Energy storage converter overcurrent protection method and system based on non-contact current sampling

Abstract

The application provides an energy storage converter overcurrent protection method and system based on non-contact current sampling, and belongs to the technical field of cross between energy storage power electronic protection and power semiconductor application. The method comprises the following steps: collecting full current analog signals of a main loop of an energy storage converter, performing insulation isolation filtering processing on the full current analog signals, and generating main loop original current sampling data; performing analog-digital conversion and feature splitting processing on the main loop original current sampling data, separating out a main current amplitude sequence and a transient change rate sequence, and generating a double-channel parallel original sampling data set; through cooperation of a double-channel non-contact sampling architecture, the insulation requirement is fully met under a 1500V high-voltage system, and the insulation safety level of the main loop of the energy storage converter is greatly improved.

Description

Technical Field

[0001] This invention proposes an overcurrent protection method and system for energy storage converters based on non-contact current sampling, which belongs to the interdisciplinary field of energy storage power electronic protection and power semiconductor application. Background Technology

[0002] The overcurrent protection system of current 1500V energy storage converters has long suffered from structural shortcomings. Many 10kW residential energy storage converters still use the shunt resistor sampling scheme, which has high insertion loss and cannot meet the requirements for enhanced insulation under high voltage systems. Insufficient creepage distance can easily cause arcing hazards in the main circuit.

[0003] The industry commonly adopts the approach of replacing the shunt with a single Hall sensor, but this solution does not cover the extreme working conditions of lightning surge intrusion. When a surge current of 10kA to 20kA enters the main circuit, the Hall magnetic core will instantly enter a magnetic saturation state, and the output signal will be directly clamped. The system sampling end will mistakenly judge that the circuit current is in the normal range. At this time, the bridge arm SiC or IGBT device has already withstood several times the rated short-circuit current, and the device withstand time is only 2 to 5 microseconds. Conventional sampling links cannot capture this abnormal state at all.

[0004] Traditional protection architectures send the Hall output signal to the ADC port of the MCU, where it undergoes digital filtering, mean calculation, and protocol encapsulation before a PWM shutdown command is issued. The delay of this entire software link far exceeds the short-circuit tolerance window of the power device, naturally creating a protection blind zone. Existing publicly available solutions lack a cross-validation mechanism for dual-channel non-contact sampling, making it impossible to perform hard-wired waveform blocking without MCU involvement under magnetically saturated surge conditions. Consequently, overcurrent-induced failures and grid disconnections of energy storage converters remain frequent occurrences. Summary of the Invention

[0005] This invention provides an overcurrent protection method and system for energy storage converters based on non-contact current sampling, to solve the problems mentioned in the background section above: The present invention proposes an overcurrent protection method for energy storage converters based on non-contact current sampling, the method comprising: S1. Acquire the full current analog signal of the main circuit of the energy storage converter, perform insulation isolation filtering on the full current analog signal to generate the original current sampling data of the main circuit; perform analog-to-digital conversion and feature decomposition on the original current sampling data of the main circuit to separate the main current amplitude sequence and transient rate of change sequence, and generate a dual-channel parallel original sampling dataset. S2. Perform main channel amplitude verification processing on the dual-channel parallel raw sampling dataset, extract the continuous current amplitude information output by the reinforced insulation Hall sensor, and generate real-time current data of the main channel; perform time-domain continuous scanning processing on the real-time current data of the main channel to capture the step jump and clamping characteristics of the current signal and generate Hall magnetic saturation state marker data. S3. Perform auxiliary channel feature matching processing on the Hall magnetic saturation state marker data, retrieve the bridge arm parasitic di / dt feature sequence output by the differential channel of the isolation operational amplifier, and generate transient current change rate feature data; perform threshold comparison calculation on the transient current change rate feature data, identify the initial time point of lightning surge and short circuit intrusion, and generate overcurrent trigger warning data. S4. Perform channel state switching processing on the overcurrent trigger warning data, cut off the main channel amplitude output during the Hall magnetic saturation trigger period, and generate no-delay protection trigger command data; perform hardware link direct transmission processing on the no-delay protection trigger command data, skip the MCU operation link and send it directly to the high-speed comparator to generate hard-wired wave blocking drive signal data. S5. The hard-wired waveguide drive signal data is isolated and transmitted. The drive signal is directly sent to the bridge arm drive unit through an automotive-grade digital isolator to generate nanosecond-level turn-off execution data for SiC devices. The nanosecond-level turn-off execution data of SiC devices is processed by operating condition backtracking to match the differentiated withstand time parameters of SiC and IGBT to generate graded load reduction linkage control data. S6. Perform grid status synchronization processing on the graded load reduction linkage control data, output flexible control signals for power-limited operation, and generate converter non-disconnection continuous operation data; perform dual-channel status reset processing on the converter non-disconnection continuous operation data, and automatically switch back to the main channel normal sampling mode after the transient surge condition ends, generating normal overcurrent protection recovery data.

[0006] The present invention proposes an overcurrent protection system for energy storage converters based on non-contact current sampling, the system comprising: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0007] The beneficial effects of this invention are as follows: Through the cooperation of a dual-channel non-contact sampling architecture, the enhanced insulation requirements are fully met under 1500V high-voltage systems, significantly improving the insulation safety level of the main circuit of the energy storage converter. Real-time verification of the di / dt characteristics of the auxiliary channel can quickly identify the Hall magnetic saturation state in the early stages of lightning surge intrusion, effectively reducing the probability of misjudgment under extreme conditions with a single Hall sampling scheme. The hard-wired direct-drive wave-blocking link skips the MCU calculation stage, compressing the overall protection action time to less than 2 microseconds, significantly enhancing the survivability of SiC and IGBT devices under short-circuit conditions. The graded load reduction linkage control logic significantly reduces the delay blind zone of traditional software protection links, avoiding the problem of the entire unit disconnecting from the grid directly after overcurrent triggering. It can complete high-precision current sampling under normal operating conditions and achieve blind-zone-free hard protection under extreme conditions of surge short circuits, fundamentally preventing energy storage system shutdown accidents caused by overcurrent failure. Attached Figure Description

[0008] Figure 1 This is a diagram illustrating the steps of the method described in this invention; Figure 2 This is a detailed flowchart of step S1 in the present invention; Figure 3 This is a detailed flowchart of step S2 in the present invention; Figure 4 This is a detailed flowchart of step S3 in the present invention; Figure 5 This is a detailed flowchart of step S4 in the present invention; Figure 6 This is a detailed flowchart of step S5 in the present invention; Figure 7 This is a detailed flowchart of step S54 in the present invention; Figure 8 This is a detailed flowchart of step S6 in this invention. Detailed Implementation

