Control method and system of bidirectional charging and discharging circuit

Abstract

The invention relates to the technical field of circuit control, in particular to a control method and system of a bidirectional charging and discharging circuit. The method comprises the following steps: collecting current monitoring data and voltage monitoring data of a bidirectional charging and discharging circuit in real time; performing arc characteristic analysis on the current monitoring data and the voltage monitoring data to obtain arc characteristic data; analyzing a high-frequency harmonic component and a current ripple fluctuation amplitude of the bidirectional charging and discharging circuit based on the arc characteristic data; performing direct-current arc detection on the bidirectional charging and discharging circuit according to the high-frequency harmonic component and the current ripple fluctuation amplitude to generate a direct-current arc fault judgment result; and calculating the instantaneous current change rate of the current monitoring data, and carrying out short-circuit detection on the bidirectional charging and discharging circuit through the instantaneous current change rate. By monitoring current and voltage data in real time, arc and short circuit characteristics are accurately analyzed, relay control is optimized, an accurate control instruction is generated, and the safety and reliability of the two-way charging and discharging circuit are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of circuit control, and particularly to a control method and system for a bidirectional charge and discharge circuit. Background Art

[0002] The initial charge and discharge circuit design mainly focuses on unidirectional energy flow and is mainly applied to traditional battery management systems. With the increasing demands in fields such as electric vehicles (EVs), smart grids, and energy storage systems, bidirectional charge and discharge technology has become a research hotspot, aiming to achieve efficient energy conversion and management during the charging and discharging processes of batteries. With the rapid development of power electronics technology, embedded control systems, and communication technology, the control methods of bidirectional charge and discharge circuits have been gradually improved. By adopting advanced digital signal processors (DSPs) and field-programmable gate arrays (FPGAs) technology, the control accuracy and response speed have been greatly improved. At the same time, the application of advanced algorithms such as model predictive control (MPC) and adaptive control has significantly enhanced the stability and reliability of bidirectional circuits under high power and complex environments. However, currently, in many existing circuits, when a short-circuit fault occurs, the response time of the detection method is relatively long, and safety measures cannot be taken quickly. At the same time, the detection and control of overcurrent at the relay contacts are usually based on simple current limiting, without fully considering the operating characteristics of the relay and its overcurrent morphology, thus resulting in relatively low safety and reliability of the bidirectional charge and discharge circuit. Summary of the Invention

[0003] Based on this, it is necessary to provide a control method and system for a bidirectional charge and discharge circuit to solve at least one of the above technical problems.

[0004] To achieve the above object, a control method for a bidirectional charge and discharge circuit, the method includes the following steps: Step S1: Real-time collect the current monitoring data and voltage monitoring data of the bidirectional charge and discharge circuit; perform arc feature analysis on the current monitoring data and voltage monitoring data to obtain arc feature data; Step S2: Analyze the high-frequency harmonic components and current ripple fluctuation amplitude of the bidirectional charge and discharge circuit based on the arc feature data; perform DC arc detection on the bidirectional charge and discharge circuit according to the high-frequency harmonic components and current ripple fluctuation amplitude, and generate a DC arc fault discrimination result; Step S3: Calculate the instantaneous current change rate of the current monitoring data, and perform short-circuit detection on the bidirectional charge and discharge circuit through the instantaneous current change rate. When a current change continuously exceeding a preset threshold is detected, start time window cumulative analysis to generate instantaneous surge data and a continuous short-circuit fault discrimination result; Step S4: Generate a control instruction based on the DC arc discrimination result and the continuous short - circuit fault discrimination result; analyze the over - current morphology of the contacts of the relay in the bidirectional charge - discharge circuit based on the instantaneous surge data analysis, and optimize the on - time control of the control instruction according to the analysis result to perform the safety control optimization operation of the bidirectional charge - discharge circuit.

[0005] Through the extraction of arc characteristics and high - frequency harmonic analysis of the real - time collected current and voltage data, the present invention can effectively identify the high - frequency characteristics during the arc generation process, significantly improving the sensitivity and accuracy of DC arc detection. Combining with the dynamic monitoring mechanism of the instantaneous current change rate, it realizes the rapid early warning and judgment of continuous short - circuit faults, effectively preventing the risk of circuit damage or battery overheating. Using the time - window mechanism to cumulatively analyze surge events can not only identify sudden current fluctuations but also distinguish instantaneous surges from continuous short - circuits, which helps to improve the fault classification ability of the system. Based on the analysis of the over - current morphology of the relay contacts to optimize the on - time control strategy can extend the service life of the relay, reduce the misoperation rate of the contactor, and improve the safety and reliability of the entire bidirectional charge - discharge system. By fusing multiple discrimination results to generate control instructions and dynamically optimizing the control, a complete closed - loop from monitoring - judgment - control is constructed, effectively enhancing the robustness and adaptability of the power management system in a complex electrical environment. Therefore, the present invention improves the safety and reliability of the bidirectional charge - discharge circuit by real - time monitoring current and voltage data, accurately analyzing arc and short - circuit characteristics, optimizing relay control, and generating accurate control instructions.

[0006] Preferably, step S1 includes the following steps: Step S11: Real - time collect the current monitoring data and voltage monitoring data of the bidirectional charge - discharge circuit; Step S12: Align the time series of the current monitoring data and voltage monitoring data to generate time - series - aligned current - voltage monitoring data; perform differential operations on the time - series - aligned current - voltage monitoring data to generate current change rate data and voltage change rate data; Step S13: Perform mutation edge detection on the current change rate data and voltage change rate data to generate suspicious arc trigger point data; extract the time window of the suspicious arc trigger point data, and perform envelope analysis on the current - voltage monitoring data according to the time window to generate arc voltage waveform feature data; Step S14: Perform spectrum analysis on the arc voltage waveform feature data to generate arc frequency - domain feature data; perform key feature screening on the arc frequency - domain feature data to generate arc feature data.

[0007] The present invention performs time-series alignment and differential operation on the collected current and voltage monitoring data, avoiding misjudgment caused by asynchronous sampling clocks, improving the perception ability of minute changes in electrical signals, and laying a solid foundation for subsequent feature extraction. By processing the current and voltage change rate data through a mutation edge detection method, potential arc trigger points can be quickly located, enhancing the capture ability of instantaneous abnormal discharge behavior. Based on the suspected trigger points, a local time window is constructed and envelope analysis is performed to effectively extract the overall trend characteristics of the arc waveform, improving the modeling accuracy of the time-domain dynamics of arc behavior. By performing spectrum analysis on the envelope voltage waveform and screening key frequency-domain features, noise interference and redundant information in irrelevant frequency bands can be eliminated, further improving the efficiency and accuracy of the arc discrimination algorithm. By successively constructing multi-stage analysis chains such as time-domain difference, mutation edge detection, envelope modeling, spectrum extraction, and key feature screening, this method has good modularity and scalability, providing a high-quality data foundation for subsequent introduction of AI or machine learning models for intelligent arc identification.

[0008] Preferably, step S2 includes the following steps: Step S21: Perform spectrum decomposition processing on the arc feature data, extract the high-frequency harmonic components of the current signal, and generate high-frequency harmonic data; perform amplitude analysis on the high-frequency harmonic data to generate high-frequency harmonic amplitude data; Step S22: Perform time-domain waveform analysis on the current monitoring data, calculate the amplitude of the current ripple fluctuation, and generate current ripple fluctuation data; calculate the arc feature index of the circuit according to the high-frequency harmonic amplitude data and the current ripple fluctuation data to generate arc feature index data; Step S23: Use the arc feature index data to perform DC arc detection on the bidirectional charge and discharge circuit to generate a DC arc fault discrimination result.

[0009] Through spectral decomposition processing of the arc feature data, the present invention can accurately extract high-frequency harmonic components from complex current signals, effectively restore the high-frequency disturbance phenomenon accompanying the occurrence of the arc, and provide a high-resolution frequency-domain basis for arc identification. By means of time-domain waveform analysis, the ripple fluctuation amplitude in the current is accurately calculated, enabling the system to perceive minute unstable behaviors during the charge and discharge processes, thereby enhancing the sensitivity detection level for abnormal discharge conditions. By fusing the high-frequency harmonic amplitude and the current ripple fluctuation amplitude, an arc feature index is constructed, effectively avoiding the misjudgment problem caused by a single feature dimension, thus significantly enhancing the reliability and generalization ability of DC arc detection. By using the arc feature index to make real-time judgments on the circuit operating state, the traditional complex multi-feature joint classification process is simplified, making the DC arc fault detection process faster and more efficient, and suitable for deployment in a real-time operating power management system. The constructed arc feature index mechanism has good model compatibility, is applicable to bidirectional charge and discharge scenarios with different operating modes such as constant voltage and constant current, and also supports future integration and expansion with intelligent discrimination algorithms to achieve multi-scenario adaptive arc detection.

[0010] Preferably, step S23 includes the following steps: Step S231: When any of the following situations occurs in the arc feature index, it is determined as a DC arc fault and the first fault discrimination data is obtained: the arc feature index is greater than 5.0; the current fluctuation exceeds ±10% and lasts for more than 3 seconds; the arc occurrence frequency is greater than 50 times per minute; the discharge voltage is greater than 400V or less than 200V; Step S232: When the relationship between the arc feature index and the arc occurrence frequency meets the following conditions, the second fault discrimination data is generated: the arc feature index is greater than 6.0 and the arc occurrence frequency is greater than 60 times per minute; the fluctuation amplitude of the arc feature index is greater than ±20%, and the change amplitude of the current value exceeds ±15%; Step S233: When determining the DC arc fault of the bidirectional charge and discharge circuit, if the following situations occur, the third fault discrimination data is generated: the current is greater than 50A or the voltage is greater than 500V, and the duration exceeds 60 seconds; the current fluctuation is greater than ±20% and the voltage fluctuation is greater than ±15%, and these fluctuations last for more than 30 seconds; Step S234: Integrate the first fault discrimination data, the second fault discrimination data, and the third fault discrimination data, and finally generate the DC arc fault discrimination result.

