Battery management system and method based on AFE chip

Abstract

The invention relates to the field of battery management, and discloses a battery management system and method based on an AFE chip, and the system comprises a main controller, an AFE data acquisition subsystem, a data anti-interference subsystem, a disconnection diagnosis subsystem, a hierarchical communication subsystem, and a BMS evaluation subsystem. The AFE data acquisition subsystem comprises a data precision calibration module, an asynchronous acquisition and analysis module and an abnormality diagnosis and processing module; the disconnection diagnosis subsystem comprises a current working condition cooperation module, a disconnection detection period module and a multi-dimensional judgment alarm module, and the BMS evaluation subsystem comprises an aging scene evaluation module, a multi-model evaluation module, a dynamic acquisition module, a driving habit quantification module and a battery state parameter correction module. Aiming at the problems of AFE sampling misalignment, cascade communication defects, evaluation system limitation and disconnection detection misjudgment, the battery management process is optimized through full-link collaborative design of data acquisition, transmission, evaluation, protection and control.

Description

Technical Field

[0001] This invention relates to battery management technology, specifically to a battery management system and method based on an AFE chip. Background Technology

[0002] Battery Management Systems (BMS) address battery management issues in electric vehicles through precise monitoring, intelligent control, and proactive protection. BMS is crucial for ensuring vehicle safety, improving performance, and reducing costs. In various application scenarios such as driving and parking, BMS needs to monitor, control, and protect the battery. However, existing BMS systems have several shortcomings in different application scenarios for electric vehicles: (i) Electromagnetic interference generated by the operation of motors and other equipment during the operation of electric vehicles causes noise fluctuations in voltage and current signals, resulting in distorted sampling data; (ii) The disconnection detection of the AFE chip is affected by the instantaneous change of the battery current. In scenarios such as rapid acceleration of electric vehicles, the battery discharges at a high rate instantaneously. If the pull-up / pull-down current source operation of the disconnection detection is encountered, the large change in the discharge current will cause the sampling voltage difference to exceed the set threshold, leading to false diagnosis and seriously affecting the reliability of disconnection detection. (iii) Existing technologies achieve synchronous sampling of different AFE chips through clock synchronization technology. This approach not only has stringent requirements for chip sampling accuracy, but also requires waiting for all AFE chips to complete their data acquisition before data processing can be performed, which reduces system response efficiency. (iv) The evaluation algorithm is trained and optimized around conventional parameters such as voltage, current and temperature. It does not take into account the impact of factors such as changes in the microstructure of motor materials, which lead to battery capacity decay, increased internal resistance and driving habits on battery status. This makes the evaluation algorithm have certain limitations in data acquisition and algorithm evaluation. The above problems frequently occur in practical applications and may lead to the following deeper issues: Distorted sampling data is transmitted as the object of cascaded communication. The signal attenuation in the cascaded communication link further amplifies the data error, causing the deviation between the sampled value received by the back-end chip and the actual value to double. Inaccurate sampling data is used as the input of the evaluation system, resulting in SOC estimation deviation and SOH judgment distortion. It provides an incorrect "voltage difference threshold" for wire break detection and increases the probability of misdiagnosis of AFE chip wiring status. The limitations of the BMS evaluation algorithm prevent the system from identifying the root causes of battery capacity degradation and increased internal resistance. It can only rely on conventional parameter estimation, which is not very accurate. Once sampling is biased, the evaluation system will lack "deep data calibration basis," leading to a further increase in the battery state prediction error. The BMS evaluation algorithm's misjudgment of battery internal resistance will prevent the detection threshold of the disconnection detection from being dynamically adjusted. In instantaneous high current scenarios, it is more likely to exceed the threshold and cause misdiagnosis. The misdiagnosis of disconnection detection will cause the BMS to erroneously disconnect the charging and discharging circuit or fail to protect the circuit in time, accelerating battery aging. In turn, it will make the AFE sampling data more distorted, forming a closed-loop negative cycle of "sampling distortion - evaluation error - misdiagnosis - more distorted sampling." Summary of the Invention

[0003] Therefore, the present invention provides a battery management system and method based on an AFE chip, which effectively solves the technical problems of inaccurate AFE sampling, defects in cascaded communication, limitations of evaluation systems, and misjudgment of disconnection detection in the prior art.

[0004] To solve the above-mentioned technical problems, the present invention specifically provides the following technical solution: a battery management system based on an AFE chip, including a main controller, an AFE data acquisition subsystem, a data anti-interference subsystem, a disconnection diagnosis subsystem, a hierarchical communication subsystem, a BMS evaluation subsystem, and also including an equalization module, a charge and discharge control module, a memory, a power supply module, and a service terminal; The AFE data acquisition subsystem includes an AFE chip, a current and voltage acquisition module, a data accuracy calibration module, an asynchronous acquisition and analysis module, and an anomaly diagnosis and processing module. The line breakage diagnosis subsystem includes a current condition coordination module, a line breakage detection cycle module, and a multi-dimensional judgment and alarm module. The hierarchical communication subsystem includes a hierarchical communication architecture module, a fault diagnosis module, and a hierarchical automatic recovery module; The BMS evaluation subsystem includes an aging scenario evaluation module, a multi-model evaluation module, a dynamic acquisition module, a driving habit quantification module, and a battery status parameter correction module.

[0005] Furthermore, the data interference immunity compensation subsystem includes an interference detection and identification module, a data optimization and processing module, a temperature compensation module, and a thermal management module; The interference detection and identification module includes an interference feature acquisition unit, an interference threshold determination unit, and an interference scene classification unit. The data optimization processing module includes a filtering algorithm execution unit and an interference data replacement unit; The temperature compensation module includes a temperature error calibration unit, a real-time temperature acquisition unit, and a sampling data dynamic compensation unit. The interference feature acquisition unit collects voltage and current data sampled by the AFE chip in real time, records the data acquisition timestamp synchronously, and extracts key feature parameters from the voltage and current data. The interference threshold determination unit presets multiple scene interference feature thresholds and determines whether interference exists. The interference scene classification unit classifies interference scenes according to the differences in interference features. The filtering algorithm execution unit adaptively selects the filtering algorithm based on the interference scene classification results. The interference data replacement unit triggers the data replacement logic when the interference detection module determines that there is transient pulse interference in the circuit. The temperature error calibration unit fits the linear compensation formulas for "temperature-voltage error" and "temperature-current error", generates a calibration table and stores it in the memory. The real-time temperature acquisition unit collects the temperature data of the AFE chip and the battery pack in real time. Before each AFE sampling, the sampling data dynamic compensation unit calls the real-time temperature data and the pre-stored calibration table to correct the current and voltage data collected by the AFE chip and outputs the compensated voltage and current data. The thermal management module performs active thermal management actions.