[0009] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0010] Example 1 One embodiment of the present invention, such as Figure 1 As shown, an overcurrent protection method for energy storage converters based on non-contact current sampling is described, the method comprising: S1. Acquire the full current analog signal of the main circuit of the energy storage converter, perform insulation isolation filtering on the full current analog signal to generate the original current sampling data of the main circuit; perform analog-to-digital conversion and feature decomposition on the original current sampling data of the main circuit to separate the main current amplitude sequence and transient rate of change sequence, and generate a dual-channel parallel original sampling dataset. S2. Perform main channel amplitude verification processing on the dual-channel parallel raw sampling dataset, extract the continuous current amplitude information output by the reinforced insulation Hall sensor, and generate real-time current data of the main channel; perform time-domain continuous scanning processing on the real-time current data of the main channel to capture the step jump and clamping characteristics of the current signal and generate Hall magnetic saturation state marker data. S3. Perform auxiliary channel feature matching processing on the Hall magnetic saturation state marker data, retrieve the bridge arm parasitic di / dt feature sequence output by the differential channel of the isolation operational amplifier, and generate transient current change rate feature data; perform threshold comparison calculation on the transient current change rate feature data, identify the initial time point of lightning surge and short circuit intrusion, and generate overcurrent trigger warning data. S4. Perform channel state switching processing on the overcurrent trigger warning data. Cut off the main channel amplitude output during the Hall magnetic saturation trigger period and directly use the transient characteristics extracted from the auxiliary channel as the protection judgment basis to generate no-delay protection trigger command data. Perform hardware link direct transmission processing on the no-delay protection trigger command data, skip the MCU operation link and send it directly to the high-speed comparator to generate hard-wired wave blocking drive signal data. S5. The hard-wired waveguide drive signal data is isolated and transmitted. The drive signal is directly sent to the bridge arm drive unit through an automotive-grade digital isolator to generate nanosecond-level turn-off execution data for SiC devices. The nanosecond-level turn-off execution data of SiC devices is processed by operating condition backtracking to match the differentiated withstand time parameters of SiC and IGBT to generate graded load reduction linkage control data. S6. Perform grid status synchronization processing on the graded load reduction linkage control data, output flexible control signals for power-limited operation, and generate converter non-disconnection continuous operation data; perform dual-channel status reset processing on the converter non-disconnection continuous operation data, and automatically switch back to the main channel normal sampling mode after the transient surge condition ends, generating normal overcurrent protection recovery data.

[0011] The working principle and effects of the above technical solution are as follows: By using a dual-channel non-contact current sampling and collaborative detection method, the accuracy and instantaneous response speed of the main circuit current sampling of the energy storage converter are significantly improved, effectively reducing the probability of current detection distortion under Hall magnetic saturation conditions. Through hardware-direct transmission of the blocking control logic, the computational delay of the protection response link is shortened, reducing the protection lag problem caused by conventional software processing. This detection and protection mechanism can accurately identify lightning surges and short-circuit faults, promptly completing the rapid shutdown of power devices and avoiding the problem of device burnout and circuit breakdown caused by instantaneous large current impacts. Differentiated adaptation of the tolerance characteristics of SiC and IGBT devices enables graded load reduction regulation, which can not only meet the equipment protection requirements under extreme fault conditions but also maintain the continuous grid-connected operation of the converter, avoiding grid fluctuations caused by transient grid disconnection during faults, and effectively enhancing equipment operational stability and grid adaptability.

[0012] One embodiment of the present invention, such as Figure 2 As shown, S1 includes: S11. A ring-shaped non-contact current acquisition coil is arranged on the DC bus of the energy storage converter and the outer wall area of ​​the three-phase AC main circuit. It continuously picks up the alternating magnetic field induced waveform generated by the movement of alternating charges inside the circuit, and fully records the full current simulation waveform formed by the superposition of steady-state load current and instantaneous impact current, and summarizes it to obtain the full current simulation signal flow with full coverage. S12. The acquired full-current analog signal is fed into a multi-stage RC passive filter array with common-mode suppression and insulation isolation module to filter out the superimposed interference components caused by grid-side harmonic noise and power device switching radiation noise, and block the signal offset problem caused by leakage coupling between high and low voltage circuits. After wave rectification and noise reduction, the normalized waveform information is output and the original current sampling data of the main circuit is generated. S13. The original current sampling data of the main circuit is sent to the high-precision successive approximation analog-to-digital converter chip to carry out continuous sampling and quantization operations, and the continuously changing analog electrical signal is converted into a discrete digital code value sequence point by point. Simultaneously, two independent data operation branches are divided and storage buffer space is allocated. S14. One branch stores the current magnitude corresponding to the sampling code value and continuously arranges it to form a time-progressive main current amplitude sequence. The other branch calculates the rate of change of the current difference between adjacent sampling nodes point by point and arranges it to form a transient rate of change sequence. The two types of time sequences are stored synchronously without interfering with each other. After integration and arrangement, a dual-channel parallel original sampling dataset is generated.

[0013] The working principle and effects of the above technical solution are as follows: By using a non-contact coil to complete the acquisition of the main circuit current signal, there is no need to intrude into the main power circuit of the equipment, which effectively improves the safety of the current signal acquisition process and reduces the potential equipment failure hazards caused by physical wiring. Multi-stage filtering and insulation isolation can weaken the signal interference caused by power grid harmonics and device switching noise, reduce the offset and distortion of current sampling data, and make the acquired current waveform more consistent with the actual operating state of the equipment. High-precision quantization processing of analog signals can refine the sampling accuracy of current data, which can not only fully capture the normal current changes in steady-state operation of the equipment, but also accurately record the dynamic characteristics of instantaneous inrush current. The split processing of dual-channel independent data branches can separately retain the time-series data of current amplitude and rate of change, enhance the comprehensiveness of subsequent current state analysis and judgment, and provide accurate and reliable raw data support for subsequent overcurrent protection calculations.

[0014] In one embodiment of the present invention, S13 includes: The extended intent data is semantically normalized and field-purified, removing redundant natural language modifiers and extracting core semantic elements that fit the device's sampling and computation requirements to generate standardized intent normalized data. The standardized intent-organized data is processed by field mapping and instruction translation, transforming natural language semantic elements into digital computing instruction formats that can be recognized by hardware chips, and generating machine-recognizable primary structured instruction data. The basic structured instruction data is processed for instruction verification and format calibration, and the standardized format of instruction timing arrangement, data call port and operation execution logic is unified to generate compliant structured search instruction data; The compliant and structured search command data is decomposed into feature dimensions to extract the core feature parameters of sampling precision, operation mode and data processing dimension contained in the command, and generate command core feature dataset; The core feature dataset of the instruction is traversed and compared within the library. The functional adaptation attributes of various computational models in the preset model library are matched, and the model subject with the highest adaptation degree is selected to generate model type matching data.

[0015] The working principle and effects of the above technical solution are as follows: By progressively refining and standardizing the intent data in natural language form, invalid semantic information can be eliminated, improving the purity of the core content of user needs and reducing interference from redundant statements in instruction generation. Through standardized field mapping and instruction translation methods, a smooth conversion from natural language to digital computational instructions can be achieved, eliminating semantic barriers in human-computer interaction and improving the overall efficiency of instruction generation. A unified instruction format calibration process can standardize the execution logic and arrangement rules of various instructions, reducing operational anomalies caused by incorrect instruction formats. Decomposing and comparing the core features of instructions and matching them with corresponding computational models not only meets diverse user analysis needs but also ensures the suitability of model calls, enhancing the accuracy and adaptability of overall business processing and ensuring the stable and orderly progress of subsequent data processing flows.