[0011] The present invention sets multiple fault discrimination conditions (first, second, and third discrimination data), conducts joint judgment from multiple characteristic perspectives, forms a structured arc fault recognition system, and effectively avoids the risks of misjudgment and missed judgment. When the current or voltage reaches the limit value (such as current > 50A, voltage > 500V) and lasts for more than a certain time, the judgment mechanism can be triggered independently, enhancing the self-protection ability of the system under abnormal peak conditions. Through the coupled analysis of the "arc occurrence frequency" and the "arc characteristic index", potential arc hazards caused by short-term high-frequency discharge behavior can be accurately identified, improving the detection sensitivity to periodic and intensive arc behaviors. Incorporating dynamic parameters such as current fluctuation amplitude, voltage fluctuation amplitude, and duration into the discrimination logic makes the arc fault recognition more conform to the dynamic evolution law of the actual operating conditions, thereby enhancing the recognition accuracy of continuous arc events. By combining multiple discrimination rules and integrating them into the final fault judgment result, it has a modular and adjustable discrimination strategy framework, facilitating subsequent dynamic adjustment and threshold optimization according to specific circuit application scenarios.

[0012] Preferably, step S3 includes the following steps: Step S31: Perform numerical differentiation calculation on the current monitoring data to obtain the instantaneous current change rate data; perform smoothing filtering on the instantaneous current change rate data to generate smoothed instantaneous current change rate data; Step S32: Compare the smoothed instantaneous current change rate data with the short-circuit discrimination threshold. When the smoothed instantaneous current change rate data is greater than or equal to the short-circuit discrimination threshold, generate short-circuit trigger flag data; Step S33: Conduct a Boolean discrimination on the short-circuit trigger flag data. When the short-circuit trigger flag data is true, start time window cumulative analysis to perform fluctuation amplitude analysis on the current monitoring data under the short-circuit trigger flag data within the time window, and generate time window cumulative data; Step S34: Conduct a continuity analysis on the time window cumulative data. If the current change amplitude in the time window cumulative data is greater than 20A or the current change duration > 30s, generate a continuous short-circuit fault discrimination result; conduct a surge analysis on the bidirectional charge and discharge circuit based on the time window cumulative data to generate instantaneous surge data.

[0013] The present invention performs numerical differentiation processing on current monitoring data, quickly extracts the instantaneous current change rate, effectively identifies short - circuit precursor characteristics such as sudden current increase or decrease, and improves the real - time performance and response speed of fault warning. By combining with a smoothing filter algorithm to process the instantaneous current change rate, high - frequency noise interference is removed, effectively preventing false alarms or missed alarms caused by instantaneous jitters, and ensuring the reliability and accuracy of fault identification. By introducing a short - circuit trigger flag and Boolean discrimination logic, combined with time - window cumulative analysis, not only can abnormal pulses within a short time be judged, but also continuous short - circuit characteristics can be identified, constructing a fault judgment model with stronger time - series perception ability. By jointly judging the amplitude and duration of current changes within the time window, two typical fault states of instantaneous surge and continuous short - circuit can be effectively distinguished, improving the perception and control ability of extreme working states. Under the background of short - circuit trigger, a surge analysis process is introduced to extract instantaneous surge characteristic data, which can provide accurate input for subsequent relay over - current protection or control instruction adjustment, enhancing the overall safety guarantee performance of the system.

[0014] Preferably, the surge analysis of the bidirectional charge - discharge circuit based on the time - window cumulative data in step S34 includes: Performing current feature extraction on the smoothed instantaneous current change rate data to obtain current instantaneous change feature data, where the current feature extraction includes current change rate, fluctuation amplitude, and high - frequency components; Mapping the current instantaneous change feature data to a high - dimensional space through a radial basis function to construct a hyperplane, obtaining a surge hyperplane; inputting the time - window cumulative data into the surge hyperplane for time - shift analysis to generate surge time - shift data; Calculating the lateral occupancy ratio of the surge hyperplane according to the surge time - shift data to obtain surge lateral occupancy ratio data; calculating the instantaneous surge intensity of the current monitoring data under the short - circuit trigger flag data based on the surge lateral occupancy ratio data to obtain the instantaneous surge intensity at each time point; Aggregating the instantaneous surge intensity at each time point to generate instantaneous surge data.

[0015] The present invention significantly enhances the recognition ability of complex current fluctuation patterns by performing multi-feature extraction of the rate of change, fluctuation amplitude, and high-frequency components on the smooth instantaneous current change rate data, providing solid data support for subsequent surge analysis. The radial basis function (RBF) is used to map the current instantaneous change characteristics to a high-dimensional feature space, effectively amplifying the differences in non-linear surge patterns, making the surge events that were originally difficult to separate have good separability in the high-dimensional hyperplane. A surge hyperplane is constructed in the high-dimensional space, and time-shift analysis is performed by projecting the time-window cumulative data onto this hyperplane, accurately capturing the occurrence time and intensity evolution trajectory of the surge event, and improving the timing sensitivity of the surge response. Based on the lateral occupancy calculation mechanism of the surge hyperplane, the coverage range of current anomalies can be judged from the spatial structure, effectively distinguishing weak surges, strong surges, and non-surge phenomena, and improving the accuracy of anomaly recognition. The introduction of the instantaneous surge intensity calculation mechanism supports the generation of a quantitative surge intensity index at each time point, and then the overall trend characteristics are obtained through data aggregation, helping with the subsequent adaptive adjustment of the control logic and early warning. This method can not only perform high-resolution surge recognition in the short-circuit trigger scenario, but also combine the short-circuit trigger flag to achieve high-accuracy abnormal event modeling and risk level classification, assisting the intelligent decision-making control of the bidirectional charge and discharge circuit system.

[0016] Preferably, generating a control instruction according to the DC arc discrimination result and the continuous short-circuit fault discrimination result in step S4 includes: Performing high-order analysis on the signal characteristics in the DC arc discrimination result to generate arc duration and waveform change data; Performing real-time pattern recognition on the current waveform in the short-circuit fault discrimination result to generate short-circuit pattern category data; Performing joint spatio-temporal analysis on the arc duration and waveform change data and the short-circuit pattern category data to generate fault collaborative impact data; Performing time-domain and frequency-domain fusion on the fault collaborative impact data to generate comprehensive fault severity data; Redundantly verifying the DC arc discrimination result and the continuous short-circuit fault discrimination result through the comprehensive fault severity data and generating a control instruction to obtain the control instruction.

[0017] By performing high-order analysis on the signal characteristics in the DC arc discrimination results, the present invention can obtain the arc duration and waveform evolution information, realize the dynamic tracking and structural understanding of the arc behavior, and enhance the ability to identify the arc risk sources. Introducing real-time pattern recognition of the short-circuit fault discrimination results can classify the short-circuit behavior into different patterns (such as burst type, continuous type, periodic type, etc.), enhance the system's ability to analyze the types of short-circuit behavior, and provide a basis for adapting control strategies. Utilizing the cross characteristics of arc and short-circuit data in the time and space dimensions to generate fault collaborative influence data can effectively reveal the arc-short circuit coupling relationship, thereby judging the fault propagation path and risk diffusion trend. By performing fusion analysis on the collaborative influence data at the time domain and frequency domain levels, not only the accuracy of depicting the fault severity is improved, but also the adaptability to multi-source abnormal signals under complex waveform backgrounds is enhanced. Introducing a redundant verification mechanism for the discrimination results before generating control instructions can avoid misoperation caused by single judgment errors, and effectively improve the accuracy of triggering control instructions and the system stability. Combining the fault severity level assessment results can automatically generate control instructions with different emergency levels, such as current limiting, power off, system reconstruction, etc., thereby realizing the differential and adaptive processing of response actions and ensuring the safe operation of the system in sudden arc or short-circuit events.

[0018] Preferably, in step S4, the morphology of the overcurrent of the relay contacts in the bidirectional charge and discharge circuit based on the instantaneous surge data analysis includes: Capturing high-frequency surge events from the current monitoring data of the bidirectional charge and discharge circuit based on the instantaneous surge data to generate instantaneous surge waveform data; Performing FIR band-pass filtering processing on the instantaneous surge waveform data to generate denoised surge feature data; Extracting the time-domain features of the denoised surge feature data to obtain surge parameter quantization data; Analyzing the contact current distribution of the relay in the bidirectional charge and discharge circuit using the surge parameter quantization data to generate contact current distribution data; Simulating the local contact current propagation of the relay based on the contact current distribution data to generate relay contact current exchange simulation data; Analyzing the change in the electronic quantum state of the relay contact current exchange simulation data to generate electronic quantum state change data; Predicting the contact wear of the relay contacts through the electronic quantum state change data to generate the analysis result of the overcurrent morphology of the contacts.

[0019] Through high-frequency surge event capture and FIR band-pass filtering denoising of instantaneous surge data, the present invention can accurately identify and isolate important surge event characteristics from current monitoring data, ensuring a clear presentation of arc and abnormal current waveforms, and providing high-quality data for subsequent analysis. By extracting time-domain characteristics from the denoised surge feature data and quantifying surge parameters, key current parameters such as surge amplitude, duration, and transient waveform can be more precisely described, enhancing the system's time-sensitive response ability to surge events and improving the accuracy of fault warning. Through detailed analysis of the current distribution of relay contacts using surge parameter quantification data, not only can the current distribution be understood, but also its potential impact on the life and performance of the relay can be predicted. This provides a scientific basis for the design optimization and practical use of relays. By using local contact current propagation simulation to dynamically model the relay, the behavior characteristics of the contacts under different current loads can be understood in advance, potential risks of contact overcurrent and poor contact can be identified, and damage caused by current fluctuations or overloads to the equipment can be prevented. By analyzing the electronic quantum state changes in the current exchange of relay contacts, the microscopic dynamics of electronic behavior can be deeply explored, and the fine change characteristics when current passes through the contacts can be revealed, which helps to further improve the prediction ability of arc and overcurrent phenomena. Through the analysis of electronic quantum state change data, the wear degree of relay contacts can be effectively evaluated, potential faults of the relay can be warned in advance, system failures caused by excessive contact wear can be avoided, and the reliability and safety of the charge-discharge circuit can be improved. Based on the above precise analysis, precise control and prediction of the relay can be carried out, risks such as overcurrent and overload can be reduced, thereby improving the stability and safety of the bidirectional charge-discharge circuit under complex working conditions and ensuring the long-term stable operation of the equipment.