[0006] Furthermore, the current and voltage acquisition module includes an AFE data acquisition unit and an acquisition parameter configuration unit; The data accuracy calibration module includes a reference voltage calibration unit, a link attenuation calibration unit, and a multi-channel calibration unit. The asynchronous acquisition and analysis module includes an asynchronous acquisition unit, a single-chip fluctuation analysis unit, an alignment unit, and a synchronization analysis unit. The anomaly diagnosis and processing module includes a reference data feedback unit, a multi-dimensional diagnosis unit, an error correction processing unit, and a communication anomaly monitoring unit. The AFE data acquisition unit controls each AFE chip to transmit the acquired analog signal to the AFE chip for analog-to-digital conversion according to the acquisition command issued by the service terminal. The acquisition parameter configuration unit presets and configures the acquisition parameters of the AFE chip according to the battery pack type and acquisition requirements. The reference voltage calibration unit periodically sends reference calibration commands to the AFE chips of each subunit. After receiving the command, the AFE chip compares the collected battery voltage data with the output value of the reference voltage source. If the deviation exceeds the threshold, it automatically adjusts the gain and offset of the internal ADC. The link attenuation calibration unit controls the main controller to periodically send standard test signals to each subunit. After receiving the signal, the subcontroller feeds back the actual amplitude of the measured data, calculates the attenuation coefficient of each link according to the formula "standard amplitude / actual amplitude", and stores the attenuation coefficient of each link in the attenuation coefficient table. The attenuation coefficient table is used to perform reverse compensation on the collected current and voltage data. The multi-channel calibration unit selects one of the AFE chips as the reference chip, calculates the error value between each AFE chip and the reference chip, and generates a channel error compensation table. The channel error compensation table is used to correct the collected data of each channel. The asynchronous acquisition unit aggregates the data collected by multiple AFE chips in the same subunit into a data group arranged in time sequence. After the sub-controller receives the data group corresponding to a single AFE chip, the single-chip fluctuation analysis unit performs internal data fluctuation analysis on the data group. After the sub-controller receives the data groups of all AFE chips, the alignment unit first aligns the timestamps of different data groups one by one. The synchronization analysis unit uses data from different data groups with the same timestamp as the vertical reference series for synchronization data analysis, analyzes the vertical reference series, and checks whether there are fluctuation differences between the data collected by a single AFE chip and the data from other AFE chips. The reference data feedback unit inputs a set of timestamp-aligned vertical reference data into the evaluation model and obtains its output temporary reference evaluation data. The multi-dimensional diagnostic unit performs multi-dimensional analysis, including: ① whether there are abnormal fluctuations in the data within the same data group; ② whether there are fluctuation differences between the data collected by a single AFE chip and the data from other AFE chips; ③ judging whether the overall data is normal by combining the changes in the temporary reference evaluation data; if any of the above ①②③ is abnormal, it can be determined that the AFE chip is abnormal. The error correction processing unit performs local and overall data fluctuation analysis on the data group of the abnormal AFE chip according to the time series. If only the local data has a different fluctuation pattern than the overall data, it indicates that there is an error in the local data, and the data replacement logic is triggered to replace the erroneous data. If all the data in the data group fluctuates irregularly, it indicates that the AFE chip is faulty, and the operation of the AFE chip is cut off. The communication anomaly monitoring unit checks whether the received data is missing a timestamp or has missing key data; if so, it determines that the communication is abnormal.

[0007] Furthermore, the method for aligning timestamps from different data groups includes the following steps: Preset several time nodes, obtain the timestamp of each data point, and align the timestamp with the time nodes; If alignment is not possible, analyze the position of the timestamp between two adjacent time nodes: Divide the area between two adjacent time nodes into a front alignment area, a prediction area, and a back alignment area in chronological order to determine the timestamp region; If the timestamp is in the front alignment area, then the timestamp will be aligned to the front time node, and the data will remain unchanged; If the timestamp is in the back alignment area, then the timestamp will be aligned to the back-end time node, and the data will remain unchanged; If the timestamp is in the prediction zone, the data at the front-end time node is estimated based on the data corresponding to the timestamp in the previous prediction zone and the data corresponding to the timestamp in the current prediction zone. The time node is then replaced with a timestamp, and the estimated data is used as the data at the new timestamp to construct a new data set.

[0008] Furthermore, the current operating condition coordination module includes a current change monitoring unit and a vehicle operating condition scheduling unit; The wire breakage detection cycle module includes a basic wire breakage detection unit and a cycle adaptive adjustment unit; The multi-dimensional alarm module includes a disconnection level determination unit and an alarm signal transmission unit. The current change monitoring unit collects battery current data in real time, calculates the current change rate and compares it with a preset threshold. When the current change rate is less than the preset threshold, it marks the "current stability window period" and triggers the disconnection detection. When the current change rate is greater than the preset threshold, the detection is paused. The vehicle operating condition scheduling unit receives the operating condition warning signal based on the transient operating condition and immediately sends a pause detection command to the disconnection detection cycle module and records the operating condition duration. After the operating condition ends, it receives the resumption detection command and starts the detection. After receiving a current stabilization window signal or a resumption of detection command, the basic disconnection detection unit starts the disconnection detection process, calculates the voltage difference, and terminates the current detection when it receives a "pause detection command". The cycle adaptive adjustment unit dynamically adjusts the detection cycle based on the battery state evaluation parameters.

[0009] Furthermore, the disconnection level determination unit determines the disconnection level based on multi-feature fusion and feeds back the disconnection detection results to the modeling unit. The disconnection level determination process is as follows: The voltage difference threshold is dynamically adjusted based on the battery state assessment model results. If the voltage difference exceeds the voltage difference threshold, and the current change rate is less than the change rate threshold and the voltage difference between adjacent cells is less than the first adjacent voltage difference threshold, it is judged as "instantaneous interference" and no alarm is triggered. If the voltage difference exceeds the voltage difference threshold, and the current change rate is less than the change rate threshold and the voltage difference between adjacent cells is greater than the second adjacent voltage difference threshold, it is judged as "suspected disconnection" and a secondary detection is initiated. If the secondary detection result is consistent with the initial detection result, it is judged as "actual disconnection" and an alarm is triggered. After receiving the alarm signal from the disconnection level determination unit, the alarm signal transmission unit generates standard alarm information and transmits it to the service terminal.

[0010] Furthermore, the hierarchical communication architecture module includes a sub-unit partitioning unit and a data frame segmentation unit; The fault diagnosis module includes a bidirectional verification unit and a bypass channel unit; The hierarchical automatic recovery module includes a feature classification unit, a real-time self-healing unit, a redundancy switching unit, and a security degradation unit; The sub-unit partitioning unit divides several AFE chips into a sub-unit, defines the sub-unit boundaries and hardware configuration, and connects each sub-unit to the main controller via a high-speed differential bus. The data frame segmentation unit splits the complete data frame containing all AFE chip acquisition information into multiple subframes and sends them to the main controller according to priority. The bidirectional verification unit enables the main controller to periodically send test data frames to each AFE chip. If the AFE chip does not respond, the main controller triggers reverse positioning to lock the abnormal chip. The bypass channel unit is immediately activated when the AFE chip fails. The feature classification unit establishes a fault feature database, collects data in real time, compares it with the fault feature database, and classifies each fault as mild, moderate, or severe. The real-time self-healing unit triggers real-time self-healing for mild faults. The redundancy switching unit starts switching the bypass channel when the fault continues to time out. The safety degradation unit enters safety degradation when the fault cannot self-heal or switch.

[0011] Furthermore, the multi-model evaluation module includes a modeling unit and a model optimization unit; The dynamic acquisition module includes a frequency linkage adjustment unit, a sampling and filtering unit, and a data preprocessing unit. The driving habit quantification module includes an operation intensity unit, an energy consumption unit, an operating condition stability unit, and a scenario classification unit; The aging scenario assessment module constructs a "scenario-aging feature" mapping library, calls the mapping library, and automatically identifies scenarios; The modeling unit builds an electrochemical sub-model and a data-driven sub-model and merges them to obtain a battery state assessment model. The battery state assessment model is used to obtain battery state assessment parameters, and the model optimization unit dynamically adjusts the weights according to the aging scenario.