[0016] One embodiment of the present invention, such as Figure 3 As shown, S2 includes: S21. Retrieve all timing sampling contents of the amplitude branch inside the dual-channel parallel original sampling dataset, and perform point-by-point amplitude deviation verification in conjunction with the reference waveform of the same source current synchronously acquired by the reinforced insulation Hall sensor unit. Eliminate small amplitude error components caused by coil arrangement deviation and environmental temperature drift, and correct the accuracy of the current value corresponding to each timing node. S22. After multi-point deviation calibration, abnormal drift sampling points are removed, and stable and continuous current value records over a long time scale are retained. The records are arranged and integrated in chronological order, and the current value records are output in a continuous and uninterrupted manner. The main channel real-time current data is then generated. S23. Construct a time-domain scanning window with a fixed sliding duration to traverse the overall time sequence of the main channel real-time current data frame by frame. Within the window, compare the fluctuation range of the values ​​before and after each sampling point to continuously identify the step jump waveform segment where the value suddenly increases and the waveform segment where the value is capped for a long time and no longer increases. S24. Mark and divide all time periods during the scanning process that are determined to be waveform capped and stagnant, and whose amplitude no longer increases synchronously with the actual current of the circuit. Assign a unique timing identifier and status code to each abnormal waveform segment, archive and arrange them in a unified manner according to the timing sequence, and generate Hall magnetic saturation status marker data.

[0017] The working principle and effects of the above technical solution are as follows: By correcting sampling data errors through point-by-point amplitude verification, the numerical deviations caused by coil layout deviations and environmental temperature changes can be offset, effectively improving the overall accuracy of main channel current sampling and reducing data drift problems caused by long-term operation. By screening and integrating continuous and stable sampling data, a complete and coherent time-series current record is output, avoiding misjudgments of current status caused by scattered abnormal sampling points. With the help of a fixed-window time-domain traversal scanning method, step fluctuations and amplitude clamping states of the current signal can be fully captured, improving the comprehensiveness of abnormal waveform identification. By annotating and archiving abnormal periods of magnetic saturation, normal current fluctuations and sensor saturation distortion conditions can be accurately distinguished, which can accurately identify real loop overcurrent anomalies and avoid sampling failure problems caused by Hall magnetic saturation, thereby improving the accuracy and reliability of subsequent overcurrent protection judgments from the source.

[0018] In one embodiment of the present invention, step S23 includes: Configure parameters for a fixed-duration time-domain sliding scan window, set a uniform frame traversal step size and sampling comparison accuracy, construct a continuous scanning detection mechanism adapted to full-time data, and generate standardized time-domain scan configuration data; The standardized time-domain scanning configuration data is called to collect the complete time sequence of the real-time current data of the main channel frame by frame, continuously recording all sampling value points in each frame window, and generating a frame-by-frame time-series scanning dataset. Perform intra-frame time-series scanning dataset comparison of numerical differences between points before and after the frame, count the current amplitude fluctuation of adjacent sampling nodes, and generate point-by-point amplitude fluctuation comparison data. By comparing the amplitude growth trend between consecutive frames, waveform segments with a significant increase in amplitude in a short period of time are selected, and step jump waveform intervals with rapid current changes are divided to generate step fluctuation interval marker data. Identify waveform segments where the amplitude value remains constant within a frame for a long period of time without incremental change, divide the clamped waveform intervals where the current amplitude stagnates, integrate the contents of the two types of abnormal waveform intervals, and generate full abnormal waveform identification data.

[0019] The working principle and effects of the above technical solution are as follows: By standardizing the time-domain scanning parameter configuration and unifying the traversal and comparison standards of current data, the regularity of current anomaly waveform detection is effectively improved, reducing the problems of missed and false judgments caused by random scanning errors. The frame-by-frame traversal of sampling points and point-by-point comparison of amplitude fluctuations can meticulously capture subtle dynamic changes in current, improving the precision of current anomaly feature identification. Classifying and distinguishing between step transitions and amplitude clamping waveform states can accurately differentiate between instantaneous fault fluctuations and sensor saturation distortion conditions, making the classification of anomaly data more consistent with the actual operating state of the equipment. The entire detection process is progressive, capable of quickly capturing short-term sudden current anomaly fluctuations and stably identifying long-term stagnant distorted waveforms, effectively enhancing the comprehensiveness and accuracy of current state identification and providing reliable data support for subsequent magnetic saturation state determination.

[0020] One embodiment of the present invention, such as Figure 4 As shown, S3 includes: S31. Retrieve all saturation anomaly time intervals recorded by Hall magnetic saturation state marker data segment by segment, lock the start and end sampling node positions corresponding to each anomaly interval, and use the node interval as the retrieval boundary to retrieve the sampling content of the isolation operational amplifier differential channel pre-stored in the auxiliary channel. S32. Extract the time-series waveform of the parasitic current change rate generated at the moment of power-on of the bridge arm power circuit within the search boundary range, remove the smooth change component caused by slow load adjustment, retain only the short-term steep rate waveform segment, and splice and integrate the multiple extracted segments to generate transient current change rate characteristic data. S33. Two types of differentiated critical judgment thresholds are preset to correspond to the power grid lightning surge impact condition and the AC / DC side metallic short circuit intrusion condition, respectively. Each steep change peak in the transient current change rate characteristic data is compared with the two types of critical thresholds in turn. S34. Mark the first sampling timing node where each peak first breaks through the corresponding threshold. This node is the initial occurrence point of various large current disturbances entering the circuit. Pack the disturbance type, intrusion time, and initial change rate values ​​together to generate overcurrent trigger warning data.

[0021] The working principle and effects of the above technical solution are as follows: By relying on dual-channel data linkage retrieval, the differential channel sampling content during the abnormal magnetic saturation period is selectively retrieved, effectively narrowing the data processing range, reducing computational consumption caused by invalid sampling data, and improving the overall efficiency of fault detection. By filtering out the smooth current change component generated by the steady-state adjustment of the load, and retaining only the waveform of the instantaneously abruptly changing current rate, the purity of transient fault characteristics is greatly improved, and the probability of fault misjudgment caused by fluctuations in normal operating conditions is reduced. By using a graded threshold comparison method to distinguish between lightning surge and metallic short circuit fault conditions, the identification criteria for different current disturbances are refined, improving the accuracy of fault type determination. By accurately locking the initial timing node of disturbance intrusion and integrating complete fault information, it can not only capture the early characteristics of sudden overcurrent faults in advance and shorten the fault response time, but also completely retain the original fault data, providing accurate and reliable judgment basis for subsequent protection actions.