[0020] Preferably, the optimization of the conduction time control of the control instruction according to the analysis result in step S4 includes: Accumulate the overcurrent duration of the analysis result of the contact overcurrent morphology to generate the total contact overcurrent duration data; Perform interval comparison processing on the total contact overcurrent duration data to generate contact overcurrent trend data; Extract the conduction signal limit of the contact overcurrent trend data to generate the contact conduction signal allowable duration data, where the contact conduction signal allowable duration data includes the maximum single duration and the minimum interval time; Optimize the continuous conduction time of the control instruction according to the maximum single duration and the minimum interval time to generate the conduction time limit parameter data; Use the conduction time limit parameter data to adjust the conduction time control of the control instruction to perform the safety control optimization operation of the bidirectional charge-discharge circuit.

[0021] By performing cumulative processing on the overcurrent duration based on the analysis results of the overcurrent morphology of the contacts, the present invention can accurately quantify the overcurrent duration of the relay contacts, monitor the load conditions of the contacts in real time, ensure timely intervention before exceeding the safety threshold, thereby preventing damage or aging of the relay contacts. By performing interval comparison processing on the total overcurrent duration data of the contacts, overcurrent trend data of the contacts is generated, providing a key basis for subsequent optimization control instructions. By identifying the overcurrent trend of the contacts, future overcurrent risks can be predicted in advance to ensure the long-term stable operation of the charge and discharge circuit. By extracting the conduction signal limit of the overcurrent trend data of the contacts, the allowable duration data of the contact conduction signal is generated, ensuring that the maximum single duration and the minimum interval time of the conduction signal can be reasonably limited during the execution of the control instructions, avoiding circuit overload or relay contact damage caused by too long conduction duration or too short interval time. Based on the maximum single duration and the minimum interval time, the circuit continuous conduction time is optimized for the control instructions to further optimize the circuit operating state. Under variable load conditions, the conduction time can be dynamically adjusted to ensure the smooth operation of the circuit under different working conditions and reduce unnecessary energy consumption. By controlling and adjusting the conduction time limit parameters, the failure of the relay contacts caused by overload can be effectively avoided, improving the safety and reliability of the bidirectional charge and discharge circuit. While ensuring the charge and discharge efficiency, the service life of the equipment is extended, and the maintenance cost and downtime are reduced. The conduction time optimization of the control instructions not only meets the requirements of different current loads, but also can automatically adjust for different types of faults or working conditions, providing a more flexible and efficient adaptive control ability for the system. This flexibility can maintain the best operating state of the circuit in various complex situations. Through intelligent analysis and control parameter optimization, the automation and intelligentization process of the bidirectional charge and discharge circuit is promoted, providing higher precision and efficiency for the real-time monitoring and regulation technology of the next-generation power system, and promoting the innovation and application of power equipment intelligent control technology.

[0022] In this specification, a control system for a bidirectional charge and discharge circuit is provided for implementing the control method of the bidirectional charge and discharge circuit as described above. The control system for the bidirectional charge and discharge circuit includes: An arc feature extraction module, configured to collect current monitoring data and voltage monitoring data of the bidirectional charge and discharge circuit in real time; perform arc feature analysis on the current monitoring data and the voltage monitoring data to obtain arc feature data; An arc fault discrimination module, configured to analyze the high-frequency harmonic components and the current ripple fluctuation amplitude of the bidirectional charge and discharge circuit based on the arc feature data; perform DC arc detection on the bidirectional charge and discharge circuit according to the high-frequency harmonic components and the current ripple fluctuation amplitude to generate a DC arc fault discrimination result; A short - circuit discrimination module, which is used to calculate the instantaneous current change rate of current monitoring data and perform short - circuit detection on the bidirectional charge - discharge circuit through the instantaneous current change rate. When detecting a current change that continuously exceeds a preset threshold, it starts time - window cumulative analysis to generate instantaneous surge data and a continuous short - circuit fault discrimination result; A contact conduction control module, which is used to generate a control instruction according to the DC arc discrimination result and the continuous short - circuit fault discrimination result; analyze the over - current morphology of the relay contacts in the bidirectional charge - discharge circuit based on the instantaneous surge data, and optimize the conduction time control of the control instruction according to the analysis result to perform the safety control optimization operation of the bidirectional charge - discharge circuit.

[0023] The beneficial effects of the present invention are as follows: Through the real - time current and voltage monitoring of the arc feature extraction module, it can accurately capture current fluctuations and voltage changes, providing a reliable data basis for subsequent arc fault discrimination and short - circuit detection. Real - time monitoring can detect potential arc problems and circuit faults at an early stage, significantly improving the safety of circuit operation. The arc fault discrimination module can effectively detect and identify DC arc faults by analyzing the high - frequency harmonic components and current ripple fluctuation amplitude in the arc feature data. This analysis method not only improves the accuracy of fault discrimination but also can give an early warning at an early stage to avoid the further development of arc faults, reducing system damage and downtime. The short - circuit discrimination module performs intelligent short - circuit detection by calculating the instantaneous current change rate and combining with a preset threshold. Once an abnormal current fluctuation is detected, this module can quickly start time - window cumulative analysis and generate instantaneous surge data, effectively identifying continuous short - circuit faults. This mechanism can respond to current anomalies in a timely manner, preventing safety risks such as equipment damage or fire caused by short - circuit faults. The contact conduction control module can not only generate a control instruction based on the arc and short - circuit fault discrimination results but also analyze the over - current morphology of the relay contacts using the instantaneous surge data, accurately predicting and regulating the contact current. This function ensures that the relay works under optimal conditions, avoiding contact wear or circuit failure caused by over - current or overload. After analyzing the over - current morphology of the relay contacts, the module can intelligently optimize the conduction time, limiting the maximum conduction duration and minimum interval time of the circuit. This optimization can achieve safety control under variable current loads, reducing unnecessary energy losses and extending the service life of electrical components. Through the linkage of multiple modules, such as the cooperation of arc fault discrimination and short - circuit detection, the system can respond to electrical faults in real time, preventing potential dangers from escalating. In addition, through the intelligent optimization of circuit control instructions, safety accidents caused by equipment overload are avoided, significantly improving the safety and reliability of the system. Therefore, the present invention improves the safety and reliability of the bidirectional charge - discharge circuit by real - time monitoring current and voltage data, accurately analyzing arc and short - circuit characteristics, optimizing relay control, and generating accurate control instructions. Description of the Drawings

[0024] Figure 1 It is a schematic diagram of the step flow of a control method for a bidirectional charge and discharge circuit; Figure 2 It is Figure 1 a schematic diagram of the detailed implementation steps of step S1 in Figure 3 It is Figure 1 a schematic diagram of the detailed implementation steps of step S2 in The realization, functional features and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. Specific embodiments

[0025] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the protection scope of the present invention.

[0026] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0027] It should be understood that although the terms "first", "second", etc. may be used here to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit can be called the second unit, and similarly the second unit can be called the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0028] To achieve the above object, please refer to Figures 1 to 3 , a control method for a bidirectional charge and discharge circuit, the method comprising the following steps: Step S1: Real-time collect the current monitoring data and voltage monitoring data of the bidirectional charge and discharge circuit; perform arc feature analysis on the current monitoring data and voltage monitoring data to obtain arc feature data; Step S2: Analyze the high-frequency harmonic components and the current ripple fluctuation amplitude of the bidirectional charging and discharging circuit based on the arc feature data; perform DC arc detection on the bidirectional charging and discharging circuit according to the high-frequency harmonic components and the current ripple fluctuation amplitude, and generate a DC arc fault discrimination result; Step S3: Calculate the instantaneous current change rate of the current monitoring data, and perform short-circuit detection on the bidirectional charging and discharging circuit through the instantaneous current change rate. When a current change continuously exceeding a preset threshold is detected, start time-window cumulative analysis to generate instantaneous surge data and a continuous short-circuit fault discrimination result; Step S4: Generate a control instruction according to the DC arc discrimination result and the continuous short-circuit fault discrimination result; analyze the overcurrent morphology of the relay contacts in the bidirectional charging and discharging circuit based on the instantaneous surge data, and optimize the conduction time control of the control instruction according to the analysis result to perform the safety control optimization operation of the bidirectional charging and discharging circuit.

[0029] Through arc feature extraction and high-frequency harmonic analysis of the real-time collected current and voltage data, the present invention can effectively identify the high-frequency features during the arc generation process, significantly improving the sensitivity and accuracy of DC arc detection. Combined with the dynamic monitoring mechanism of the instantaneous current change rate, it realizes the rapid early warning and judgment of continuous short-circuit faults, effectively preventing the risks of circuit damage or battery overheating. Using the time-window mechanism to perform cumulative analysis on surge events can not only identify sudden current fluctuations but also be used to distinguish instantaneous surges from continuous short circuits, helping to improve the fault classification ability of the system. Optimizing the conduction time control strategy based on the overcurrent morphology analysis of the relay contacts can extend the service life of the relay, reduce the misoperation rate of the contactor, and improve the safety and reliability of the entire bidirectional charging and discharging system. By fusing multiple discrimination results to generate control instructions and performing dynamic optimization control on them, a complete closed-loop from monitoring - judgment - control is constructed, effectively enhancing the robustness and adaptability of the power management system in a complex electrical environment. Therefore, the present invention improves the safety and reliability of the bidirectional charging and discharging circuit by real-time monitoring current and voltage data, accurately analyzing arc and short-circuit characteristics, optimizing relay control, and generating accurate control instructions.

[0030] In the embodiment of the present invention, refer to Figure 1 As shown, it is a schematic diagram of the step flow of a control method for a bidirectional charging and discharging circuit of the present invention. In this example, the control method for a bidirectional charging and discharging circuit includes the following steps: Step S1: Real-time collect the current monitoring data and voltage monitoring data of the bidirectional charging and discharging circuit; perform arc feature analysis on the current monitoring data and voltage monitoring data to obtain arc feature data; Step S2: Analyze the high-frequency harmonic components and the current ripple fluctuation amplitude of the bidirectional charging and discharging circuit based on the arc feature data; perform DC arc detection on the bidirectional charging and discharging circuit according to the high-frequency harmonic components and the current ripple fluctuation amplitude, and generate a DC arc fault discrimination result; Step S3: Calculate the instantaneous current change rate of the current monitoring data, and perform short-circuit detection on the bidirectional charging and discharging circuit through the instantaneous current change rate. When a current change continuously exceeding a preset threshold is detected, start time window cumulative analysis to generate instantaneous surge data and a continuous short-circuit fault discrimination result; Step S4: Generate a control instruction according to the DC arc discrimination result and the continuous short-circuit fault discrimination result; analyze the overcurrent morphology of the relay contacts in the bidirectional charging and discharging circuit based on the instantaneous surge data, and optimize the conduction time control of the control instruction according to the analysis result to perform the safety control optimization operation of the bidirectional charging and discharging circuit.