[0012] Furthermore, the frequency linkage adjustment unit receives driving characteristics, analyzes the operating conditions based on the driving characteristics, and switches the acquisition frequency according to the operating conditions. The sampling and filtering unit samples in layers according to the core layer, auxiliary layer, and redundant layer. The data preprocessing unit determines different frequency modes according to the frequency and performs redundant removal on the stable segments of data generated in the high-frequency mode where the deviation of multiple consecutive sampling data is less than the deviation threshold. The operation intensity unit collects driving characteristics in real time, calculates the average acceleration intensity and instantaneous peak intensity, distinguishes different intensities of braking, and calculates the proportion of different intensities of braking to quantify the operation intensity characteristics. The energy consumption unit analyzes the energy consumption deviation coefficient in real time based on current integral and driving mileage, and counts the vehicle idling time and battery discharge current during idling to quantify the contribution of idling conditions to total energy consumption, thereby constructing energy consumption characteristics. The operating condition stability unit calculates the vehicle speed fluctuation coefficient and counts the switching frequency of discharge-energy recovery-discharge to construct operating condition stability characteristics. The scenario classification unit combines driving characteristics and environmental data to divide the vehicle scenario into several typical dynamic scenarios and clarifies the driving habit characteristics and patterns under each scenario. The battery state parameter correction module dynamically calibrates the battery state assessment parameters based on the driving characteristic data provided by the driving habit quantification module and the current scene determined by the dynamic scene recognition module.

[0013] To address the aforementioned technical problems, the present invention further provides the following technical solution: A battery management method based on an AFE chip includes the following steps: The system is initialized and parameters are configured. Each AFE chip collects voltage and current data from the battery pack, acquires driving characteristics, and performs operating condition analysis. The sampling frequency of the data is adjusted according to the operating condition. The AFE chip's disconnection detection is enabled or disabled based on the current stability and operating conditions, the disconnection detection cycle is dynamically optimized, and an alarm is issued based on the disconnection detection results. The voltage and current data are subjected to anti-interference, parameter compensation, attenuation calibration and multi-channel calibration data correction processing. The corrected data is combined into several data groups according to different AFE chips, and the timestamps of different data groups are aligned. Multi-dimensional anomaly analysis is performed based on the aligned data groups, and data replacement or chip cutting is performed based on the anomaly analysis results. Based on the adaptive optimization of the battery state assessment model for battery aging scenarios, the battery state assessment model is input with aligned timestamp data to output battery state assessment parameters, and the battery state assessment parameters are calibrated according to driving characteristics and the current scenario. Different faults generated by the system are classified, and the fault status is fed back in real time. For minor faults, a self-healing program is initiated.

[0014] Compared with the prior art, the present invention has the following advantages: This invention addresses four interconnected problems in the system: inaccurate AFE sampling, defects in cascaded communication, limitations of the evaluation system, and misjudgment of disconnection detection. Through a collaborative design across the entire data acquisition-transmission-evaluation-protection-control chain, it optimizes the battery management process. By combining filtering, temperature compensation and thermal management, the influence of motor interference on voltage and current signals is eliminated, providing "distortion-free raw data" for subsequent testing, reducing the possibility of misjudgment by the evaluation system and lowering the probability of misdiagnosis of disconnection due to data fluctuations. The cascading process uses real-time dynamic calibration, link error correction, and asynchronous data acquisition and synchronous analysis to reduce cascading attenuation distortion and communication delays, ensuring that the accurate data output from the sampling stage is delivered to the evaluation stage in a "complete and timely" manner, providing "high-quality and high-timeliness" data input for subsequent evaluation stages; It expands the acquisition of in-depth data such as electrode material structure and electrolyte evolution to make up for the one-sidedness of conventional parameters. By subdividing aging scenarios and using multi-model fusion algorithms, it adapts to different driving habits and ambient temperatures, overcomes the limitations of evaluation algorithms, improves algorithm accuracy, and ensures the accuracy of battery status prediction and provides accurate data reference for wire breakage detection. By using "timing optimization + multi-dimensional judgment" to solve the problem of "overlapping detection timing and interference" by using the working condition linkage triggering mechanism to avoid instantaneous high current scenarios such as rapid acceleration, the problem of "overlapping detection timing and interference" is solved. The sampling voltage difference and current change trend are combined with the internal resistance data output from the evaluation stage to avoid misjudgment by a single threshold. This system achieves multiple optimizations and upgrades through end-to-end technology optimization: it achieves accurate judgment of battery status through multi-dimensional data collection and scenario-based algorithms, and its output data provides dynamic thresholds for disconnection detection. Disconnection detection reduces the risk of misjudgment with the support of sampling and evaluation through timing optimization and multi-dimensional judgment. At the same time, the detection results are fed back to the evaluation stage to fill in local data gaps. The multi-stage linkage upgrades the battery management system from "single-point function compliance" to "overall performance reliability", effectively avoiding the risk of overall failure caused by defects in a certain stage of the traditional system. Attached Figure Description

[0015] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0016] Figure 1 A structural block diagram of a battery management system based on an AFE chip is provided for an embodiment of the present invention; Figure 2 This is a schematic diagram of the operation of the data interference resistance compensation subsystem in an embodiment of the present invention; Figure 3 This is a flowchart of the AFE data acquisition subsystem in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the timestamp alignment process in an embodiment of the present invention; Figure 5 This is a schematic diagram of the operation of the disconnection diagnosis subsystem in an embodiment of the present invention; Figure 6 This is a schematic diagram of the layered communication connection in an embodiment of the present invention; Figure 7 This is a schematic diagram of the operation of the BMS evaluation subsystem in an embodiment of the present invention; Figure 8 This is a flowchart of the relevant part of the data substitution logic in an embodiment of the present invention; Figure 9 This is a flowchart of the relevant part of the disconnection level determination logic in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figure 1 As shown, this invention provides a battery management system based on an AFE chip, including a main controller, an AFE data acquisition subsystem, a data anti-interference subsystem, a disconnection diagnosis subsystem, a hierarchical communication subsystem, and a BMS evaluation subsystem. It also includes an equalization module, a charge / discharge control module, a memory, a power supply module, and a service terminal. The equalization module is used to maintain the voltage balance of the battery pack, the charge / discharge control module is used to control the charging and discharging of the battery pack, the memory stores various data and programs, and the power supply module provides power to the system.

