequency is increased to 2kHz; if it is >0.05%, it is considered high-to-medium degradation, the calibration cycle is shortened to 12 hours, a warning is generated, and it is recommended to prepare spare parts in advance. If the current monitoring accuracy is below 95%, the calibration cycle of all sampling nodes will be shortened by 50% globally until the accuracy is restored to above 95%. Stricter parameter configurations are implemented for critical high-voltage branches (such as the power battery bus), the calibration cycle is 24 hours by default, and the alarm threshold is reduced by 50%. The central diagnostic unit packages the optimized operating parameters (including calibration cycle, sampling frequency and mirror channel self-test frequency) into instructions and sends them to the corresponding sampling nodes via the hardwired synchronization bus; Upon receiving the instruction, the sampling node immediately updates its internal parameters and sends an acknowledgment signal to the central diagnostic unit. The operation data of three complete calibration cycles were continuously monitored to verify the optimization effect: whether the calibration coefficient deviation tended to stabilize, whether the monitoring accuracy was improved to over 95%, and whether there were no new abnormal alarms or calibration failure records. If the verification passes, keep the current parameter configuration; if the verification fails, restore the original parameters and re-analyze the cause of degradation, and adjust the optimization strategy. The calibration results and the optimization effects of the operating parameters are fed back to the degradation trend analysis logic to continuously correct the calculation parameters of the daily degradation rate (which are the hyperparameters commonly used in the sliding window multi-point linear regression method to calculate the daily degradation rate, such as the sliding window size, the importance weight of calibration data at each time point, and the threshold used to identify and remove obviously abnormal calibration data points, etc.) to improve the prediction accuracy. Example of feedback correction logic: After each calibration, the actual calibration coefficient is compared with the calibration coefficient predicted based on the previous daily degradation rate, and the prediction deviation is calculated; If the prediction deviation of three consecutive calibrations is greater than 20%, the degradation rate calculation parameter correction process is triggered; the central diagnostic unit adjusts the hyperparameters and recalculates the daily degradation rate of the most recent 10 calibrations using the adjusted parameters; Verify the accuracy of the adjusted predictions: if the prediction deviation drops to within 20%, retain the new parameters; otherwise, continue adjusting. Regularly generate equipment lifecycle health reports, including the degradation status of each sampling node, remaining service life prediction, calibration plan and maintenance recommendations; When a sampling node has a calibration coefficient deviation exceeding ±10% that cannot be recovered through calibration, or fails three consecutive calibrations, or experiences a fatal hardware failure, a scrapping and replacement recommendation is generated: Once the hardware replacement is completed, the full lifecycle baseline file reconstruction is triggered, starting a new full lifecycle management cycle. This ultimately forms a closed-loop management chain covering the entire lifecycle, from data collection, degradation analysis, parameter optimization to maintenance and replacement, ensuring the long-term stable operation of the equipment.

[0026] Example 2: Please refer to Figure 3 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. A method for intelligent monitoring of the insulation performance of high-voltage components in new energy sources is provided, including: S1: Wake up each sampling node sequentially according to the global synchronization clock to perform initial self-calibration, establish an independent full life cycle baseline file, and generate an initial ready signal; S2: Based on the full life cycle baseline file and initial ready signal, time slot multiple access scheduling and anti-interference sampling are performed according to the global synchronization clock. Each sampling node performs insulation measurement and self-check of its own status in parallel, dynamically adapts to the working conditions to perform insulation measurement compensation, and generates insulation status data and sampling node health status data. S3: Based on real-time insulation status data, transient anomaly detection is performed on insulation resistance. If a transient anomaly exists, the corresponding sampling node is triggered to perform active diagnosis in the next available time slot. The anomaly type is distinguished by preset hard logic rules, and an anomaly event judgment result is generated. S4: Based on the abnormal event judgment result, generate a fault diagnosis request signal, simultaneously verify the abnormal insulation state, trigger high voltage safety disconnection through two physically independent signals, locate the fault location based on the preset hard-binding mapping table and report it. S5: Based on the health status data and fault records of the sampling nodes collected throughout the entire operation cycle, all sampling nodes are periodically triggered to perform full-range self-calibration and functional verification in sequence, track hardware degradation trends, optimize sampling node parameters, and form a closed-loop management link for the entire life cycle.