[0022] In one embodiment of the present invention, S32 includes: Based on the start and end sampling nodes of the locked abnormal interval, the continuous sampling waveform of the differential channel of the auxiliary channel isolation operational amplifier is precisely segmented, and the dynamic current change waveform data generated at the moment of power-on of the bridge arm power circuit is completely collected to generate full-domain time-period current change waveform segment data. The waveform segment data of current change over the entire time period is decomposed into layers and components. The slow current fluctuation component corresponding to the normal load adjustment action of the equipment is separated, and the smooth waveform unit with the fluctuation rate conforming to the normal working condition is selected to generate steady-state load fluctuation component data. Remove all smooth waveform content corresponding to steady-state load fluctuation component data, retain current waveform units with waveform change rate exceeding the normal operating range and instantaneous change characteristics, and generate short-time steep-change waveform segment data. The time sequence alignment and seamless splicing of multiple discrete short-time abruptly changing waveform segments are performed to unify the sampling accuracy and waveform amplitude benchmark of each segment and integrate them into a complete and continuous set of current abrupt change waveforms. The integrated set of current change waveforms is processed for data normalization and noise reduction, eliminating invalid clutter points caused by sampling noise, standardizing the waveform timing layout format, and generating transient current change rate characteristic data.

[0023] The working principle and effects of the above technical solution are as follows: This solution performs precise waveform extraction for faulty and abnormal intervals, processing only the current data within the effective time period, reducing resource consumption caused by full data processing and improving the overall speed of data processing. By decomposing the current waveform components in layers, it separates the slow fluctuations generated by daily load regulation, effectively filtering out operating condition interference caused by normal equipment operation and reducing fault misjudgments caused by steady-state fluctuations. Targeted retention of transiently changing waveform content allows for precise focusing on the sudden current anomalies in the power circuit, significantly improving fault signal identification. Timing alignment and splicing regularization of multiple transient waveform segments eliminate data gaps and accuracy differences, making the waveform data more complete and coherent. The post-processing noise reduction and regularization operation removes invalid data from sampling noise, ensuring the complete preservation of transient current change characteristics and improving the purity of the final feature data, providing high-quality data support for subsequent fault threshold comparison and operating condition identification.

[0024] One embodiment of the present invention, such as Figure 5 As shown, S4 includes: S41. Analyze the disturbance start and end time information recorded in the overcurrent trigger warning data, and synchronously send a switching level signal to the sampling channel switching logic hardware unit. Within the complete duration of the Hall device entering the magnetic saturation range, disconnect the data output path of the main current amplitude channel to the subsequent circuit. S42, the transmission link of transient change rate characteristic data of the auxiliary channel is activated. All subsequent protection limit judgment calculations are carried out based on the current rate waveform collected by the differential channel, eliminating the judgment lag problem caused by current value distortion due to magnetic saturation, and integrating the channel switching logic to generate protection trigger command data without delay. S43. Abandoning the conventional data transmission process of microcontroller cache read / write and floating-point operation relay, the delay-free protection trigger instruction data is directly transmitted to the input of the high-speed voltage comparator at the back end via dedicated high-speed traces on the inner layer of the PCB in the form of hardware level pulses. S44. The high-speed comparator receives the pulse level in real time and completes the level amplitude comparison. It outputs a turn-off control pulse waveform with a fixed level amplitude. After normalizing the pulse rising edge and falling edge timing parameters, it outputs them in a unified manner to generate hard-wired waveform drive signal data.

[0025] The working principle and effects of the above technical solution are as follows: By actively switching the sampling channel when the Hall device is in a magnetic saturation state, the distorted amplitude data of the main channel is discarded, and the transient characteristics of the auxiliary channel are used to complete the protection judgment, effectively eliminating the protection delay and judgment error caused by the distortion of current values. The hardware-level channel switching mode eliminates the software calculation and allocation process, greatly improving the response speed of the protection logic and reducing the risk of equipment loss of control during a fault. The traditional transmission process of relay processing by the microcontroller is eliminated, and the protection command is directly transmitted by a dedicated high-speed line, shortening the signal transmission and calculation time and reducing the protection lag defect caused by system delay. By using a high-speed comparator to regulate the control pulse timing parameters, the output of the wave-blocking drive signal is ensured to be stable and reliable. It can quickly complete accurate protection actions under magnetic saturation fault conditions and maintain the consistency of the protection signal output, significantly improving the real-time performance and stability of the overcurrent protection of the energy storage converter and avoiding device damage caused by sudden large current disturbances.

[0026] One embodiment of the present invention, such as Figure 6 As shown, S5 includes: S51. The hard-wired wave-blocked drive signal data is sent into the automotive-grade digital isolation chip arranged in multiple parallel circuits to isolate the potential difference interference between the front and rear strong and weak circuits, avoid the problem of high voltage on the power side entering the control side and causing drive logic disorder, and synchronously compress the signal transmission delay. S52. After electrical isolation, the turn-off pulse signal is sent to the drive ports of all power devices in the upper and lower bridge arms of the energy storage converter to drive the internal channel of the silicon carbide power chip to turn off quickly. The timing delay and action status information of each device to complete the turn-off action are recorded and the nanosecond-level turn-off execution data of the SiC device are generated. S53. Retrieve the timing record of the entire disturbance action recorded by the nanosecond-level turn-off execution data of the SiC device, reverse the duration of the overcurrent disturbance, the peak current magnitude, the disturbance decay trend and complete the working condition process, and completely replicate the working condition curve of the entire process from disturbance intrusion to initial decay. S54. Retrieve the short-time high current withstand limit duration parameters of silicon carbide devices and insulated gate bipolar transistors pre-recorded inside the memory chip, combine them with the disturbance severity obtained from backtracking to divide the power reduction level into three levels, adapt the corresponding load reduction range according to the disturbance strength, and integrate the level control logic to generate graded load reduction linkage control data.

[0027] The working principle and effects of the above technical solution are as follows: By using a digital isolation chip to separate the strong and weak current circuits, high voltage potential is prevented from entering the control area, reducing the probability of high voltage surges disrupting the drive program and shortening the overall transmission time of the drive signal. Synchronously sending shutdown pulses to all bridge arm power devices enables synchronized shutdown of each device, completely preserving the timing information of each device's actions and improving the completeness of data retention throughout the entire fault process. Reverse reconstruction of the overcurrent disturbance's complete change curve based on the shutdown record allows for a comprehensive understanding of the fault's strength and duration, compensating for the inability to observe the internal operating conditions in real time during the fault occurrence phase. Multiple load reduction strategies are defined based on the overload capacity of the two types of power devices, and the power adjustment range is matched according to the severity of the fault. This not only fully protects the power devices from damage during severe overcurrent but also avoids excessive reduction in equipment output under minor disturbances, enhancing the converter's operational resilience under fault conditions and ensuring a smooth transition to subsequent grid connection and control processes.