[0019] This invention addresses four interconnected problems in the system: inaccurate AFE sampling, defects in cascaded communication, limitations of the evaluation system, and misjudgment of disconnection detection. Through a collaborative design across the entire data acquisition-transmission-evaluation-protection-control chain, it optimizes the battery management process. By combining filtering, temperature compensation and thermal management, the influence of motor interference on voltage and current signals is eliminated, providing "distortion-free raw data" for subsequent testing, reducing the possibility of misjudgment by the evaluation system and lowering the probability of misdiagnosis of disconnection due to data fluctuations. The cascading process uses real-time dynamic calibration, link error correction, and asynchronous data acquisition and synchronous analysis to reduce cascading attenuation distortion and communication delays, ensuring that the accurate data output from the sampling stage is delivered to the evaluation stage in a "complete and timely" manner, providing "high-quality and high-timeliness" data input for subsequent evaluation stages; It expands the acquisition of in-depth data such as electrode material structure and electrolyte evolution to make up for the one-sidedness of conventional parameters. By subdividing aging scenarios and using multi-model fusion algorithms, it adapts to different driving habits and ambient temperatures, overcomes the limitations of evaluation algorithms, improves algorithm accuracy, and ensures the accuracy of battery status prediction and provides accurate data reference for wire breakage detection. By using "timing optimization + multi-dimensional judgment" to solve the problem of "overlapping detection timing and interference" by using the working condition linkage triggering mechanism to avoid instantaneous high current scenarios such as rapid acceleration, the problem of "overlapping detection timing and interference" is solved. The sampling voltage difference and current change trend are combined with the internal resistance data output from the evaluation stage to avoid misjudgment by a single threshold. The data interference mitigation subsystem addresses the noise, fluctuations, and deviations in sampling data caused by high-frequency electromagnetic interference from motors and other equipment in the complex electromagnetic environment of electric vehicles. Through end-to-end control of "interference detection - data optimization - state assurance," it ensures the accuracy of voltage and current sampling data, supporting the reliability of battery charge / discharge state judgment and SOC estimation. Specifically: The data interference immunity compensation subsystem includes an interference detection and identification module, a data optimization and processing module, a temperature compensation module, and a thermal management module. The interference detection and identification module includes an interference feature acquisition unit, an interference threshold determination unit, and an interference scene classification unit; The data optimization and processing module includes a filtering algorithm execution unit and an interference data replacement unit; The temperature compensation module includes a temperature error calibration unit, a real-time temperature acquisition unit, and a sampling data dynamic compensation unit; The interference feature acquisition unit collects voltage and current data sampled by the AFE chip in real time, records the data acquisition timestamp synchronously (with an accuracy of 1ms), and extracts key feature parameters from the voltage and current data, including the magnitude of data mutations (such as the instantaneous change in current) and the frequency components of fluctuations (high-frequency fluctuations are initially identified through the time-domain waveform of the data), providing basic data for subsequent interference judgment. The interference threshold determination unit presets interference feature thresholds for multiple scenarios, and sets a core threshold for the high-frequency operation scenario of electric vehicle motors: Data mutation thresholds: current mutation exceeding 5A within 1ms, voltage mutation exceeding 0.5V within 1ms; High-frequency component proportion threshold: The time domain data is converted into frequency domain data through FFT (Fast Fourier Transform). When the proportion of high-frequency components above 10kHz exceeds 30%, an interference warning is triggered. The system compares the collected feature parameters with the corresponding thresholds in real time. If any threshold condition is met, it is determined that there is interference.

[0020] The interference scene classification unit classifies interference scenes based on differences in interference characteristics, including: 1. High-frequency interference from motor operation (high proportion of high-frequency components, stable frequency abrupt changes). 2. Transient pulse interference in the circuit (short pulse characteristics, large abrupt change amplitude but short duration, <500μs); 3. Interference from multiple devices (complex characteristics, with both high-frequency fluctuations and transient changes). 4. Low-frequency interference scenarios (low-frequency noise when the motor is running at low speed, with high-frequency components accounting for <10%). After classification, the types of interference scenarios are output, providing a basis for matching subsequent data optimization solutions.

[0021] The filtering algorithm execution unit adaptively selects the filtering algorithm based on the classification results of the interference scene: For low-frequency interference scenarios: execute the moving average filtering algorithm, dynamically adjust the sliding window size N (value range 5-10, default 8), calculate the delay time according to the AFE chip sampling frequency (e.g. 1kHz) (the delay is about 8ms when N=8, which meets the 100ms level real-time requirement of the battery management system), take the arithmetic mean of N consecutive sampled values, smooth random low-frequency noise, and eliminate small data fluctuations; For dynamic scene interference (interference scenarios include high-frequency motor operation interference or multi-device superposition interference): Kalman filtering algorithm is executed to establish an equivalent circuit model for battery charging and discharging. Parameters such as battery internal resistance and capacitance are input. Through the Kalman filtering "prediction-update" loop, the theoretical values ​​of current and voltage at the current moment are predicted based on the model. Combined with the actual measured values ​​sampled by AFE, the weights of the predicted and measured values ​​are adjusted according to the noise covariance matrix, and the optimal estimated value is output. This ensures data response speed in dynamic scenarios, with current deviation controlled within 2% and voltage deviation controlled within 1%.

[0022] When the interference detection module determines that there is transient pulse interference in the circuit, the interference data substitution unit triggers the data substitution logic (the relevant part of the data substitution logic of the interference data substitution unit is as follows). Figure 8 (as shown) 1. Call up the valid data from the sampling period before the interference occurred to ensure data continuity, load the battery state assessment model, and estimate the theoretical values ​​of current and voltage at the current moment; 2. A weighted fusion method of "valid data from the previous moment × a + (theoretical values ​​of current and voltage) × b" is adopted (where a and b are weights that can be adjusted according to the actual situation) to generate alternative data and avoid erroneous data from entering subsequent calculations.

[0023] The temperature compensation stage addresses the issue of temperature variations around the AFE chip caused by battery heat generation, which can lead to temperature sampling errors, decreased voltage / current sampling accuracy, and SOH assessment deviations. Through comprehensive control of the entire process—temperature compensation, thermal management, and state verification—the AFE sampling error is kept within acceptable limits (voltage deviation ≤ 0.5%, current deviation ≤ 1%), ensuring the accuracy of battery health status assessment. Specifically: The temperature error calibration unit fits the linear compensation formulas for "temperature-voltage error" and "temperature-current error" using the least squares method [e.g., compensation formula: true value = sampled value - (temperature - 25℃) × temperature coefficient]. Then, it generates a calibration table based on the linear compensation formula and stores it in the memory.

[0024] The real-time temperature acquisition unit collects temperature data from the AFE chip and battery pack in real time. 1. AFE chip temperature: acquired by the temperature sensor built into the AFE chip, with the sampling frequency synchronized with the AFE chip sampling frequency (e.g., 1kHz), and an accuracy of ±0.5℃; 2. Battery pack temperature: Arrange 3 to 5 distributed temperature sensors (NTC thermistors) inside the battery pack to collect the surface and internal temperature of the battery, and take the average value as the representative temperature of the battery pack. The sampling interval is 1 second. The sampling data dynamic compensation unit calls up the real-time temperature value and the pre-stored calibration table before each AFE sampling: based on the current temperature, it substitutes the calibration table to correct the current and voltage data collected by the AFE chip, outputs the compensated voltage and current data, and marks the compensation temperature and compensation amount for easy subsequent traceability.

[0025] The thermal management module sets the battery temperature control range to 25℃~45℃ (optimal operating range). Preheating is triggered below 0℃, and heat dissipation is triggered above 45℃, executing active thermal management actions. 1. Heat dissipation control: When the battery temperature is >45℃, the liquid cooling / air cooling system is activated. In liquid cooling mode, the coolant flow rate is controlled (e.g., 5~10L / min) to remove heat through the liquid cooling plate. In air cooling mode, the fan is turned on (speed 2000~3000rpm) to accelerate air circulation. 2. Preheating control: When the battery temperature is <0℃, start the PTC heater and control the heating power (e.g., 500~1000W) to raise the battery temperature to above 5℃, so as to avoid the impact of low temperature on battery performance and AFE chip sampling.

[0026] The working process of the above data interference compensation subsystem is as follows: Figure 2 As shown.