[0031] In the embodiment of the present invention, high-precision Hall current sensors and differential voltage sampling circuits are deployed at the positive and negative electrodes of the bidirectional charging and discharging circuit to synchronously collect real-time data at a frequency of 10 kHz, and signal noise reduction preprocessing is performed using a moving average filter and a hardware RC low-pass filter; subsequently, the wavelet packet decomposition algorithm is used to extract the energy characteristics of the 3-10 kHz frequency band, and an arc feature database is formed by combining the analysis of the voltage mutation rate. In the fault detection link, the proportion of high-frequency harmonic components is calculated by three groups of FIR digital filter banks of 20 kHz / 50 kHz / 100 kHz, and the dynamic threshold method is used synchronously to monitor the current ripple fluctuation amplitude. When the harmonic continuously exceeds 5%, the ripple exceeds 20% of the rated value, and the voltage mutation rate exceeds 10 times per second, a three-layer decision tree model is triggered to determine the DC arc fault; at the same time, the three-point central difference method is used to calculate the current change rate in real time, and a hardware comparator is configured to start double-time window cumulative analysis when di / dt>500 A / ms (the short window exceeds the limit 8 times in 10 ms and the long window is continuously triggered in 100 ms), and the three elements of the surge current are extracted by combining the 100 kHz fault recording wave to accurately discriminate the short-circuit fault. Finally, the state machine decision system integrates the arc and short-circuit detection results, preferentially executes the 5 ms-level arc rapid breaking, and dynamically optimizes the breaking time sequence based on the contact ablation model established by HIL simulation: the pre-arc extinguishing control or the hardware direct cut protection is enabled based on the integral value of the surge current in stages. After the fault is eliminated, the integrity of the circuit is verified through three-level self-checks (insulation resistance measurement, small current conduction test, impedance spectrum analysis). The entire system uses the STM32H7+FPGA heterogeneous architecture to achieve real-time processing and hardware redundancy, and combines Model-Based Design to generate code and back-to-back tests to ensure the operation reliability.

[0032] As an example of the present invention, refer to Figure 2 As shown, in this example, step S1 includes: Step S11: Collect the current monitoring data and voltage monitoring data of the bidirectional charging and discharging circuit in real time; Step S12: Align the current monitoring data and voltage monitoring data in time series to generate time-series aligned current and voltage monitoring data; perform differential operations on the time-series aligned current and voltage monitoring data to generate current change rate data and voltage change rate data; Step S13: Detect the mutation edges of the current change rate data and voltage change rate data to generate suspicious arc trigger point data; extract the time window of the suspicious arc trigger point data, and perform envelope analysis on the current and voltage monitoring data according to the time window to generate arc voltage waveform feature data; Step S14: Perform spectrum analysis on the arc voltage waveform feature data to generate arc frequency domain feature data; perform key feature screening on the arc frequency domain feature data to generate arc feature data.

[0033] In the embodiments of the present invention, a highly responsive Hall current sensor (such as LEM LA55-P) and an isolated differential voltage probe (such as TI ISO224) are respectively deployed at the positive and negative poles of the bidirectional charge and discharge circuit. The multi-channel synchronous sampling ADC module (ADI AD7606C-16) synchronously captures current and voltage signals at a sampling rate of 20 kHz to ensure that the clocks of the two sets of data are of the same source (the FPGA provides hardware-level timestamp alignment). The current signal is passed through a second-order Butterworth low-pass filter (cutoff frequency 15 kHz) to eliminate switching noise, and the voltage signal suppresses ground loop interference through a common-mode choke. The hardware FIFO buffer of the FPGA is used to interpolate and resample the current and voltage data to compensate for the sensor transmission delay (calibrated within a synchronous error of ±1 μs), generating a data stream with strictly aligned time. The hardware-accelerated three-point central difference method (sliding window: n-1, n, n+1) is used to calculate the current change rate (di / dt) and voltage change rate (dv / dt), where Δt = 50 μs, and the differential results are stored as 16-bit fixed-point numbers to reduce calculation latency. A dual-threshold comparator is designed: when the instantaneous value of the current change rate > 200 A / ms or the absolute value of the voltage change rate > 30% of the nominal value, it is marked as a suspicious point, and combined with the sliding window variance detection (window width 10 ms) to exclude occasional interference, generating a trigger event queue with timestamps. For the original data segment 50 ms before and after the trigger point, a full-wave rectifier circuit (such as using OPA2182 to construct a precision rectifier) combined with a low-pass filter with a cutoff frequency of 2 kHz is used to extract the current / voltage envelope, and the envelope peak value, rise time (10%-90%), and duration are calculated as waveform features. The FFT acceleration core (FPGA built-in IP core) is used to perform a 4096-point FFT operation on the arc waveform segment, with a frequency resolution of 4.88 Hz, focusing on analyzing the frequency band of 0.5-20 kHz; three key features are extracted through the energy ratio algorithm: the high-frequency resonance peak (the frequency of the maximum amplitude point in the 5-10 kHz interval), the fundamental frequency harmonic distortion rate (the ratio of the amplitude of the third harmonic to the fundamental wave), and the spectral flatness (the energy variance in the 10-20 kHz frequency band). Based on the historical fault data set, a random forest model is trained to select the 5 frequency domain features with the highest discrimination (such as the energy ratio at 8.2 kHz and the spectral steepness slope at 15 kHz), and real-time feature mapping is achieved through a hardware look-up table (LUT), and finally a compressed arc feature vector is output.

[0034] As an example of the present invention, refer to Figure 3 As shown, in this example, step S2 includes: Step S21: Perform spectral decomposition processing on the arc feature data, extract the high-frequency harmonic components of the current signal, generate high-frequency harmonic data; perform amplitude analysis on the high-frequency harmonic data to generate high-frequency harmonic amplitude data; Step S22: Perform time-domain waveform analysis on the current monitoring data, calculate the current ripple fluctuation amplitude, and generate current ripple fluctuation data; calculate the arc characteristic index of the circuit according to the high-frequency harmonic amplitude data and the current ripple fluctuation data, and generate arc characteristic index data; Step S23: Use the arc characteristic index data to perform DC arc detection on the bidirectional charge and discharge circuit, and generate a DC arc fault discrimination result.

[0035] In the embodiments of the present invention, by using a hardened discrete wavelet transform (DWT) module (selecting the db6 wavelet basis) in an FPGA, the arc feature data is decomposed into 6 layers, and the high-frequency harmonic components in the frequency band of 5 - 20 kHz (corresponding to the 4th - 5th layer detail coefficients) are extracted. The full-wave rectification is performed on the decomposed high-frequency sub-band signals, and the root mean square (RMS) values of each frequency band are calculated through sliding window integration (window width 10 ms), and the results are normalized to a percentage of the fundamental wave amplitude (e.g., the proportion of the 5 kHz harmonic = high-frequency RMS / 50 Hz fundamental wave RMS × 100%). A white noise test signal is injected before harmonic calculation to dynamically calibrate the background noise threshold of each frequency band, and only the effective harmonic data exceeding 3 times the standard deviation of the noise baseline is retained. An IIR digital filter with a cut-off frequency of 1 kHz is applied to the original current signal to filter out high-frequency interference, and then a dynamic baseline tracking algorithm is adopted: the dynamic baseline current value is calculated based on a moving average (window length 100 ms), the absolute value of the ripple is obtained through full-wave rectification, the peak-to-peak value (Max - Min) and the RMS value are statistically calculated within each 10 ms window, and a weighted exponential formula is constructed: arc feature index = 0.6 × (proportion of the 5 kHz harmonic / 10%) + 0.4 × (peak-to-peak value of the ripple / 20% of the rated current). When the index > 1.0, a primary alarm is triggered, and when > 1.5, it is determined as a high-risk arc. A two-level determination mechanism is designed: if the feature index is > 1.5 at 3 consecutive sampling points (interval 0.1 ms), the breaking protection is immediately started. When the proportion of the time duration exceeding the threshold (> 1.0) within a 200 ms time window is > 60%, and accompanied by a voltage drop > 15% of the nominal value, a confirmed fault is output. In the scenario of load mutation (such as charge-discharge switching), a pre-marking mechanism is enabled: the change of the control signal is monitored 10 ms in advance, the determination threshold is temporarily increased to 1.8, and the current polarity inversion is monitored in real time through a hardware comparator to exclude the false harmonics caused by the reverse recovery current. A dedicated data path is allocated inside the FPGA to implement the parallel pipelining processing of harmonic decomposition (DWT) and ripple calculation (IIR filtering + sliding statistics), ensuring that the end-to-end delay from data input to fault discrimination < 500 μs. A CAN bus communication port is reserved to support the host computer to inject an analog arc waveform (such as the IEC 62619 standard test template) and adjust the harmonic weight coefficient and threshold parameter online. After the arc fault is determined, a high-speed storage unit (SRAM cache) is triggered to record the complete data stream of 50 ms before and after the fault, including the original current / voltage, harmonic components, and feature index curve, for offline backtracking analysis. A hardware protection circuit independent of the main control MCU is set up. When the feature index > 2.0 or the current mutation rate > 1000 A / ms, the main circuit is directly forced to be cut off by driving the MOSFET through the CPLD, and the response time < 20 μs.

[0036] Preferably, step S23 includes the following steps: Step S231: When any of the following conditions occurs for the arc feature index, it is determined as a DC arc fault and the first fault discrimination data is obtained: the arc feature index is greater than 5.0; the current fluctuation exceeds ±10% and lasts for more than 3 seconds; the arc occurrence frequency is greater than 50 times per minute; the discharge voltage is greater than 400V or less than 200V; Step S232: When the relationship between the arc feature index and the arc occurrence frequency meets the following conditions, the second fault discrimination data is generated: the arc feature index is greater than 6.0 and the arc occurrence frequency is greater than 60 times per minute; the fluctuation range of the arc feature index is greater than ±20%, and the change range of the current value exceeds ±15%; Step S233: When determining the DC arc fault of the bidirectional charge and discharge circuit, if the following situation occurs, the third fault discrimination data is generated: the current is greater than 50A or the voltage is greater than 500V, and the duration lasts for more than 60 seconds; the current fluctuation is greater than ±20% and the voltage fluctuation is greater than ±15%, and these fluctuations last for more than 30 seconds; Step S234: Integrate the first fault discrimination data, the second fault discrimination data, and the third fault discrimination data, and finally generate the DC arc fault discrimination result.