[0028] One embodiment of the present invention, such as Figure 7 As shown, S54 includes: Read the set of power device withstand parameters stored in the memory chip, extract the short-time high current withstand time threshold of silicon carbide device and the short-time overload withstand time threshold of insulated gate bipolar transistor, and generate a dataset of device-differentiated withstand parameters. Retrieve all disturbance condition curve data from the completed operating condition backtracking, extract multiple core disturbance indicators such as peak disturbance current, duration, and current decay rate, and generate a quantitative dataset of overcurrent disturbance intensity. The overcurrent disturbance intensity quantization dataset is divided into gradient levels, and three disturbance levels are distinguished according to the strength of the disturbance impact: mild overload, moderate impact, and severe fault, generating disturbance level classification data. Match the power reduction range corresponding to different disturbance levels, configure progressively increasing power load reduction coefficients accordingly, form a one-to-one matching relationship between levels and control parameters, and generate graded load reduction parameter adaptation data. Integrate all graded load reduction parameter adaptation data and device tolerance protection logic, unify and standardize the linkage control rules for equipment power regulation, and generate graded load reduction linkage control data.

[0029] The working principle and effects of the above technical solution are as follows: By separately retrieving the overload tolerance parameters of the two types of power devices, the differences in the load-bearing capacity of the devices themselves are fully considered, reducing the adaptation deviation problem caused by the unified control standard. Multiple key indicators in the disturbance condition are extracted to complete the intensity quantification, no longer relying on a single current value to judge the degree of fault, significantly improving the comprehensiveness of fault severity assessment. Overcurrent impact is divided into three gradient levels, allowing for more detailed differentiation of fault scenarios and reducing control errors caused by general judgments. Corresponding power reduction coefficients are matched according to different levels; the stronger the disturbance, the greater the power reduction, balancing device protection and equipment operation requirements. After the entire set of control rules is unified and standardized, it can firmly safeguard the device safety boundary in severe short-circuit faults, preventing chip damage caused by long-term high-current impacts, while retaining most of the equipment's output power in minor overload conditions, enhancing the operational adaptability of the energy storage converter during fault stages, and making power control more delicate and reasonable.

[0030] One embodiment of the present invention, such as Figure 8 As shown, S6 includes: S61. Collect and analyze the real-time grid parameters of three-phase voltage amplitude and frequency fluctuation at the grid connection point, and adapt and calibrate the internal power reduction command of the graded load reduction linkage control data with the real-time grid operation parameters to avoid the grid voltage drastic fluctuation caused by sudden power drop. S62. Based on the adaptation calibration results, continuously output a gradually decreasing power-limiting flexible control level signal. The energy storage converter adjusts the active power output under this signal, maintaining the grid connection between the equipment and the AC grid throughout the process without performing emergency grid disconnection operations. Record the parameters of the entire grid-connected power-limited operation process to generate non-disconnection continuous operation data of the converter. S63: Continuously monitors the attenuation trend of AC grid voltage waveform and main circuit current waveform, continuously determines whether transient surge disturbance energy has completely dissipated, and triggers dual-channel sampling hardware reset logic after the circuit current returns to the daily load fluctuation range. S64. Gradually close the main channel current amplitude sampling output path, disconnect the auxiliary channel full-authority judgment working mode, restore the Hall main channel amplitude sampling as the main body of regular current monitoring, and switch the auxiliary channel to the background backup verification branch. S65. Record the complete timing sequence of channel reset, the sampling deviation values ​​before and after reset, and all contents of the current grid-connected operation status of the equipment. Archive and store them in a unified manner to complete the closed-loop protection process and generate normal overcurrent protection recovery data.

[0031] The working principle and effects of the above technical solution are as follows: By calibrating power control commands in conjunction with real-time grid operating parameters, the power reduction process is made smoother and more stable, reducing grid fluctuations caused by sudden power changes and avoiding grid oscillations caused by equipment protection actions. A flexible power-limiting operation mode is adopted to maintain the equipment's grid connection status, eliminating the need for emergency grid disconnection and shutdown, effectively improving the continuity of grid-connected operation of the energy storage converter and reducing equipment losses and power outages caused by frequent start-stop operations. Real-time monitoring of the grid and loop current status determines the dissipation of disturbances, ensuring that the sampling channel reset timing matches the actual operating conditions and reducing the probability of secondary false protection caused by premature reset. After fault recovery, it automatically switches back to the conventional main channel sampling mode, while the auxiliary channel retains backup verification functions, ensuring both stable accuracy of normal current monitoring and maintaining fault redundancy verification capabilities. The reset process and operating parameters are fully recorded, improving the closed-loop process of equipment protection and enhancing the long-term reliability and operating condition traceability of the converter.

[0032] One embodiment of the present invention provides an overcurrent protection system for an energy storage converter based on non-contact current sampling, the system comprising: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0033] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0034] Example 2 Based on Example 1, this example further refines the process for determining the Hall magnetic saturation state. By cross-verifying the change in the amplitude of the main channel current and the transient change in the auxiliary channel within the same time window, the reliability of the Hall magnetic saturation state is determined.

[0035] The method includes steps S1-S7.

[0036] S1. Acquire the full current analog signal of the main circuit of the energy storage converter in accordance with the method described in Example 1. After insulation isolation filtering, analog-to-digital conversion and feature decomposition, generate a dual-channel parallel raw sampling dataset including the main current amplitude sequence and the transient rate of change sequence.

[0037] S2. Perform main channel amplitude verification and time domain continuity scanning on the dual-channel parallel raw sampling dataset in accordance with the method described in Example 1, identify the step jump waveform segment and amplitude clamping waveform segment in the real-time current data of the main channel, and determine the abnormal segment that meets the initial judgment condition of Hall magnetic saturation in Example 1 as the Hall magnetic saturation state data to be verified.

[0038] S3. Perform auxiliary channel feature matching on the Hall magnetic saturation state marker data verified in step S7, retrieve the transient rate of change feature data of the differential channel output of the isolation operational amplifier within the corresponding time interval, and perform threshold comparison in the manner described in Example 1 to generate overcurrent trigger warning data.

[0039] S4. Based on the overcurrent trigger warning data, perform the channel state switching and hardware link direct transmission processing described in Example 1. During the period when the Hall magnetic saturation is confirmed, cut off the main channel amplitude output and send the protection trigger signal directly to the high-speed comparator to generate hard-wired wave-blocking drive signal data.

[0040] S5. The hard-wired waveguide drive signal data is isolated and transmitted and the bridge arm is turned off in accordance with the method described in Example 1. The graded load reduction linkage control data is generated by combining the overcurrent disturbance condition and the differential withstand time parameters of SiC and IGBT.

[0041] S6. Implement power limiting flexible regulation based on the graded load reduction linkage control data as described in Example 1, and complete the dual-channel state reset after the transient disturbance dissipates, restoring the main channel's normal sampling mode.