[0027] The AFE data acquisition subsystem addresses issues such as accuracy deviation, insufficient synchronization, and abnormal interference in data acquisition under AFE cascade scenarios. Through the collaborative operation of multiple modules, it realizes core functions such as data acquisition, dynamic calibration, asynchronous analysis, anomaly handling, and status assessment, ensuring the accuracy of battery status assessment. Specifically, the AFE data acquisition subsystem includes an AFE chip, a current and voltage acquisition module, a data accuracy calibration module, an asynchronous acquisition and analysis module, and an anomaly diagnosis and processing module. The current and voltage acquisition module includes an AFE data acquisition unit and an acquisition parameter configuration unit; The data accuracy calibration module includes a reference voltage calibration unit, a link attenuation calibration unit, and a multi-channel calibration unit; The asynchronous acquisition and analysis module includes an asynchronous acquisition unit, a single-chip fluctuation analysis unit, an alignment unit, and a synchronous analysis unit; The anomaly diagnosis and processing module includes a reference data feedback unit, a multi-dimensional diagnosis unit, an error correction and processing unit, and a communication anomaly monitoring unit. The current and voltage acquisition module, as the source of system data acquisition, is responsible for controlling the AFE chip to accurately acquire the current and voltage data of the battery pack, providing raw data support for subsequent asynchronous analysis, accuracy calibration and status assessment.

[0028] According to the acquisition instructions issued by the service terminal or the main controller, the AFE data acquisition unit controls each AFE chip to connect to the designated cells in the battery pack, and collects the current and voltage signals of the cell pack in real time. The acquired analog signals are transmitted to the ADC (analog-to-digital converter) inside the AFE chip to complete the analog-to-digital conversion and generate digital current and voltage data, providing basic data for the subsequent generation of time series data sets.

[0029] The data acquisition parameter configuration unit presets and configures the acquisition parameters of the AFE chip according to the battery pack type (such as lithium battery, lead-acid battery) and acquisition requirements, including data accuracy (such as ADC sampling bit depth) and acquisition trigger conditions (such as command trigger, timer trigger).

[0030] The data accuracy calibration module dynamically eliminates various errors during the AFE cascade data acquisition process through real-time reference voltage comparison, cascade link attenuation compensation, and multi-channel consistency correction, ensuring that the acquired data accuracy meets usage requirements. Specifically: The reference voltage calibration unit periodically sends reference calibration commands to the AFE chips of each subunit. After receiving the command, the AFE chip compares the collected battery voltage data with the output value of the reference voltage source. If the deviation exceeds the set ±0.5%, it will automatically adjust the gain and offset of the internal ADC (analog-to-digital converter) to correct the acquisition error.

[0031] The link attenuation calibration unit controls the main controller to periodically send standard test signals (amplitude 1V, frequency 1kHz) to each sub-unit. After receiving the signals, the sub-controllers feed back the actual amplitude of the measured data, calculate the attenuation coefficient of each link according to the formula "standard amplitude / actual amplitude", and store each link attenuation coefficient in the attenuation coefficient table for subsequent automatic retrieval. This allows for reverse compensation of the collected current and voltage data, effectively eliminating the error caused by link attenuation.

[0032] The multi-channel calibration unit selects one of the more stable AFE chips as the reference chip (the reference chip can be randomly changed according to the period). Calibration is triggered once per hour - the reference chip and other AFE chips simultaneously collect the voltage data of the same cell, calculate the error value between each AFE chip and the reference chip, generate a channel error compensation table, and then correct the data collected by each channel according to the channel error compensation table to ensure the consistency of multi-channel data and avoid the overall data accuracy from decreasing due to channel differences.

[0033] The asynchronous acquisition and analysis module acquires data from the AFE chip in asynchronous mode (without requiring fully synchronous acquisition). Combined with timestamp alignment and multi-dimensional analysis methods, it achieves synchronous data analysis and simultaneously performs single-chip data fluctuation screening and multi-chip longitudinal synchronization verification. Specifically: The asynchronous acquisition unit aggregates the data collected by multiple AFE chips in the same subunit into data groups arranged in time sequence; it aggregates the data into multiple such data groups in time order, each data group corresponding to a set of current and voltage data of a battery cell (i.e., data collected by an AFE chip at different times), and the group contains current and voltage data at different time points. Each data is accompanied by a corresponding timestamp, and finally the current and voltage data are sorted and arranged in time order.

[0034] After the sub-controller receives the data set corresponding to a single AFE chip, the single-chip fluctuation analysis unit performs internal data fluctuation analysis on the data set (to determine whether there are abnormal fluctuations) and initially screens the stability of the data collected by the chip.

[0035] After the sub-controller receives all the data groups from the AFE chips, the alignment unit first aligns the timestamps of the different data groups one by one. like Figure 4 As shown, the alignment method is as follows: Several time nodes are preset. The timestamp of each data point is obtained, and the timestamp is aligned with the time nodes. If alignment is not possible, the position of the timestamp between two adjacent time nodes is analyzed. The area between two adjacent time nodes is divided into a front alignment zone, a prediction zone, and a back alignment zone according to time sequence. The region of the timestamp is determined. If the timestamp is in the front alignment zone, it is aligned to the front time node, and the data remains unchanged. If the timestamp is in the back alignment zone, it is aligned to the back time node, and the data remains unchanged. If the timestamp is in the prediction zone, the data at the front time node is estimated based on the data corresponding to the timestamp in the previous prediction zone and the data corresponding to the timestamp in the current prediction zone (generally, the arithmetic mean is taken). The time nodes are replaced with timestamps in this way, and the estimated data is used as the data for the new timestamp to construct a new data set. The synchronous analysis unit uses data from different data groups with the same timestamp as the longitudinal reference series for synchronous data analysis. It analyzes the longitudinal reference series to check whether there are fluctuation differences between the data collected by a single AFE chip and the data from other AFE chips. The anomaly diagnosis and processing module combines synchronous analysis results with temporary reference evaluation data to accurately identify data anomaly types (including data and timestamp mismatch, abnormal data fluctuations, chip failures, and communication anomalies), and executes targeted processing strategies to ensure the validity of collected data and the security of system operation. Specifically: The reference data feedback unit inputs a set of timestamp-aligned vertical reference data into the evaluation model and obtains its output temporary reference evaluation data.

[0036] All processes involved in the asynchronous acquisition and analysis module are "synchronous analysis processes".

[0037] The multi-dimensional diagnostic unit performs multi-dimensional analysis based on the analysis results of the synchronous analysis process. The analysis aspects include: ① Are there any abnormal fluctuations in the data within the same data set (i.e., is there a certain pattern to the data change trend over a certain period of time? If not, there are abnormal fluctuations); ② Are there any fluctuation differences between the data collected by a single AFE chip and the data collected by other AFE chips; ③ Combine the changes in the temporary reference evaluation data (if there are significant differences in the temporary reference evaluation data obtained from different longitudinal reference series, it indicates that there may be an anomaly) to determine whether the overall data is normal; If any of the above ①②③ is abnormal, it can be preliminarily determined that the data collected by the AFE chip is abnormal.

[0038] The error correction processing unit performs local and overall data fluctuation analysis on the data sets of abnormal AFE chips according to the time series: If only local data fluctuates differently from the overall data, it indicates that there is an error in the local data, triggering the data replacement logic to replace the erroneous data; if all data in the data group fluctuates irregularly, it indicates that the AFE chip is faulty, and the operation of the AFE chip is cut off.