[0027] Example 3: This example discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the intelligent monitoring system for insulation performance of new energy high-voltage components described above.

[0028] Since the electronic device described in this embodiment is the electronic device used to implement the intelligent monitoring method for the insulation performance of new energy high-voltage components in this application embodiment, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the intelligent monitoring method for the insulation performance of new energy high-voltage components described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any electronic device used by those skilled in the art to implement the intelligent monitoring method for the insulation performance of new energy high-voltage components in this application embodiment falls within the scope of protection of this application.

[0029] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0030] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An intelligent monitoring system for the insulation performance of high-voltage components in new energy sources, characterized in that, include: Initial calibration module: wakes up each sampling node sequentially according to the global synchronization clock to perform initial self-calibration, establishes an independent full life cycle baseline file, and generates an initial ready signal; Real-time monitoring module: Based on the full life cycle baseline file and initial ready signal, it performs time slot multiple access scheduling and anti-interference sampling according to the global synchronization clock. Each sampling node performs insulation measurement and self-check of its own status in parallel, dynamically adapts to the working conditions to perform insulation measurement compensation, and generates insulation status data and sampling node health status data. Anomaly Diagnosis Module: Based on real-time insulation status data, it performs transient anomaly detection on insulation resistance. If a transient anomaly is found, it triggers the corresponding sampling node to perform active diagnosis in the next available time slot. It distinguishes the anomaly type through preset hard logic rules and generates anomaly event judgment results. Safety protection module: Based on the abnormal event judgment result, it generates a fault diagnosis request signal, synchronously verifies the abnormal insulation state, triggers high voltage safety disconnection through two physically independent signals, locates the fault location based on the preset hard-binding mapping table and reports it. Health Management Module: Based on the health status data and fault records of sampling nodes collected throughout the entire operation cycle, it periodically triggers all sampling nodes to perform full-range self-calibration and functional verification, tracks hardware degradation trends, optimizes sampling node parameters, and forms a closed-loop management link for the entire life cycle.

2. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 1, characterized in that, The methods for establishing the full lifecycle baseline profile include: After the device is powered on, it performs an internal end-to-end hardware self-test and generates and broadcasts a global synchronization clock. Based on a global synchronization clock, each sampling node is woken up sequentially according to its physical address. Each sampling node activates multiple dynamic reference sources within its dedicated measurement time slot to perform full-range self-calibration and injects a non-characteristic frequency single pulse for the first coupled network diagnosis. Based on the self-calibration and coupled network diagnostic results, an independent full life cycle baseline profile is established for each sampling node, and an initial ready signal is generated synchronously.

3. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 2, characterized in that, The methods for performing time-slot multiple access scheduling and anti-interference sampling include: Based on the received initial ready signal and global synchronization clock, time is divided into fixed dedicated time slots, and time slot tokens are broadcast cyclically to authorize a single sampling node to exclusively work, while the other sampling nodes maintain high-impedance isolation. Simultaneously, the operating status of all converters is acquired in real time, and non-integer multiple injection signal frequency commands are dynamically sent to the currently working sampling nodes. After receiving the frequency command, the sampling nodes sample within the quiet window where the interference of the converter switching cycle is weakest, and obtain raw measurement data with high anti-interference.

4. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 3, characterized in that, The methods for generating insulation status data and sampling node health status data include: Based on the original measurement data, the original value of insulation resistance is generated; online diagnosis of the status of the coupling network is performed in time slots, and if an abnormality is detected, it is determined to be a coupling network fault and reported. Based on the full life cycle baseline archive, the initial equivalent resistance of the coupling network is extracted as the baseline value. The current equivalent resistance of the coupling network is compared with the original value of the insulation resistance that exceeds the preset deviation range, and dynamic compensation is performed to obtain the final insulation status data. Throughout the main measurement channel operation, the mirror self-test channel synchronously and in parallel performs hardware comparison and self-testing, monitors the health status of the measurement link in real time, and generates health status data of the sampling nodes.

5. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 4, characterized in that, The methods for transient anomaly detection of insulation resistance include: It continuously receives real-time insulation status data uploaded by each sampling node and monitors the maximum transient change in insulation resistance through a preset sliding time window; If the detected change exceeds the preset transient anomaly threshold, it is determined that a transient anomaly event has occurred. The event is sorted by occurrence time and the next available dedicated measurement time slot is allocated. An active diagnostic command is then sent to the corresponding sampling node.

6. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 5, characterized in that, The methods for generating abnormal event determination results include: After receiving the active diagnostic command, the sampling node transmits a diagnostic pulse with specified parameters in the allocated dedicated diagnostic time slot and collects the complete current response waveform; then it extracts the response duration and peak stability as core features, and distinguishes four types of anomalies, namely Y capacitor disturbance, real insulation fault, external interference and measurement circuit fault, through preset hard logic rules. For current response waveforms with unclear core characteristics, multi-pulse consistency verification is triggered to generate abnormal event judgment results.

7. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 6, characterized in that, The method of triggering high-voltage safety disconnection by two physically independent signals includes: When the abnormal event is determined to be a real insulation fault, a fault diagnosis request signal is generated; the common mode voltage of the corresponding high-voltage branch is monitored synchronously, and after verifying and determining the insulation abnormality, an insulation crisis signal is generated. The fault diagnosis request signal and the insulation failure signal are sent simultaneously to the non-bypassable hardware AND gate of the high-voltage safety circuit to jointly trigger the high-voltage safety disconnection; based on the preset time slot, sampling node and high-voltage branch hard-binding mapping table, the faulty high-voltage branch is located, fault information is generated and reported.

8. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 7, characterized in that, The method of periodically triggering all sampling nodes to perform full-range self-calibration and functional verification sequentially includes: The system continuously tracks the cumulative running time, historical calibration records, and health status data of all sampling nodes, and evaluates the monitoring accuracy by combining historical fault information. When a preset calibration trigger threshold is triggered, idle time slots are allocated to each sampling node for periodic full-range self-calibration, while simultaneously performing coupled network diagnostics and measurement link function verification. If the verification passes, the corresponding sampling node parameters are updated and the calibration results are recorded. If the verification fails, an alarm is reported and the node is marked as pending maintenance.

9. The intelligent monitoring system for insulation performance of new energy high-voltage components according to claim 8, characterized in that, The methods for tracking hardware degradation trends and optimizing sampling node parameters include: Based on historical calibration data, calibration results, and full-cycle operation fault records, the long-term change trend of calibration coefficients at each sampling node is analyzed through degradation analysis logic to identify hardware degradation trends, thereby identifying hardware degradation types and classifying degradation levels. Based on the degradation level and monitoring accuracy, the calibration cycle and sampling frequency of the sampling nodes are dynamically adjusted to verify the optimization effect and provide feedback to correct the degradation analysis logic, forming a closed-loop management link for the entire life cycle.

10. A method for intelligent monitoring of insulation performance of high-voltage components in new energy sources, implemented based on claims 1 to 9, characterized in that it includes: S1: Wake up each sampling node sequentially according to the global synchronization clock to perform initial self-calibration, establish an independent full life cycle baseline file, and generate an initial ready signal; S2: Based on the full life cycle baseline file and initial ready signal, time slot multiple access scheduling and anti-interference sampling are performed according to the global synchronization clock. Each sampling node performs insulation measurement and self-check of its own status in parallel, dynamically adapts to the working conditions to perform insulation measurement compensation, and generates insulation status data and sampling node health status data. S3: Based on real-time insulation status data, transient anomaly detection is performed on insulation resistance. If a transient anomaly exists, the corresponding sampling node is triggered to perform active diagnosis in the next available time slot. The anomaly type is distinguished by preset hard logic rules, and an anomaly event judgment result is generated. S4: Based on the abnormal event judgment result, generate a fault diagnosis request signal, simultaneously verify the abnormal insulation state, trigger high voltage safety disconnection through two physically independent signals, locate the fault location based on the preset hard-binding mapping table and report it. S5: Based on the health status data and fault records of the sampling nodes collected throughout the entire operation cycle, all sampling nodes are periodically triggered to perform full-range self-calibration and functional verification in sequence, track hardware degradation trends, optimize sampling node parameters, and form a closed-loop management link for the entire life cycle.

Images