[0042] S7. Perform dynamic reliability verification of the main and auxiliary channels on the Hall magnetic saturation state data to be verified obtained in step S2; step S7 is a further processing procedure for the Hall magnetic saturation state determination in step S2, and is performed synchronously with the time domain scanning process of step S2, and is completed before the Hall magnetic saturation state marker data is called in step S3, including: S71. Within the same time-domain window corresponding to the Hall magnetic saturation state data to be verified, determine the change in current amplitude of the main channel, the change in current characteristic of the auxiliary channel, and the dispersion of the amplitude of the main channel window respectively. Based on the difference between the current changes characterized by the main channel and the auxiliary channel, and combined with the stability of the main channel amplitude within the window, determine the mismatch characterization value of the main and auxiliary channels to characterize the degree of inconsistency between the main and auxiliary channels when the main channel amplitude tends to be flattened while the actual current of the main circuit continues to change. S72. Continue calling step S71 to determine the change in the amplitude of the main channel current and the change in the characteristic current of the auxiliary channel within the same time domain window, and determine the consistency of the change direction of the sampling points of the effective transient change rate of the auxiliary channel. Based on the proportion of the characteristic current change of the auxiliary channel in the total change of the main and auxiliary channels and the consistency of the change direction, determine the transient change duration characteristic value to determine whether the transient change detected by the auxiliary channel belongs to the continuous current change of the main circuit or the positive and negative alternating disturbance formed by the high-frequency switching of the power device. S73. Correlate the main and auxiliary channel mismatch characterization values ​​and transient change duration characterization values ​​corresponding to the same time domain window to determine the Hall magnetic saturation confidence level. Based on the change state of the Hall magnetic saturation confidence level in the continuous time domain window, determine whether the Hall magnetic saturation state data to be verified has been converted into valid Hall magnetic saturation state marker data. Send the valid Hall magnetic saturation state marker data to steps S3 and S4 as the trigger basis for auxiliary channel feature calling and sampling channel switching.

[0043] Furthermore, in step S71, each amplitude-clamped waveform segment identified in step S2 is synchronously divided according to the time-domain sliding scan window adopted in step S2, so that the real-time current data of the main channel and the transient rate of change sequence of the auxiliary channel maintain the same time boundary.

[0044] For any time-domain window, read the real-time current values ​​of the main channel corresponding to the starting and ending sampling nodes of the window, and determine the change in the amplitude of the main channel current based on the difference between the two. Simultaneously retrieve the transient rate of change sampling data of the output of the differential channel of the isolation operational amplifier within the time-domain window, determine the degree of current change represented by the auxiliary channel within the window based on the transient rate of change of each sampling node and the corresponding sampling time interval, and obtain the current change represented by the auxiliary channel.

[0045] Simultaneously, the dispersion of the real-time current sampling values ​​of the main channel within the time-domain window is statistically analyzed to determine the amplitude dispersion of the main channel window. The amplitude dispersion level of the main channel during normal operation before the overcurrent disturbance occurs is used as the benchmark fluctuation scale. The lower the amplitude dispersion of the main channel window, the closer the output of the reinforced insulation Hall sensor is to a stable clamping state within the current window.

[0046] Based on this, the degree of difference between the change in the amplitude of the main channel current and the change in the characteristic current of the auxiliary channel is taken as the first evaluation factor, and the degree of reduction of the amplitude dispersion of the main channel window relative to the normal operation reference fluctuation scale is taken as the second evaluation factor. The two evaluation factors are correlated to obtain the main and auxiliary channel mismatch characterization value.

[0047] When both the main channel and the auxiliary channel represent a relatively stable current, even if the main channel forms a short-term current plateau, there will be no significant mismatch between the main and auxiliary channels. When the reinforced insulation Hall sensor enters a magnetic saturation state, the output of the main channel gradually flattens out, while the auxiliary channel can still represent the continuous change of the main circuit current, which increases the difference in change between the main and auxiliary channels. At the same time, the dispersion of the main channel window amplitude decreases, thereby increasing the mismatch representation value of the main and auxiliary channels accordingly.

[0048] Furthermore, in step S72, the same time-domain window and the corresponding main channel current amplitude change and auxiliary channel characterization current change determined in step S71 are used without re-dividing the analysis period, so as to ensure that the two characterization results correspond to the same current change process.

[0049] The transient rate of change sampling points of the differential channel output of the isolation operational amplifier within the time-domain window are screened. Small rate of change sampling points located within the normal operating background fluctuation range are removed, and effective transient rate of change sampling points exceeding the normal background fluctuation range are retained. Then, based on the overall current change direction represented by the auxiliary channel within the current window, the proportion of sampling points with the same overall change direction among the effective transient rate of change sampling points is statistically analyzed to determine the consistency of the change direction of the auxiliary channel.

[0050] When the large transient rate of change detected by the auxiliary channel mainly originates from the high-frequency switching of power devices or local oscillations, the effective transient rate of change sampling points are usually distributed alternately in positive and negative directions, and the consistency of their change directions is low. When the main circuit experiences a continuously increasing transient overcurrent, the effective transient rate of change sampling points within the same short time window have a more obvious directional continuity, and the consistency of their change directions increases accordingly.

[0051] Therefore, based on the proportion of the current change in the auxiliary channel relative to the overall current change in the main and auxiliary channels, and constrained by the consistency of the change direction in the auxiliary channel, the duration of transient change is determined. This process prevents large single transient rate peaks from being directly identified as continuous current changes in the main circuit, thereby reducing the impact of high-frequency switching disturbances on Hall magnetic saturation judgment.

[0052] Furthermore, in step S73, the mismatch characterization values ​​of the primary and secondary channels and the transient change persistence characterization values ​​corresponding to the same time-domain window are correlated and fused. During the correlation processing, the increase in Hall magnetic saturation confidence is limited when any of the characterization values ​​is at a low level, and the Hall magnetic saturation confidence of the corresponding time-domain window is increased only when both the degree of mismatch between the primary and secondary channels and the degree of persistence of change in the secondary channel reach a high level.

[0053] Specifically, when the main and auxiliary channel mismatch characterization value is high while the transient change characterization value is low, it indicates that although the main channel has amplitude clamping characteristics, the auxiliary channel lacks continuous directionality. The current anomaly is primarily judged as local oscillation or high-frequency switching disturbance. When the transient change characterization value is high while the main and auxiliary channel mismatch characterization value is low, it indicates that although there is continuous current change in the main circuit, the main channel has not yet shown a clear clamping mismatch state corresponding to magnetic saturation, and Hall magnetic saturation is not directly confirmed. Only when both characterization values ​​increase simultaneously can the credibility of Hall magnetic saturation be improved.

[0054] The Hall magnetic saturation confidence level is divided into a normal range, a pending confirmation range, and a confirmation range. When the Hall magnetic saturation confidence level is in the normal range, the main channel continues to operate according to the conventional sampling method of Example 1. When it enters the pending confirmation range, the acquisition status of the transient rate of change characteristic data of the auxiliary channel is maintained, and subsequent time domain windows are verified without immediately cutting off the main channel. When the Hall magnetic saturation confidence level enters the confirmation range and remains in the confirmation range for two consecutive time domain windows, the corresponding Hall magnetic saturation state data to be verified is converted into valid Hall magnetic saturation state marker data.