[0039] The communication anomaly monitoring unit checks whether the received data is missing a timestamp or whether there is a lack of key data (such as current or voltage data not being transmitted). If so, it is determined to be a communication anomaly. At the same time, a real-time monitoring and anomaly recording alarm function has been added to ensure that communication anomalies can be detected and recorded in a timely manner. In the case of communication anomalies, communication can be cut off or wait for a period of time to see if communication is restored.

[0040] After the data anomaly handling is completed (such as replacing erroneous data through interpolation or disconnecting faulty chips), the battery status assessment is carried out based on the corrected and accurate data to ensure the accuracy of the assessment results.

[0041] The working process of the above AFE data acquisition subsystem is as follows: Figure 3 As shown.

[0042] The disconnection diagnosis subsystem is designed to improve the accuracy of disconnection detection of the battery pack sampling line. By dynamically optimizing the detection timing, adaptively adjusting the cycle, and making multi-dimensional judgments, it avoids instantaneous current interference and misjudgment of battery degradation, thus achieving accurate and intelligent disconnection detection.

[0043] Specifically, the disconnection diagnosis subsystem includes a current condition coordination module, a disconnection detection cycle module, and a multi-dimensional judgment and alarm module; The current condition coordination module includes a current change monitoring unit and a vehicle condition scheduling unit; The disconnection detection cycle module includes a basic disconnection detection unit and a cycle adaptive adjustment unit; The multi-dimensional alarm module includes a disconnection level determination unit and an alarm signal transmission unit; The current change monitoring unit collects battery current data in real time, calculates the current change rate (di / dt) and compares it with a preset threshold (5A / ms for conventional batteries, 3A / ms for lithium iron phosphate batteries, etc.). When the absolute value of di / dt is less than the corresponding threshold, it marks the "current stability window" and triggers the disconnection detection. When it is greater than the threshold, it pauses the detection (sends a pause detection command) to avoid instantaneous current interference.

[0044] The vehicle operating condition scheduling unit receives the operating condition warning signal based on the transient operating condition, and immediately sends a pause detection command to the disconnection detection cycle module and records the operating condition duration. After the operating condition ends, it receives the resume detection command and starts the detection, matching the detection timing with the operating condition.

[0045] Basic wire breakage detection unit Function: Upon receiving the "current stabilization window period" signal or "resume detection command", the disconnection detection process is initiated, the AFE chip operates on the pull-up and pull-down current sources of the multi-sampling lines, collects cell voltage data, calculates the voltage difference, and provides basic data for subsequent judgment; upon receiving the "pause detection command", the current detection is terminated to avoid misjudgment due to superposition with instantaneous large current.

[0046] The cycle adaptive adjustment unit dynamically adjusts the detection cycle based on battery state assessment parameters: for example, when SOH≥80% and SOC20%-80%, a normal cycle is set (1 detection every 1 second); when SOH<80% (high internal resistance, weak anti-interference) or SOC<20% or>80% (easy to charge and discharge with high current), the cycle is extended to 1 detection every 3 seconds, and "multi-round verification" is triggered (three consecutive tests are consistent and enter the judgment), balancing efficiency and accuracy, and reducing the false judgment rate of battery degradation.

[0047] The disconnection level determination unit determines the disconnection level based on multiple features (voltage difference from the basic disconnection detection unit, current change rate from the current change monitoring unit, and voltage difference between adjacent cells) (the disconnection level determination logic is as follows). Figure 9 (As shown), and the broken wire detection results are fed back to the modeling unit: 1. Dynamically adjust the voltage difference threshold based on the battery state assessment model evaluation results (battery internal resistance, etc.); 2. If the voltage difference exceeds the voltage difference threshold, but the current change rate is <5A / ms and the voltage difference between adjacent cells is <50mV, it is judged as "transient interference" and no alarm is triggered. If the voltage difference exceeds the voltage difference threshold, and the current change rate is <5A / ms, and the voltage difference between adjacent cells is >100mV, it is judged as "suspected open circuit" and a secondary detection is initiated. 3. If the secondary detection still meets the criteria of "voltage difference exceeding voltage difference threshold", "voltage difference between adjacent cells > 100mV" and "current stable", it is judged as "actual disconnection" and an alarm is triggered. This unit can distinguish between false current interference differences and real wire breakage anomalies, with an accuracy rate of over 98% in battery degradation scenarios.

[0048] After receiving the alarm signal from the disconnection level determination unit, the alarm signal transmission unit generates standard alarm information (including disconnection sampling line number, cell location, determination basis, etc.) and transmits it to the service terminal, triggering local audible and visual alarms (fault light on, buzzer prompt), allowing drivers and maintenance personnel to check the fault in a timely manner.

[0049] The working process of the above-mentioned disconnection diagnosis subsystem is as follows: Figure 5 As shown.

[0050] The hierarchical communication subsystem addresses issues in battery management such as communication latency, difficulty in locating communication faults in AFE chips, and reliance on manual recovery. Through a hierarchical architecture, graded diagnostic system, and automatic recovery strategy, it improves data transmission efficiency, achieves accurate fault location and rapid self-healing, reduces safety hazards in unmanned scenarios, and ensures the stable operation of the battery management system.

[0051] Specifically, the hierarchical communication subsystem includes a hierarchical communication architecture module, a fault diagnosis module, and a hierarchical automatic recovery module; The layered communication architecture module includes a sub-unit partitioning unit and a data frame segmentation unit; The fault diagnosis module includes a two-way verification unit and a bypass channel unit; The graded automatic recovery module includes a feature classification unit, a real-time self-healing unit, a redundancy switching unit, and a security degradation unit; like Figure 6 As shown, the sub-unit partitioning unit divides the large-scale battery pack into sub-units of 3 to 15 AFE chips, configures a sub-controller for each sub-unit, clarifies the sub-unit boundaries and hardware configuration, and connects each sub-unit using a high-speed differential bus. Within a sub-unit, low-speed data interaction (such as voltage acquisition) is realized between AFE chips, and only summary data (average voltage, fault status) is transmitted between sub-units to reduce the amount of data across units. The latency within a sub-unit is controlled within 3μs, the rate between sub-units is increased to 8Mbps, and the overall latency is reduced by more than 60%. All sub-units are connected to the main controller.

[0052] The data frame segmentation unit splits the complete data frame containing all AFE chip acquisition information into multiple subframes. Each subframe contains only the key data (voltage, current) of the AFE chip. The main controller or sub-controller receives the subframes according to priority and processes abnormal subframes first, reducing the single frame transmission time from 800ns to 350ns.

[0053] The bidirectional verification unit enables the main controller to periodically send test data frames to each AFE chip. The AFE chip needs to return a response frame with a verification code. If the chip does not respond or the verification fails, the main controller triggers reverse positioning, indirectly forwarding data through adjacent AFE chips to confirm whether there is a hidden fault of "one-way pass, reverse break" and quickly lock the abnormal chip. The bypass channel unit is activated immediately when the AFE chip fails, temporarily skipping the faulty chip and maintaining overall communication through the data forwarding function of the adjacent AFE chip. At the same time, it records the faulty chip ID and triggers the repair process when the system is idle to avoid communication interruption.

[0054] The feature classification unit establishes a fault feature library containing "communication frequency fluctuations, data frame error types, signal amplitude attenuation, and sudden changes in bus load rate". It collects data in real time, compares it with the fault feature library, and classifies each fault as mild, moderate, or severe.