[0037] In the embodiments of the present invention, six groups of independent hardware comparators (such as ADcmp601) are deployed to monitor key parameters in real time and in parallel: Comparator 1: Arc feature index > 5.0 → Trigger a red LED alarm and latch the flag bit; Comparator 2: Absolute value of current fluctuation > 10% (output by the RMS calculation module) + 3-second timer (32-bit counter built into the FPGA) → Trigger when timing out; Comparator 3: Arc frequency counter (based on the rising-edge detection circuit) > 50 times / minute → Trigger when the pulse accumulator exceeds the limit; Comparator 4: Voltage monitoring channel > 400V or < 200V → Trigger by threshold comparison; Comparator 5: Arc feature index > 6.0 + Arc frequency > 60 times / minute → Trigger by AND logic; Comparator 6: Arc index fluctuation > ±20% (calculated by the sliding window standard deviation) + Current change > ±15% → Trigger when the window variance exceeds the limit. For the long-time determination conditions of S233 (such as 60 seconds / 30 seconds): Use a double-buffered circular queue to store current / voltage fluctuation data, calculate the volatility ((Max - Min) / Avg) every 1 second, and configure a hardware watchdog timer (such as MAX6818) to monitor the duration: When the volatility exceeds the limit continuously for 30 seconds, trigger the timer overflow interrupt. Build a three-level fault level: Level 1 (Emergency disconnection): Voltage in S231 > 500V or current > 50A → Trigger the hardware direct cut circuit (<100 μs response); Level 2 (Confirmed disconnection): Any one of the conditions in S231 / S232 is satisfied + Not restored for 200 ms → The MCU drives the relay to disconnect; Level 3 (Early warning and load reduction): Only the condition of S233 is satisfied → Reduce the charge and discharge power to 30% and report to the management system. The data integration uses a hardware OR gate array (SN74LVC1G32), and any fault signal triggers a global interrupt. At the moment of switching the charge and discharge mode (obtain the control instruction through the CAN bus), automatically increase the arc feature index threshold to 7.0 and restore it after 500 ms. When the ambient temperature > 85°C (obtained through the I2C temperature sensor), the current / voltage fluctuation threshold is relaxed to ±25%. Configure 4MB SRAM as the waveform memory, and automatically save the following data when a fault is triggered: The original current / voltage waveforms (sampled at 20 kHz) from 10 seconds before the fault to 2 seconds after the fault, the arc feature index change curve and the comparator trigger timestamp, and the system operation log (charge and discharge status, temperature, load impedance). The key determination signal (such as the Level 1 disconnection instruction) is generated by three independent comparators and is output using a 2-out-of-3 voting mechanism. Set up an analog protection circuit independent of the MCU: When the current > 80A, trigger the thyristor (BT152) through the Zener diode (BZX55C) to force the short-circuit fuse.Use a programmable load (such as Chroma 63200A) to inject a standard arc waveform (compliant with UL 1699B test specifications), automatically traverse test cases through the GPIB interface: 50 - 1000V voltage range, 10 - 200A current step, 50 - 500Hz arc frequency. Perform a hardware self - test when powering on: trigger virtual fault signals in sequence to verify the response time and action accuracy at each level. Automatically run a calibration sequence every 24 hours: inject a known - amplitude harmonic signal to dynamically correct the deviation of wavelet decomposition coefficients and RMS calculations.

[0038] Preferably, step S3 includes the following steps: Step S31: Perform numerical differentiation on the current monitoring data to obtain instantaneous current change rate data; perform smoothing filtering on the instantaneous current change rate data to generate smoothed instantaneous current change rate data; Step S32: Compare the smoothed instantaneous current change rate data with the short - circuit discrimination threshold. When the smoothed instantaneous current change rate data is greater than or equal to the short - circuit discrimination threshold, generate short - circuit trigger flag data; Step S33: Perform a Boolean discrimination on the short - circuit trigger flag data. When the short - circuit trigger flag data is true, start time - window cumulative analysis to analyze the fluctuation amplitude of the current monitoring data under the short - circuit trigger flag data within the time window, and generate time - window cumulative data; Step S34: Perform continuity analysis on the time - window cumulative data. If the current change amplitude in the time - window cumulative data is greater than 20A or the current change duration > 30s, generate a continuous short - circuit fault discrimination result; perform surge analysis on the bidirectional charge - discharge circuit based on the time - window cumulative data to generate instantaneous surge data.

[0039] In the embodiments of the present invention, an analog differentiator is built using an operational amplifier (such as OPA2182) to perform real-time differential calculation on the output signal of the Hall current sensor, realizing the hardware-level extraction of di / dt (bandwidth limited to 50 kHz to avoid high-frequency noise amplification). Synchronously, a digital three-point central difference method (formula: di / dt = (I_{n + 1}-I_{n - 1}) / (2Δt), Δt = 20 μs) is implemented in the FPGA as a redundant verification channel. An adaptive Kalman filter (implemented by the FPGA hardcore) is used: the filtering gain is dynamically adjusted according to the variance of the current change rate, enhancing smoothing when the noise is large (cutoff frequency 500 Hz), and retaining a 1 kHz bandwidth in the steady state. A second-order Butterworth low-pass filter (cutoff frequency 2 kHz) is configured as hardware preprocessing to eliminate the influence of switching noise on differentiation. It is dynamically calculated according to the rated current of the circuit (such as threshold = 0.2×I_rated, when I_rated = 100 A, the threshold is 20 A / ms), and updated online through the SPI interface. A fixed value of 50 A / ms is used as an emergency trigger line (such as MOSFET direct cut-off protection). A high-speed comparator (such as TI TLV3201) is deployed to compare the smoothed di / dt with the threshold in real time. When the limit is exceeded: immediately pull up the FPGA interrupt pin to trigger a priority task; latch the current timestamp and current value into the register for subsequent modules to read. Only when three consecutive sampling points (interval 10 μs) all exceed the threshold, a stable short-circuit trigger flag (True) is output. Short window (100 ms): stores the current data of 1000 points per second, and calculates the peak-to-peak value (Max - Min) and the root mean square value (RMS) within the window in real time. Long window (30 s): records the duration of the short-circuit trigger flag, and the cumulative duration of the True state is counted by a 32-bit counter (built into the FPGA). In the short window, if the current fluctuation > 10 A (absolute value) and the change rate > 5 A / ms, it is marked as an "effective fluctuation event". The density of effective events within the statistical window (such as ≥5 events within 100 ms) is used as an accumulation index. When the long window counter shows that True lasts > 25 seconds and the short window fluctuation density > 70%, a continuous short-circuit fault (Level 2) is triggered. If the current mutation > 20 A and cannot recover to within ±5% of the baseline within 30 seconds, combined with the temperature sensor data (such as > 90 °C), it is forcibly determined as an irreversible short circuit. After the short circuit is triggered, a 100 kHz oversampling mode is started (the ADC switches to the high-speed mode), and the current waveform from 5 ms before the fault to 50 ms after the fault is recorded.Calculate the three surge elements: capture the absolute maximum value of the waveform, the reciprocal of the time difference between 10% and 90% of the peak value (for example, if it reaches 90% of the peak value within 0.1 ms, the slope = 800 A / ms), perform the ∫I²dt operation on the surge section (accelerated by the FPGA multiplier-accumulator MAC core) for contact life prediction, thereby generating instantaneous surge data. It is also possible to read the radiator temperature (such as MAX30205) through the I2C interface. For every 10°C increase in temperature, the dynamic threshold is reduced by 5%. When the voltage fluctuation > 15%, the short-circuit determination is automatically frozen for 500 ms to avoid misjudgment. The three-level segmentation strategy is cited: Level 1 (<100 μs): When di / dt > 50 A / ms, the CPLD directly drives the SiC MOSFET (such as C3M0065090D) to forcibly cut off the main circuit. Level 2 (<5 ms): After the continuous short-circuit fault is confirmed, the MCU controls the relay to disconnect, and at the same time triggers the pre-charge resistor to be connected to prevent the arc from reigniting. Level 3 (early warning): When the fluctuation exceeds the limit but does not last, a load reduction instruction is sent through the CAN bus (such as the BMS limits the current to 50%).

[0040] Preferably, the surge analysis of the bidirectional charge and discharge circuit based on the time-window cumulative data in step S34 includes: Extract the current characteristics from the smoothed instantaneous current change rate data to obtain the instantaneous current change characteristic data, where the current characteristic extraction includes the current change rate, the fluctuation amplitude, and the high-frequency component; Map the instantaneous current change characteristic data to a high-dimensional space through a radial basis function to construct a hyperplane, obtaining a surge hyperplane; input the time-window cumulative data into the surge hyperplane for time-shift analysis to generate surge time-shift data; Calculate the lateral occupancy ratio of the surge hyperplane according to the surge time-shift data to obtain the surge lateral occupancy ratio data; calculate the instantaneous surge intensity of the current monitoring data under the short-circuit trigger flag data based on the surge lateral occupancy ratio data to obtain the instantaneous surge intensity at each time point; Aggregate the instantaneous surge intensity at each time point to generate instantaneous surge data.