[0055] After valid Hall magnetic saturation state marker data is generated, the corresponding abnormal timing interval is sent to step S3, and step S3 retrieves the sampling content of the isolation operational amplifier differential channel corresponding to the abnormal timing interval; at the same time, the Hall magnetic saturation state marker data is sent to step S4 as the trigger basis for the sampling channel switching logic hardware unit to execute the main channel cut-off and auxiliary channel turn-on.

[0056] Through the above processing, this embodiment, based on the identification of step transitions and amplitude clamping, further utilizes existing sampling data from the main channel and auxiliary channel within the same time domain window for cross-verification. On one hand, the output distortion of the reinforced insulation Hall sensor is identified by the degree of amplitude clamping in the main channel and the degree of mismatch between the main and auxiliary channel changes; on the other hand, transient interference caused by high-frequency switching of power devices is continuously eliminated by the direction of transient changes in the auxiliary channel. Finally, the Hall magnetic saturation state is confirmed only when both types of characteristics are met simultaneously, thereby reducing the possibility of incorrect switching of sampling channels and false triggering of hard-wired blocking caused by normal current limiting platform and high-frequency switching disturbances, and improving the reliability of the overcurrent protection process.

Claims

1. An overcurrent protection method for energy storage converters based on non-contact current sampling, characterized in that, The method includes: S1. Acquire the full current analog signal of the main circuit of the energy storage converter, perform insulation isolation filtering on the full current analog signal to generate the original current sampling data of the main circuit; perform analog-to-digital conversion and feature decomposition on the original current sampling data of the main circuit to separate the main current amplitude sequence and transient rate of change sequence, and generate a dual-channel parallel original sampling dataset. S2. Perform main channel amplitude verification processing on the dual-channel parallel raw sampling dataset, extract the continuous current amplitude information output by the reinforced insulation Hall sensor, and generate real-time current data of the main channel; perform time-domain continuous scanning processing on the real-time current data of the main channel to capture the step jump and clamping characteristics of the current signal and generate Hall magnetic saturation state marker data. S3. Perform auxiliary channel feature matching processing on the Hall magnetic saturation state marker data, retrieve the bridge arm parasitic di / dt feature sequence output by the differential channel of the isolation operational amplifier, and generate transient current change rate feature data; perform threshold comparison calculation on the transient current change rate feature data, identify the initial time point of lightning surge and short circuit intrusion, and generate overcurrent trigger warning data. S4. Perform channel state switching processing on the overcurrent trigger warning data, cut off the main channel amplitude output during the Hall magnetic saturation trigger period, and generate no-delay protection trigger command data; perform hardware link direct transmission processing on the no-delay protection trigger command data, skip the MCU operation link and send it directly to the high-speed comparator to generate hard-wired wave blocking drive signal data. S5. The hard-wired waveguide drive signal data is isolated and transmitted. The drive signal is directly sent to the bridge arm drive unit through an automotive-grade digital isolator to generate nanosecond-level turn-off execution data for SiC devices. The nanosecond-level turn-off execution data of SiC devices is processed by operating condition backtracking to match the differentiated withstand time parameters of SiC and IGBT to generate graded load reduction linkage control data. S6. Perform grid status synchronization processing on the graded load reduction linkage control data, output flexible control signals for power-limited operation, and generate converter non-disconnection continuous operation data; perform dual-channel status reset processing on the converter non-disconnection continuous operation data, and automatically switch back to the main channel normal sampling mode after the transient surge condition ends, generating normal overcurrent protection recovery data.

2. The overcurrent protection method for energy storage converters based on non-contact current sampling according to claim 1, characterized in that, S1 includes: S11. A ring-shaped non-contact current acquisition coil is arranged on the DC bus of the energy storage converter and the outer wall area of ​​the three-phase AC main circuit. It continuously picks up the alternating magnetic field induced waveform generated by the movement of alternating charges inside the circuit, and fully records the full current simulation waveform formed by the superposition of steady-state load current and instantaneous impact current, and summarizes it to obtain the full current simulation signal flow with full coverage. S12. The acquired full-current analog signal is fed into a multi-stage RC passive filter array with common-mode rejection and insulation isolation module. After wave rectification and noise reduction, the normalized waveform information is output and the original current sampling data of the main circuit is generated. S13. The original current sampling data of the main circuit is sent to the high-precision successive approximation analog-to-digital converter chip to carry out continuous sampling and quantization operations, and the continuously changing analog electrical signal is converted into a discrete digital code value sequence point by point. Simultaneously, two independent data operation branches are divided and storage buffer space is allocated. S14. One branch stores the current magnitude corresponding to the sampling code value and continuously arranges it to form a time-progressive main current amplitude sequence. The other branch calculates the rate of change of the current difference between adjacent sampling nodes point by point and arranges it to form a transient rate of change sequence. The two types of time sequences are stored synchronously without interfering with each other. After integration and arrangement, a dual-channel parallel original sampling dataset is generated.

3. The overcurrent protection method for energy storage converters based on non-contact current sampling according to claim 2, characterized in that, S13 includes: The extended intent data is semantically normalized and field-purified, removing redundant natural language modifiers and extracting core semantic elements that fit the device's sampling and computation requirements to generate standardized intent normalized data. The standardized intent-organized data is processed by field mapping and instruction translation, transforming natural language semantic elements into digital computing instruction formats that can be recognized by hardware chips, and generating machine-recognizable primary structured instruction data. The basic structured instruction data is processed for instruction verification and format calibration, and the standardized format of instruction timing arrangement, data call port and operation execution logic is unified to generate compliant structured search instruction data; The compliant and structured search command data is decomposed into feature dimensions to extract the core feature parameters of sampling precision, operation mode and data processing dimension contained in the command, and generate command core feature dataset; The core feature dataset of the instruction is traversed and compared within the library. The functional adaptation attributes of various computational models in the preset model library are matched, and the model subject with the highest adaptation degree is selected to generate model type matching data.

4. The overcurrent protection method for energy storage converters based on non-contact current sampling according to claim 1, characterized in that, The S2 includes: S21. Retrieve all timing sampling contents of the amplitude branch inside the dual-channel parallel original sampling dataset, and combine them with the reference waveform of the same source current synchronously acquired by the enhanced insulation Hall sensor unit to perform point-by-point amplitude deviation verification and correct the accuracy of the current value corresponding to each timing node. S22. After multi-point deviation calibration, abnormal drift sampling points are removed, and stable and continuous current value records over a long time scale are retained. The records are arranged and integrated in chronological order, and the current value records are output in a continuous and uninterrupted manner. The main channel real-time current data is then generated. S23. Construct a time-domain scanning window with a fixed sliding duration to traverse the overall time sequence of the main channel real-time current data frame by frame. Within the window, compare the fluctuation range of the values ​​before and after each sampling point to continuously identify the step jump waveform segment where the value suddenly increases and the waveform segment where the value is capped for a long time and no longer increases. S24. Mark and divide all time periods during the scanning process that are determined to be waveform capped and stagnant, and whose amplitude no longer increases synchronously with the actual current of the circuit. Assign a unique timing identifier and status code to each abnormal waveform segment, archive and arrange them in a unified manner according to the timing sequence, and generate Hall magnetic saturation status marker data.