[0055] The real-time self-healing unit triggers real-time self-healing for minor faults such as transient interference and slight delays. For example, when the AFE chip experiences a single delay, the main controller or sub-controller sends a "resynchronization command" to reset the timing, etc. The recovery time is ≤100ms, requiring no manual intervention and ensuring uninterrupted parameter acquisition.

[0056] The redundancy switching unit initiates the switching bypass channel when the fault persists for more than 1 second (such as when the chip fails to respond for 3 consecutive times or the node continuously sends incorrect frames); if it is a software error, it automatically issues a "parameter reset command" to attempt recovery and reduce the scope of the fault's impact.

[0057] When a fault cannot be self-healed or switched over (such as hardware damage or bus short circuit), the safety degradation unit enters safety degradation mode: it cuts off the power supply to the faulty node to prevent its spread, limits the charging and discharging power to below 50% of the rated value, sends a "fault alarm and corresponding location" to the service terminal, and starts the backup power supply to maintain core monitoring and prevent battery runaway.

[0058] The BMS assessment subsystem aims to address the issues of incomplete data, insufficient collection frequency and accuracy in traditional BMS systems through scenario-based precise assessment, multi-model fusion calculation, and dynamic data collection optimization. It achieves accurate assessment of battery status and remaining lifespan, adapting to complex in-vehicle conditions and individual battery differences. Furthermore, it solves the problem of current in-vehicle BMS systems lagging in tracking dynamic changes in driving habits. Through multi-module collaborative work, it achieves accurate quantification of driving habits, real-time identification of dynamic scenarios, and adaptive adjustment of SOC prediction. Ultimately, it improves the driver's accuracy in judging vehicle range, alleviates battery anxiety, and prevents vehicles from breaking down due to depleted battery power.

[0059] Specifically, the BMS evaluation subsystem includes an aging scenario evaluation module, a multi-model evaluation module, a dynamic acquisition module, a driving habit quantification module, and a battery state parameter correction module. The multi-model evaluation module includes a modeling unit and a model optimization unit; The dynamic acquisition module includes a frequency linkage adjustment unit, a sampling filtering unit, and a data preprocessing unit; The driving habit quantification module includes an operation intensity unit, an energy consumption unit, an operating condition stability unit, and a scenario classification unit; The aging scenario assessment module constructs a "scenario-aging feature" mapping library, calls the mapping library, and automatically identifies scenarios: vehicle scenarios are divided into multiple typical scenarios, including high temperature and high load (temperature > 35℃, discharge rate > 1C, such as long-distance highway driving in summer), low temperature and low load (temperature < -10℃, discharge rate < 0.5C, such as short-distance commuting in winter), and frequent fast charging (fast charging once a day, charging rate > 1.5C), etc. Boundary parameters are defined, and aging feature data of different scenarios are collected. For example, in the high temperature and high load scenario, "internal resistance increases by 1%, SEI film thickness increases by 5nm, and capacity decreases by 0.8%", and in the low temperature and low load scenario, "internal resistance increases by 1%, corresponding to lithium deposition increases by 2μg / cm², and capacity decreases by 1.2%". The data is entered into the "scenario-aging feature" mapping library, and aging feature data such as "ambient temperature, charge and discharge rate, driving time, and mileage" are received. The scenario is identified according to the preset logic ("scenario-aging feature" mapping library).

[0060] The multi-model evaluation module combines the interpretability of the physical mechanism of the electrochemical model with the adaptability of the data-driven model to complex scenarios. Through dynamic weight adjustment and online learning, it improves the accuracy and generalization of battery capacity decay and SOH evaluation, and adapts to the limited computing power of the vehicle MCU.

[0061] Specifically: The modeling process for the modeling unit is as follows: 1. Build an electrochemical sub-model: Based on the simplified single-particle model (SPM), characterize the aging mechanism of SEI film growth, lithium deposition, etc., input "internal resistance, expansion force, temperature" (adding depth data), and output "theoretical capacity decay trend"; 2. Build a data-driven sub-model: Use the lightweight GBDT algorithm (adapted to vehicle MCU, training time <10 seconds), input "voltage, current, number of cycles", output "actual capacity decay correction value"; 3. Fusion Output: A battery state assessment model is constructed using a weighted algorithm (60% electrochemistry and 40% data-driven by default, which can be dynamically adjusted). Battery state assessment parameters are obtained based on the battery state assessment model, avoiding the high computational power consumption of full-order electrochemistry and making up for the "black box" defects of data-driven models.

[0062] The model optimization unit dynamically adjusts the weights according to the aging scenario to achieve adaptation to the scenario and optimization of the model.

[0063] The dynamic acquisition module dynamically adjusts the data acquisition frequency based on vehicle operating conditions and driving characteristics. It combines hierarchical sampling and data filtering to solve the problems of insufficient acquisition frequency and data redundancy in traditional BMS, and balances accuracy and power consumption.

[0064] The frequency linkage adjustment unit receives driving characteristics and switches the acquisition frequency based on these characteristics: it acquires driving characteristics such as acceleration opening, braking status, and vehicle speed through a high-speed differential bus, determines different operating conditions, and adapts the frequency accordingly: For stable operating conditions (vehicle speed 30~60km / h, acceleration <20%), it uses the base frequency (current / voltage 500Hz, temperature 10Hz, adjusted as needed); for transient operating conditions (acceleration >80% / emergency braking, vehicle speed change >10km / h·s) (this unit sends a condition warning signal to the vehicle operating condition scheduling unit based on the transient condition), it uses a high frequency (current / voltage 2kHz, temperature 50Hz), adjusted as needed; for static operating conditions (engine off, current <10mA), it uses a low-power frequency (current / voltage 10Hz, temperature 1Hz, adjusted as needed), reducing power consumption from 10mA to below 1mA (adapting to static <100μA). Pre-trigger mechanism: Based on 100,000+ sets of "acceleration degree-current" data, a pre-identification model is built. If the acceleration from 20% to 60% is detected in 0.3 seconds, a large current is predicted 0.5 seconds later, and high frequency is started 0.2 seconds in advance to avoid lag.

[0065] The sampling and filtering unit samples in layers: core layer, auxiliary layer, and redundancy layer. The core layer (current, voltage) has a minimum sampling frequency of 500Hz. The auxiliary layer (temperature, expansion force) has a minimum transient frequency of 50Hz and a minimum stable frequency of 10Hz. The redundancy layer (humidity, equalization current) is triggered by events (minimum 10Hz when humidity > 80% / equalization current > 100mA). The data preprocessing unit determines different frequency modes (such as high frequency, low frequency, and medium frequency) based on the frequency. For data generated in the high frequency mode, redundant segments with a deviation of less than the deviation threshold from multiple consecutive sampling data (the number of times can be set according to the actual situation) are removed.

[0066] The driving habit quantification module constructs a feature system from multiple dimensions and realizes real-time calculation and updating of features, completing the transformation of driving habits from "macro-classification" to "micro-dynamic characterization", providing accurate feature data support for subsequent scene recognition and SOC prediction adjustment.

[0067] The driving habit quantification module is responsible for establishing a multi-dimensional feature system that includes "operational intensity, energy consumption, and operational stability" to comprehensively and meticulously depict driving habits, as detailed below: The operational intensity unit collects driving characteristics in real time via a high-speed differential bus, such as accelerator pedal opening (0%-100%), opening change rate, brake pedal travel (0%-100%), and braking frequency (number of brakes per kilometer). It calculates "average acceleration intensity" (the percentage of time within each 10-minute period where the accelerator pedal opening is >50%) and "instantaneous peak intensity" (the maximum opening of the accelerator pedal in a single instance and the corresponding current change rate). It distinguishes between "light braking" (travel <30%), "medium braking" (30%-60%), and "emergency braking" (>60%), and calculates the percentage of braking at different intensities to quantify operational intensity characteristics.