[0041] In the embodiment of the present invention, three parallel processing channels are constructed in the FPGA: the instantaneous di / dt is calculated by using the three-point central difference method (Δt = 10 μs), and the result is output in 16-bit fixed-point numbers. The current data is statistically analyzed by a sliding window (window length 100 ms), and the peak-to-peak value (Max-Min) and standard deviation (σ) are calculated in real time. The high-frequency signal is extracted by a sixth-order Butterworth high-pass filter (cutoff frequency 5 kHz), and its root mean square value (RMS) is calculated. The outputs of the three channels are packed into a 32-bit feature vector (12-bit di / dt + 10-bit peak-to-peak value + 10-bit high-frequency RMS). A programmable RBF core is configured in the FPGA, and 256 center points of the basis functions are pre-stored (generated by clustering historical surge data), and the Gaussian kernel function (σ = 0.5) is used. The calculation of the distance between the input feature vector and the center point is realized by a LUT (look-up table): 4096 pre-calculated Gaussian response values are stored (the address line corresponds to the quantized Euclidean distance value), and the real-time calculation is as follows: Euclidean distance between the feature vector and the center point → LUT address → kernel output value. The outputs of 256 kernels are weighted and summed (the weights are updated by online learning) to generate an 8-bit hyperplane state value. A FIFO buffer with a depth of 1024 is constructed (corresponding to 10.24 ms @ 100 kHz) to store the hyperplane state sequence within the time window. The offset between the current state sequence and the historical template is calculated by the sliding window cross-correlation algorithm (window length 50 points), and 50 groups of dot products are calculated in parallel by using the built-in DSP48 unit of the FPGA, and the peak detection determines the maximum correlation offset. A three-dimensional feature space coordinate system (di / dt, peak-to-peak value, high-frequency RMS) is established, and the lateral component is extracted: the current feature vector is projected onto the vertical plane of the hyperplane normal vector, and the calculation formula is: lateral vector = original vector - (original vector × normal vector) × normal vector; the calculation of the squared modulus value is completed by the MAC (multiply-accumulate) hard core, and the result is normalized to the occupancy ratio of 0 - 100%. Based on the lateral occupancy ratio (X) and the current di / dt value (Y), a two-dimensional intensity matrix is constructed: intensity = 0.7×X + 0.3×(Y / Y_max), where Y_max = 100 A / ms (configurable), and the matrix outputs an 8-bit intensity value (0 - 255), corresponding to the surge intensity of 0 - 100%. The intensity value is subjected to a moving average to output a smooth surge trend, and the area integral of the intensity value (∫ intensity dt) is calculated for energy evaluation. The maximum intensity value is recorded and latched in real time and continuously output before manual reset. The instantaneous intensity, trend, and integral value are packed into a 64-bit data frame (32-bit timestamp + 16-bit intensity value + 16-bit energy value) and transmitted to the dual-port RAM through the DMA channel for real-time reading by the host computer.

[0042] Preferably, generating a control instruction according to the DC arc discrimination result and the continuous short-circuit fault discrimination result in step S4 includes: Performing high-order analysis on the signal characteristics in the DC arc discrimination result to generate arc duration and waveform change data; Perform real-time pattern recognition on the current waveforms in the short-circuit fault discrimination results to generate short-circuit pattern category data; Conduct joint spatio-temporal analysis on the arc duration, waveform change data, and short-circuit pattern category data to generate fault collaborative impact data; Fuse the time domain and frequency domain of the fault collaborative impact data to generate comprehensive fault severity data; Redundantly verify the DC arc discrimination results and continuous short-circuit fault discrimination results through the comprehensive fault severity data and generate control instructions to obtain control instructions.

[0043] In the embodiments of the present invention, by performing high-order feature extraction on the high-frequency disturbance segments in the voltage and current signals included according to the DC arc discrimination result, the analysis methods include but are not limited to: wavelet transform to obtain multi-scale time local changes; instantaneous frequency extraction to judge the arc starting and extinguishing moments; arc envelope feature extraction to analyze its duration fluctuation; finally generating arc duration and waveform change data, which can accurately describe the evolution trend and interference characteristics of the arc phenomenon. Based on the continuous short-circuit fault discrimination result, key features of the current waveform (such as rising edge slope, stable region current amplitude, oscillation frequency, etc.) are extracted, and a real-time signal classification algorithm (such as support vector machine SVM, convolutional neural network CNN or time series clustering method) is used to perform pattern recognition on the short-circuit signal, and the output is: short-circuit type (such as single-point short circuit, continuous short circuit, intermittent short circuit); short-circuit response mode category (such as burst type, transition type, oscillation type); finally generating short-circuit mode category data, which can be used to discriminate the fault evolution path. Align the arc duration and waveform change data with the short-circuit mode category data on the time axis and perform spatial association. Through the following joint analysis operations: establish a collaborative interference time window; analyze the triggering and mutual influence relationship between the two types of faults; establish a spatio-temporal feature cross-mapping matrix to extract coupling behavior features; output fault collaborative influence data, which is used to further describe the global state of the system affected by the composite fault disturbance. Apply a spectrum fusion method (such as the STFT+EMD joint algorithm, spectrum energy clustering method) to the fault collaborative influence data, and combine time domain indicators (such as average value, variance, fluctuation rate, etc.) for feature fusion analysis, and the output is: comprehensive frequency domain energy distribution; fault evolution trend curve; severity level identification label (such as L1~L4 level division); thus obtaining comprehensive fault severity data, which provides accuracy support for subsequent control strategies. Use the comprehensive fault severity data to cross-verify the DC arc discrimination result and the short-circuit fault discrimination result: if the two results are consistent, improve the fault judgment confidence; if there are differences in the results, perform priority correction through the fault model weight mechanism; based on the verified results, call the preset control strategy library and automatically match the following instructions: cut off the DC power supply; start the protection device (such as an arc extinguisher); notify the remote system for manual review; finally output the control instruction to achieve the real-time and accurate response of the system to complex fault events.

[0044] Preferably, in step S4, the analysis of the overcurrent morphology of the relay contacts in the bidirectional charge and discharge circuit based on the instantaneous surge data includes: Capture high-frequency surge events from the current monitoring data of the bidirectional charge and discharge circuit based on the instantaneous surge data to generate instantaneous surge waveform data; Perform FIR band-pass filtering on the instantaneous surge waveform data to generate denoised surge feature data; Extract the time domain features of the denoised surge feature data to obtain surge parameter quantization data; Analyze the contact current distribution of the relay in the bidirectional charge and discharge circuit using surge parameter quantization data to generate contact current distribution data; simulate the local contact current propagation of the relay based on the contact current distribution data to generate relay contact current exchange simulation data; Analyze the change of electronic quantum state of the relay contact current exchange simulation data to generate electronic quantum state change data; Predict the contact wear of the relay contacts through the electronic quantum state change data to generate the analysis results of the contact overcurrent morphology.

[0045] In the embodiments of the present invention, real-time current monitoring data is obtained from a bidirectional charge and discharge circuit, with particular attention paid to instantaneous surge events in the current. This data is typically obtained by real-time acquisition using a current sensor or a current probe. Signal processing techniques are used to capture high-frequency surge events in the current signal. By adopting an instantaneous surge data acquisition technique, the high-frequency components in the surge signal can be identified in real time. These components usually contain sudden changes or instantaneous variations in the current, such as short-term overcurrents or short pulses. Through a high-frequency capture algorithm, the mutation regions in the current waveform are quickly extracted and processed to generate accurate instantaneous surge waveform data, providing a basis for subsequent filtering and feature extraction. The generated instantaneous surge waveform data is fed into a FIR band-pass filter. The function of the FIR band-pass filter is to filter out the unwanted low-frequency and high-frequency noises in the current signal. Generally, these noises do not contribute to the analysis of the current characteristics, and the surge information in the middle frequency band of the signal is the key focus. An appropriate frequency band is set in the band-pass filter to ensure that the signals passing through the intermediate frequency band can filter out the irrelevant noises. After filtering, the surge characteristics in the signal are retained while the unnecessary noises are removed. The data after filtering is the denoised surge feature data. These data are clearer, with reduced interference, facilitating the subsequent extraction of time-domain features. Time-domain analysis is performed on the denoised surge feature data to extract important parameters in the current waveform. These time-domain features include peak current, overcurrent duration, waveform amplitude, rise time, decay time, etc., which directly reflect the characteristics of the current transient behavior. The extracted time-domain features are quantized to be converted into standardized surge parameter data. These parameter data can clearly quantify the behavior of the current signal, facilitating subsequent analysis and comparison. Based on the quantized surge parameter data, the current distribution at the relay contacts is calculated using an electromagnetic field simulation model or circuit analysis software. By the flow characteristics of the current on the contacts, the current density and distribution of the contacts can be obtained, and the regions where the contacts bear a large current can be determined. By analyzing the spatial distribution of the contact current, contact current distribution data is generated, which will contribute to subsequent wear analysis of the contacts and current propagation simulation. Based on the contact current distribution data, local current propagation simulation of the relay contacts is carried out using circuit simulation software. During the simulation process, the process of current exchange and propagation between the contacts is studied, including the direction and intensity of the current and its interaction with the contact surface. During the simulation, factors such as contact impedance and contact surface friction between the contacts are considered to generate relay contact current exchange simulation data. These data can help understand the dynamic effects generated during current exchange at the contacts, such as contact heating and arc generation. Using the principles of electronic quantum physics, based on the relay contact current exchange simulation data, the change in the quantum state of electrons during current transmission is analyzed. At this time, the thermal effect, arc effect caused by the current, and the scattering and acceleration process of electrons at the contact points are considered.Through the quantum physical model and the theory of electron state changes, analyze the process of electron state changes and generate electron quantum state change data, which reflect the microscopic behavior and physical effects of electrons during the contact exchange process. According to the electron quantum state change data, combined with the current distribution of relay contacts, the simulation results of contact current exchange, and quantum state analysis, construct a wear prediction model for relay contacts. The model takes into account the contact current, electron state changes, and their physical effects on the contact surface. Through this model, calculate the wear degree, wear rate of relay contacts, and their potential impact on current transmission efficiency. The generated analysis results can help identify potential overcurrent risks and provide an assessment of the relay contact life. Through the prediction of relay contact wear, generate a complete analysis result of the contact overcurrent morphology. These results include the prediction of contact wear and the performance changes of contacts under different working conditions. These data provide strong support for the design optimization and maintenance of relays.

[0046] Of particular importance, the local contact current propagation simulation of the relay based on the contact current distribution data also includes: Perform ensemble empirical mode decomposition on the relay based on the contact current distribution data to generate an intrinsic mode function data IMF set; perform Hilbert spectral analysis on the IMF set data to generate time-frequency energy distribution data; Perform phase space reconstruction and recursive network construction processing on the time-frequency energy distribution data to generate recursive network adjacency matrix data; perform persistent homology analysis on the recursive network adjacency matrix data to generate topological feature Betti number sequence data; Perform fractional-order differentiation on the Betti number sequence data to generate fractional-order current diffusion response data; perform transfer entropy calculation on the fractional-order current diffusion response data to generate direction coupling coefficient matrix data; Perform Monte Carlo sampling processing on the direction coupling coefficient matrix data to generate relay contact current exchange simulation data.