5. The overcurrent protection method for energy storage converters based on non-contact current sampling according to claim 1, characterized in that, The S3 includes: S31. Retrieve all saturation anomaly time intervals recorded by Hall magnetic saturation state marker data segment by segment, lock the start and end sampling node positions corresponding to each anomaly interval, and use the node interval as the retrieval boundary to retrieve the sampling content of the isolation operational amplifier differential channel pre-stored in the auxiliary channel. S32. Extract the time-series waveform of the parasitic current change rate generated at the moment of power-on of the bridge arm power circuit within the search boundary range, remove the smooth change component caused by slow load adjustment, retain only the short-term steep rate waveform segment, and splice and integrate the multiple extracted segments to generate transient current change rate characteristic data. S33. Two types of differentiated critical judgment thresholds are preset to correspond to the power grid lightning surge impact condition and the AC / DC side metallic short circuit intrusion condition, respectively. Each steep change peak in the transient current change rate characteristic data is compared with the two types of critical thresholds in turn. S34. Mark the first sampling timing node where each peak first breaks through the corresponding threshold. This node is the initial occurrence point of various large current disturbances entering the circuit. Pack the disturbance type, intrusion time, and initial change rate values ​​together to generate overcurrent trigger warning data.

6. The overcurrent protection method for energy storage converters based on non-contact current sampling according to claim 5, characterized in that, S32 includes: Based on the start and end sampling nodes of the locked abnormal interval, the continuous sampling waveform of the differential channel of the auxiliary channel isolation operational amplifier is precisely segmented, and the dynamic current change waveform data generated at the moment of power-on of the bridge arm power circuit is completely collected to generate full-domain time-period current change waveform segment data. The waveform segment data of current change over the entire time period is decomposed into layers and components. The slow current fluctuation component corresponding to the normal load adjustment action of the equipment is separated, and the smooth waveform unit with the fluctuation rate conforming to the normal working condition is selected to generate steady-state load fluctuation component data. Remove all smooth waveform content corresponding to steady-state load fluctuation component data, retain current waveform units with waveform change rate exceeding the normal operating range and instantaneous change characteristics, and generate short-time steep-change waveform segment data. The time sequence alignment and seamless splicing of multiple discrete short-time abruptly changing waveform segments are performed to unify the sampling accuracy and waveform amplitude benchmark of each segment and integrate them into a complete and continuous set of current abrupt change waveforms. The integrated set of current change waveforms is processed for data normalization and noise reduction, eliminating invalid clutter points caused by sampling noise, standardizing the waveform timing layout format, and generating transient current change rate characteristic data.

7. The overcurrent protection method for energy storage converters based on non-contact current sampling according to claim 1, characterized in that, The S4 includes: S41. Analyze the disturbance start and end time information recorded in the overcurrent trigger warning data, and synchronously send a switching level signal to the sampling channel switching logic hardware unit. Within the complete duration of the Hall device entering the magnetic saturation range, disconnect the data output path of the main current amplitude channel to the subsequent circuit. S42, the transmission link of transient rate of change characteristic data of the auxiliary channel is activated. All subsequent protection limit judgment calculations are carried out based on the current rate waveform collected by the differential channel, and the channel switching logic is integrated to generate protection trigger command data without delay. S43. Abandoning the conventional data transmission process of microcontroller cache read / write and floating-point operation relay, the delay-free protection trigger instruction data is directly transmitted to the input of the high-speed voltage comparator at the back end via dedicated high-speed traces on the inner layer of the PCB in the form of hardware level pulses. S44. The high-speed comparator receives the pulse level in real time and completes the level amplitude comparison. It outputs a turn-off control pulse waveform with a fixed level amplitude. After normalizing the pulse rising edge and falling edge timing parameters, it outputs them in a unified manner to generate hard-wired waveform drive signal data.

8. The overcurrent protection method for energy storage converter based on non-contact current sampling according to claim 1, characterized in that, The S5 includes: S51. The hard-wired wave-blocked drive signal data is sent into the automotive-grade digital isolation chip arranged in multiple parallel circuits to isolate the potential difference interference between the front and rear strong and weak circuits, avoid the problem of high voltage on the power side entering the control side and causing drive logic disorder, and synchronously compress the signal transmission delay. S52. After electrical isolation, the turn-off pulse signal is sent to the drive ports of all power devices in the upper and lower bridge arms of the energy storage converter to drive the internal channel of the silicon carbide power chip to turn off quickly. The timing delay and action status information of each device to complete the turn-off action are recorded and the nanosecond-level turn-off execution data of the SiC device are generated. S53. Retrieve the timing record of the entire disturbance action recorded by the nanosecond-level turn-off execution data of the SiC device, reverse the duration of the overcurrent disturbance, the peak current magnitude, the disturbance decay trend and complete the working condition process, and completely replicate the working condition curve of the entire process from disturbance intrusion to initial decay. S54. Retrieve the short-time high current withstand limit duration parameters of silicon carbide devices and insulated gate bipolar transistors pre-recorded inside the memory chip, combine them with the disturbance severity obtained from backtracking to divide the power reduction level into three levels, adapt the corresponding load reduction range according to the disturbance strength, and integrate the level control logic to generate graded load reduction linkage control data.

9. The overcurrent protection method for energy storage converters based on non-contact current sampling according to claim 1, characterized in that, The S6 includes: S61. Collect and analyze the real-time grid parameters of three-phase voltage amplitude and frequency fluctuation at the grid connection point, and adapt and calibrate the internal power reduction command of the graded load reduction linkage control data with the real-time grid operation parameters to avoid the grid voltage drastic fluctuation caused by sudden power drop. S62. Based on the adaptation calibration results, continuously output a gradually decreasing power-limiting flexible control level signal. The energy storage converter adjusts the active power output under this signal, maintaining the grid connection between the equipment and the AC grid throughout the process without performing emergency grid disconnection operations. Record the parameters of the entire grid-connected power-limited operation process to generate non-disconnection continuous operation data of the converter. S63: Continuously monitors the attenuation trend of AC grid voltage waveform and main circuit current waveform, continuously determines whether transient surge disturbance energy has completely dissipated, and triggers dual-channel sampling hardware reset logic after the circuit current returns to the daily load fluctuation range. S64. Gradually close the main channel current amplitude sampling output path, disconnect the auxiliary channel full-authority judgment working mode, restore the Hall main channel amplitude sampling as the main body of regular current monitoring, and switch the auxiliary channel to the background backup verification branch. S65. Record the complete timing sequence of channel reset, the sampling deviation values ​​before and after reset, and all contents of the current grid-connected operation status of the equipment. Archive and store them in a unified manner to complete the closed-loop protection process and generate normal overcurrent protection recovery data.

10. An overcurrent protection system for energy storage converters based on non-contact current sampling, characterized in that, The system includes: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 9.

Images