[0047] In the embodiments of the present invention, current sampling devices are respectively arranged on a relay contact array to synchronously record the current time series of each contact at a fixed sampling frequency (for example, 10,000 times per second). First, a band-pass filter is used to remove the DC offset and noise far above or below the target frequency band; then, the average value of the filtered data for each channel is subtracted and divided by its standard deviation to make the data amplitude scales of all channels consistent for subsequent analysis. For each preprocessed current signal, multiple groups of random noise are repeatedly added and traditional empirical mode decomposition is performed. The first mode in the noise-assisted decomposition results is averaged to obtain the first intrinsic mode function of this round. The remaining signal after stripping the first mode continues to perform the same steps until the remaining signal tends to be monotonic or below the noise level. The number of noise groups can be up to hundreds, and the noise amplitude is set to about one-twentieth of the original signal amplitude to ensure the stability and accuracy of the decomposition. For each obtained intrinsic mode function, the Hilbert transform technique is applied to combine its real part and imaginary part to generate the corresponding analytic signal. The instantaneous amplitude and instantaneous frequency are respectively extracted from the analytic signal to characterize the evolution of the intensity change and frequency change of the current oscillation over time. The squared instantaneous amplitudes of all mode functions are accumulated to the corresponding time-frequency positions to form a two-dimensional data matrix reflecting the dual distribution of energy over time and frequency. Appropriate time delay and embedding dimension are selected, and the time-frequency energy sequence is converted into a series of multi-dimensional vectors in a "sliding window" manner to obtain the state space trajectory of the system. The reconstructed multi-dimensional state vector sequence is input into a gated recurrent unit or a long short-term memory network for training, so that the network learns to capture the time dependence and dynamic characteristics of the sequence. According to the output of the hidden layer after network training, the similarity between the hidden states at any two moments is calculated, and a weighted matrix is formed according to the similarity magnitude as the adjacency matrix of the recurrent network. Based on the adjacency matrix, a filtering process of the network is gradually constructed according to a series of similarity thresholds, and a multi-layer network structure is generated in order from the closest to the loosest. The number of connected components and the number of loops of the network are extracted at each threshold level, and these topological features changing with the threshold are recorded to form a continuous Betti number sequence. Taking the Betti number sequence as the input, the fractional difference method is applied to perform a differential operation with long-range memory characteristics on the sequence to obtain a new sequence reflecting the diffusion response of topological features. A fractional order between zero and one (for example, 0.5) is selected to balance the sensitivity to short-term mutations and long-term trends. For each pair of channels, based on their fractional diffusion response sequences, an index measuring the information transfer amount "from channel A to channel B" is calculated to reveal the direction of dynamic causal influence. The transfer entropy values of all channel pairs are arranged and combined to form a complete direction coupling coefficient matrix, and each item in the matrix reflects a specific current propagation intensity and direction. The probability distribution of the direction coupling coefficient matrix is fitted (such as normal distribution or kernel density estimation), and sample matrices are randomly drawn from it multiple times.Set the same initial current distribution for each sampling matrix and iterate the current exchange process in fixed steps. At each step, multiply the current vector by the sampling matrix to update the current distribution. Collect the current distribution results of all sampling cases at each iteration step, calculate statistics such as the average curve, variance, maximum and minimum values, etc., to form the final simulation dataset. Finally, generate a complete simulation report including time-frequency energy diagrams, Betti number persistence diagrams, coupling coefficient heat maps, and Monte Carlo distribution statistical diagrams.

[0048] Of particular importance, the analysis of the electronic quantum state changes in the relay contact current exchange simulation data also includes: Extract the initial quantum state values of the relay contact current exchange simulation data; Conduct an analysis of the electronic energy level distribution for the initial quantum state values to generate electronic energy level change data; Conduct a simulation of the quantum state time evolution for the electronic energy level change data to generate electronic quantum state time change data; Conduct an analysis of the electronic spin state for the electronic quantum state time change data to generate electronic spin change data; Conduct an evaluation of the quantum state changes for the electronic spin change data to generate electronic quantum state change evaluation data.

[0049] In the embodiments of the present invention, on the basis of simulating the current exchange of relay contacts, it is first necessary to establish the correspondence between current signals and quantum states. By measuring the current data of relay contacts, the electronic states of each contact at the initial moment can be deduced. Based on the current exchange process, the initial values of quantum states related to current can be extracted. At this time, the quantum state can be mapped to the state of a quantum bit through information such as the magnitude and phase of the contact current, and the quantum state parameters of each contact are initially constructed, such as electron energy, electron spin, etc. Based on the principles of quantum mechanics, an energy distribution model of the quantum state is established. The current data of each relay contact will be mapped to a specific energy level. Fourier transform or wavelet transform is performed on the contact current data to extract the frequency components related to energy from it. The energy information obtained from the frequency domain analysis is matched with the energy levels of the quantum state, and through statistical analysis, the distribution of each quantum state at different energy levels is obtained. By analyzing the time evolution of the relay contact current exchange, a curve of the electron energy level changing with time is plotted to generate electron energy level change data. The time-dependent Schrödinger equation is used to simulate the evolution of the quantum state, considering the external influences caused by current exchange (such as changes in electric and magnetic fields, etc.). The initial quantum state is evolved through the simulated time evolution process, and the electron energy distribution and the changes in its state at each time point are calculated. The evolution of the quantum state at different time steps is sampled, and the change data such as electron energy, phase, and spin state are recorded to generate complete quantum state time change data. Based on the mathematical representation of the quantum state, the electron spin components at each time step are extracted. The electron spin state is usually represented by the wave function of the quantum bit, and the spin direction of each electron can be extracted from the quantum state. Quantum physical methods such as the spin Hall effect or the spin transfer torque effect are used to analyze the influence of the current exchange process on electron spin and detect the changes in spin during the time evolution process. Through the combination of simulation and experimental data, the change trend of the spin state with time is obtained, and a spin change curve is plotted to form electron spin change data. The stability of the evolution process of the quantum state is evaluated. Analyze whether the evolution of the quantum state tends to be stable, or whether phenomena such as volatility, chaos, or decoherence occur at certain time points. Based on indicators such as quantum entanglement and quantum coherence in quantum information theory, evaluate whether the changes in the quantum state during the relay contact current exchange process will lead to the loss or reduction of the quality of quantum information. The evaluation results are quantified through statistical methods to generate electron quantum state change evaluation data, including indicators such as quantum state stability, entanglement degree change, and coherence loss.

[0050] Preferably, the optimization of the conduction time control of the control instruction according to the analysis result in step S4 includes: The analysis results of the overcurrent morphology of the contacts are cumulatively processed for the overcurrent duration to generate the total overcurrent duration data of the contacts; The total overcurrent duration data of the contacts are processed by interval comparison to generate the overcurrent trend data of the contacts; Extract the conduction signal limit of the overcurrent trend data of the contact, and generate the allowable duration data of the contact conduction signal, where the allowable duration data of the contact conduction signal includes the maximum single duration and the minimum interval time; Optimize the continuous conduction time of the circuit for the control instruction according to the maximum single duration and the minimum interval time, and generate the conduction time limit parameter data; Use the conduction time limit parameter data to control and adjust the conduction time of the control instruction, so as to perform the safety control optimization operation of the bidirectional charge and discharge circuit.

[0051] In the embodiments of the present invention, through a current monitoring system, real-time current data of the relay contacts in the bidirectional charging and discharging circuit is collected. This data includes the current waveforms during each conduction process of the contacts. When the contact current exceeds the set overcurrent threshold, the overcurrent duration at this time is recorded and calibrated as the "overcurrent duration". On this basis, the durations of all overcurrent events are accumulated to generate the total overcurrent duration data of the contacts. For example, when analyzing an actual working charging and discharging circuit, it is found that the relay contacts frequently experience sudden increases in current during multiple charging and discharging cycles. The durations of each overcurrent are 10 milliseconds, 8 milliseconds, 12 milliseconds, etc. respectively. After adding these durations, the total overcurrent duration of the contacts is obtained as 30 milliseconds. Further analysis is carried out on the total overcurrent duration data of the contacts. The overcurrent events are compared in intervals according to a certain time interval (such as every hour or every minute). In this way, the law that the contacts bear a large overcurrent load during certain periods can be identified. For example, through specific analysis, it is obtained that the relay contacts experience more overcurrent events between 8 am and 10 am, and the total overcurrent duration is 20 milliseconds, while between 3 pm and 5 pm, the overcurrent duration is significantly reduced, only 5 milliseconds. Through such interval comparison processing, the overcurrent trend data of the contacts is obtained, that is, the change of the overcurrent of the contacts in different time periods. According to the overcurrent trend data of the contacts, the limiting conditions of the conduction signal can be extracted. Specifically, it is necessary to calculate: the maximum single duration: that is, the longest duration that the relay contacts can safely withstand each time they conduct. By analyzing the historical overcurrent data, it is found that when the relay contacts bear an overcurrent for more than 15 milliseconds, the contact wear will accelerate. Therefore, the maximum single conduction duration is set to 15 milliseconds. The minimum interval time: that is, the shortest time interval that needs to be maintained between two conductions. It is assumed that it is found that when the time interval between two conductions is less than 2 seconds, the temperature of the contacts is too high and overcurrent damage is likely to occur. Therefore, the minimum interval time is set to 2 seconds. According to the above settings, the allowable duration data of the contact conduction signal is generated, which includes the maximum single duration of 15 milliseconds and the minimum interval time of 2 seconds. According to the generated maximum single duration and minimum interval time, the control instructions are optimized and adjusted. In the design and control system of the circuit, PWM (pulse width modulation) signals are usually used to control the conduction duration of the relay. Therefore, according to the maximum single duration and minimum interval time, these parameters are used as constraint conditions to optimize the conduction duration setting in the control instructions. Through an optimization algorithm, appropriate conduction duration limit parameters are calculated to ensure that the maximum duration of each conduction is 15 milliseconds and to ensure that there is at least a 2-second interval between two conductions. In this way, the conduction time limit parameter data in the control system is set as: the maximum single duration = 15 milliseconds, the minimum interval time = 2 seconds. The instructions in the control system are adjusted by using the generated conduction time limit parameter data. Specifically, in the bidirectional charging and discharging circuit, the conduction control signal of the relay will be adjusted according to the above optimization parameters.By limiting the maximum single conduction duration to 15 milliseconds and setting a minimum interval of 2 seconds between two conductions, it is ensured that the relay contacts will not suffer from overcurrent damage due to continuous conduction for too long, thus effectively extending the service life of the relay and improving the stability of the system. For example, during actual operation, the system will automatically adjust the conduction time according to the above settings. When the contacts receive a conduction instruction in the working state, the control system will determine whether the conduction duration meets the maximum single duration requirement. If it exceeds 15 milliseconds, the conduction will be automatically interrupted, and it can only be conducted again after waiting for at least 2 seconds, thus preventing the risk of overcurrent in the relay contacts.

[0052] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to cover all changes falling within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0053] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A control method for a bidirectional charge and discharge circuit, characterized in that: The following steps are involved: Step S1: real-time acquisition of current monitoring data and voltage monitoring data of a bidirectional charge and discharge circuit; Perform arc characteristic analysis on the current monitoring data and the voltage monitoring data to obtain arc characteristic data; Step S2: analyzing the high-frequency harmonic components and current ripple fluctuation amplitude of the bidirectional charge and discharge circuit based on the arc characteristic data; performing DC arc detection on the bidirectional charge and discharge circuit according to the high-frequency harmonic components and current ripple fluctuation amplitude, and generating a DC arc fault determination result; Step S3: Calculate the instantaneous current change rate of the current monitoring data, and perform short circuit detection on the bidirectional charge and discharge circuit through the instantaneous current change rate. When a current change that continuously exceeds a preset threshold is detected, start the time window accumulation analysis to generate instantaneous surge data and continuous short circuit fault judgment results; Step S4: generating a control instruction according to the DC arc determination result and the continuous short circuit fault determination result; The contact overcurrent profile of the relay in the bidirectional charging and discharging circuit is analyzed based on the transient surge data, and the on-time control of the control instruction is optimized according to the analysis results to perform the safety control optimization operation of the bidirectional charging and discharging circuit.

2. The control method of the bidirectional charge and discharge circuit according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: collecting current monitoring data and voltage monitoring data of the bidirectional charge and discharge circuit in real time; Step S12: performing time alignment on the current monitoring data and the voltage monitoring data to generate time-aligned current and voltage monitoring data; performing differential operation on the time-aligned current and voltage monitoring data to generate current change rate data and voltage change rate data; Step S13: performing mutation edge detection on the current change rate data and the voltage change rate data to generate suspicious arc trigger point data; extracting the time window of the suspicious arc trigger point data, and performing envelope analysis on the current and voltage monitoring data according to the time window to generate arc voltage waveform characteristic data; Step S14: performing spectrum analysis on the arc voltage waveform characteristic data to generate arc frequency domain characteristic data; performing key feature screening processing on the arc frequency domain characteristic data to generate arc characteristic data.

3. The control method of the bidirectional charge and discharge circuit according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing spectrum decomposition processing on the arc characteristic data, extracting the high-frequency harmonic components of the current signal, and generating high-frequency harmonic data; performing amplitude analysis on the high-frequency harmonic data, and generating high-frequency harmonic amplitude data; Step S22: performing time domain waveform analysis on the current monitoring data, calculating the current ripple fluctuation amplitude, and generating current ripple fluctuation data; calculating the arc characteristic index of the circuit according to the high-frequency harmonic amplitude data and the current ripple fluctuation data, and generating arc characteristic index data; Step S23: using the arc characteristic index data to perform DC arc detection on the bidirectional charge and discharge circuit to generate a DC arc fault determination result.

4. The control method of the bidirectional charge and discharge circuit according to claim 3, characterized in that: Step S23 includes the following steps: Step S231: When the arc characteristic index shows any of the following conditions, it is determined to be a DC arc fault and the first fault discrimination data is obtained: the arc characteristic index is greater than 5.0; the current fluctuation exceeds ±10% and lasts for more than 3 seconds; the arc occurrence frequency is greater than 50 times / minute; the discharge voltage is greater than 400V or less than 200V; Step S232: When the relationship between the arc characteristic index and the arc occurrence frequency meets the following conditions, the second fault discrimination data is generated: the arc characteristic index is greater than 6.0 and the arc occurrence frequency is greater than 60 times / minute; the fluctuation range of the arc characteristic index is greater than ±20%, and the current value variation range exceeds ±15%; Step S233: when determining the DC arc fault of the bidirectional charge and discharge circuit, if the following conditions occur, the third fault determination data is generated: the current is greater than 50A or the voltage is greater than 500V, and the duration exceeds 60 seconds; the current fluctuation is greater than ±20% and the voltage fluctuation is greater than ±15%, and the duration of these fluctuations exceeds 30 seconds; Step S234: Integrate the first fault determination data, the second fault determination data and the third fault determination data to finally generate a DC arc fault determination result.

5. The control method of the bidirectional charge and discharge circuit according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: performing numerical differentiation calculation on the current monitoring data to obtain instantaneous current change rate data; performing smoothing filtering on the instantaneous current change rate data to generate smoothed instantaneous current change rate data; Step S32: comparing the smooth instantaneous current change rate data with the short circuit determination threshold, and when the smooth instantaneous current change rate data is greater than or equal to the short circuit determination threshold, generating short circuit trigger flag data; Step S33: Performing Boolean discrimination on the short-circuit trigger flag data, when the short-circuit trigger flag data is true, starting the time window cumulative analysis to analyze the fluctuation amplitude of the current monitoring data under the short-circuit trigger flag data within the time window, and generating the time window cumulative data; Step S34: Perform continuity analysis on the accumulated data in the time window. If the current change amplitude in the accumulated data in the time window is greater than 20A or the current change duration is >30s, a continuous short circuit fault judgment result is generated; perform surge analysis on the bidirectional charging and discharging circuit based on the accumulated data in the time window to generate instantaneous surge data.

6. The control method of the bidirectional charge and discharge circuit according to claim 4, characterized in that: In step S34, the surge analysis of the bidirectional charge and discharge circuit based on the accumulated data in the time window includes: Performing current feature extraction on the smooth instantaneous current change rate data to obtain current instantaneous change feature data, wherein the current feature extraction includes current change rate, fluctuation amplitude and high-frequency component; The instantaneous change characteristic data of the current is mapped to the high-dimensional space through the radial basis function to construct a hyperplane, thereby obtaining a surge hyperplane; the accumulated data of the time window is input into the surge hyperplane for time shift analysis to generate surge time shift data; The plane lateral proportion of the surge hyperplane is calculated according to the surge time-shift data to obtain the surge lateral proportion data; based on the surge lateral proportion data, the instantaneous surge intensity of the current monitoring data under the short-circuit trigger mark data is calculated to obtain the instantaneous surge intensity at each time point; The instantaneous surge intensity at each time point is aggregated to generate instantaneous surge data.

7. The control method of the bidirectional charge and discharge circuit according to claim 1, characterized in that: In step S4, generating a control instruction according to the DC arc determination result and the continuous short circuit fault determination result includes: Perform high-level analysis on the signal characteristics in the DC arc identification results to generate arc duration and waveform change data; Performing real-time pattern recognition on the current waveform in the short-circuit fault identification result to generate short-circuit pattern category data; Conduct joint spatiotemporal analysis on arc duration and waveform change data and short-circuit mode category data to generate fault synergistic impact data; Fuse the fault collaborative impact data in the time domain and frequency domain to generate comprehensive fault severity data; The DC arc identification result and the continuous short circuit fault identification result are redundantly verified by comprehensive fault severity data, and control instructions are generated to obtain control instructions.

8. The control method of the bidirectional charge and discharge circuit according to claim 1, characterized in that: Analyzing the contact overcurrent morphology of the relay in the bidirectional charge and discharge circuit based on the instantaneous surge data in step S4 includes: Based on the instantaneous surge data, the current monitoring data of the bidirectional charging and discharging circuit is used to capture high-frequency surge events and generate instantaneous surge waveform data; Perform FIR bandpass filtering on the instantaneous surge waveform data to generate denoised surge characteristic data; Extract the time domain features of the denoised surge characteristic data to obtain surge parameter quantification data; Use surge parameter quantification data to analyze the contact current distribution of the relay in the bidirectional charging and discharging circuit and generate contact current distribution data; simulate the local contact current propagation of the relay based on the contact current distribution data and generate relay contact current exchange simulation data; Performing electron quantum state change analysis on relay contact current exchange simulation data to generate electron quantum state change data; The contact wear of the relay contacts is predicted through the data of electron quantum state changes, and the analysis results of the contact overcurrent morphology are generated.

9. The control method of the bidirectional charge and discharge circuit according to claim 1, characterized in that: In step S4, optimizing the on-time control of the control instruction according to the analysis result includes: The analysis results of the contact overcurrent morphology are processed by accumulating the overcurrent duration to generate the total contact overcurrent duration data; Perform interval comparison processing on the total contact overcurrent duration data to generate contact overcurrent trend data; Extract the conduction signal limit of the contact overcurrent trend data, and generate the contact conduction signal allowable duration data, wherein the contact conduction signal allowable duration data includes the maximum single duration and the minimum interval time; Optimize the continuous conduction time of the circuit for the control instruction according to the maximum single duration and the minimum interval time, and generate the conduction time limit parameter data; The on-time limit parameter data is used to adjust the on-time control of the control instruction to perform the safety control optimization operation of the bidirectional charging and discharging circuit.

10. A control system for a bidirectional charge and discharge circuit, characterized in that: The control method for the bidirectional charge-discharge circuit according to claim 1 is used to execute the control method of the bidirectional charge-discharge circuit, wherein the control system of the bidirectional charge-discharge circuit comprises: The arc feature extraction module is used to collect the current monitoring data and voltage monitoring data of the bidirectional charging and discharging circuit in real time; perform arc feature analysis on the current monitoring data and the voltage monitoring data to obtain arc feature data; An arc fault identification module is used to analyze the high-frequency harmonic components and current ripple fluctuation amplitude of the bidirectional charging and discharging circuit based on arc characteristic data; perform DC arc detection on the bidirectional charging and discharging circuit according to the high-frequency harmonic components and current ripple fluctuation amplitude, and generate a DC arc fault identification result; The short-circuit identification module is used to calculate the instantaneous current change rate of the current monitoring data, and to perform short-circuit detection on the bidirectional charge and discharge circuit through the instantaneous current change rate. When a current change that continuously exceeds a preset threshold is detected, the time window accumulation analysis is started to generate instantaneous surge data and continuous short-circuit fault identification results; The contact conduction control module is used to generate control instructions based on the DC arc discrimination results and the continuous short circuit fault discrimination results; based on the instantaneous surge data, the contact overcurrent morphology of the relay in the bidirectional charging and discharging circuit is analyzed, and the conduction time control of the control instructions is optimized according to the analysis results to perform the safety control optimization operation of the bidirectional charging and discharging circuit.

Images