[0068] Energy Consumption Unit: Based on current integration and driving mileage, it calculates the "current energy consumption per 100 kilometers" (e.g., 15 kWh / 100 km) in real time and compares it with the average energy consumption of the same model (e.g., 12 kWh / 100 km) to obtain the "energy consumption deviation coefficient" (e.g., 1.25, reflecting the energy consumption level of the current driving habits); it also counts the vehicle idling time (e.g., idling accounts for 30% during morning and evening rush hours) and the battery discharge current during idling (e.g., 5A when the air conditioner is on and 1A when it is off), quantifying the contribution of idling conditions to total energy consumption, thereby constructing energy consumption characteristics.

[0069] Operating condition stability unit: It uses vehicle speed sensors to collect the number of times the vehicle speed changes every 10 seconds (e.g., 2 changes in smooth traffic conditions and 8 changes in congested traffic conditions) and calculates the "vehicle speed fluctuation coefficient"; it also counts the number of times "discharge-energy recovery-discharge" switches per kilometer (e.g., 10 switches when there is frequent rapid acceleration / brake and 2 switches when driving smoothly) to reflect the impact of operating conditions on the battery charge and discharge cycle, and thus constructs operating condition stability characteristics.

[0070] The scenario classification unit combines two types of data: driving characteristics (operation intensity characteristics, energy consumption characteristics, and operating condition stability characteristics) and environmental data (which can be obtained by image sensors, etc.) to divide in-vehicle scenarios into several typical dynamic scenarios and clarify the driving habit characteristics and patterns under each scenario.

[0071] The battery status parameter correction module dynamically calibrates the battery status assessment parameters based on the feature data provided by the driving habit quantification module and the current scene determined by the dynamic scene recognition module. This ensures that the battery status prediction continuously matches the actual situation and provides the driver with accurate range information.

[0072] By combining driving characteristics and environmental data, the mapping relationship between "SOC-driving range" (SOC is a battery state parameter) is adjusted in real time to adjust the battery state parameters and ensure that the driver obtains accurate remaining driving range. The specific correction process is as follows: The system adjusts the scenario range coefficient according to different scenarios. Based on the current SOC value (output by the battery state assessment model), the vehicle's rated range, and the current scenario range coefficient, the remaining range is calculated in real time according to the formula "range = current SOC × rated range × scenario range coefficient". The displayed data is refreshed every 10 seconds to ensure that the driver can obtain the latest and most accurate range information in a timely manner, avoiding battery anxiety and vehicle breakdown caused by misjudgment of range.

[0073] The main working process of the above-mentioned BMS evaluation subsystem is as follows: Figure 7 As shown.

[0074] In addition, the present invention also provides a battery management method based on an AFE chip, comprising the following steps: The system is initialized and parameters are configured. Each AFE chip collects voltage and current data from the battery pack, acquires driving characteristics, and performs operating condition analysis. The sampling frequency of the data is adjusted according to the operating condition. The AFE chip's disconnection detection is enabled or disabled based on the current stability and operating conditions, the disconnection detection cycle is dynamically optimized, and an alarm is issued based on the disconnection detection results. The voltage and current data are subjected to anti-interference, parameter compensation, attenuation calibration and multi-channel calibration data correction processing. The corrected data is combined into several data groups according to different AFE chips, and the timestamps of different data groups are aligned. Multi-dimensional anomaly analysis is performed based on the aligned data groups, and data replacement or chip cutting is performed based on the anomaly analysis results. Based on the adaptive optimization of the battery state assessment model for battery aging scenarios, the battery state assessment model is input with aligned timestamp data to output battery state assessment parameters, and the battery state assessment parameters are calibrated according to driving characteristics and the current scenario. Different faults generated by the system are classified, and the fault status is fed back in real time. For minor faults, a self-healing program is initiated.

[0075] Two implementation methods are provided for the above solutions: Example 1: New energy passenger vehicle scenario (500km range, 70kWh: ternary lithium battery pack) The system uses 16 TI BQ79616 AFE chips paired with 128GB of on-board memory and a CAN bus communication architecture. Targeting the high-frequency start-stop and dynamic operating conditions of passenger vehicles, the AFE chip sampling frequency is initially set to 20kHz, and can be dynamically adjusted according to driving conditions (such as dropping to 12kHz at a constant speed of 60km / h). A multi-level data calibration strategy of "reference calibration + link attenuation compensation + channel error correction" is adopted. The communication architecture is divided into layers according to "4 chips / sub-unit", prioritizing the transmission of voltage data to ensure real-time data during driving. After multi-level calibration and anti-interference processing, the voltage acquisition error is controlled within ±1mV, and the data acquisition accuracy reaches 99.8%, providing accurate data support for SOC calculation. After SOC calibration, the error is only 0.5% (from 80% to 80.5%). The anomaly diagnosis response time is <1s, which can accurately identify local data anomalies of a single chip (such as an abnormal value of 1.25V) and complete the replacement. The disconnection detection cycle is dynamically adjusted according to SOC (5s / time when SOC>50%), and there is no disconnection alarm throughout the process. Based on driving habit quantification (gentle operation, low energy consumption) and scenario adaptation (urban constant speed scenario), the charging and discharging control is optimized (charging current ≤10A, discharging current ≤20A), effectively reducing battery loss.

[0076] Example 2: New energy commercial vehicle scenario (10-ton load capacity, 120kWh battery pack) Equipped with 256GB of memory and CAN-FD bus, it adopts vibration filtering algorithm to deal with interference from bumpy roads, combined with the "cell balancing + fan cooling" collaborative thermal management strategy, and adopts a hierarchical architecture to prioritize the transmission of voltage data, adapting to the safety monitoring needs of commercial vehicles. Under complex operating conditions such as frequent acceleration under heavy loads and bumpy roads, the data acquisition frequency is adjusted in real time, and the voltage data fluctuation is controlled within ±1.5mV. It can accurately quantify driving habits (acceleration intensity 0.8m / s², heavy braking rate 20%), with a scene recognition accuracy of 100%, SOC calibration error of 0.3%, and voltage difference controlled within 0.008V (threshold 0.01V). The thermal management system starts fan cooling at 30℃, and the charging and discharging current is limited to ≤30A to avoid overload risk. Abnormal data identification and replacement time is <1s, with no hardware failures. The acquisition accuracy reaches 99.6%, meeting the endurance and safety requirements of commercial vehicles operating at high intensity every day.

[0077] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.

Claims

1. A battery management system based on an AFE chip, characterized in that, It includes a main controller, an AFE data acquisition subsystem, a data interference suppression subsystem, a line disconnection diagnosis subsystem, a hierarchical communication subsystem, a BMS evaluation subsystem, as well as an equalization module, a charge and discharge control module, a memory, a power supply module, and a service terminal; The AFE data acquisition subsystem includes an AFE chip, a current and voltage acquisition module, a data accuracy calibration module, an asynchronous acquisition and analysis module, and an anomaly diagnosis and processing module. The line breakage diagnosis subsystem includes a current condition coordination module, a line breakage detection cycle module, and a multi-dimensional judgment and alarm module. The hierarchical communication subsystem includes a hierarchical communication architecture module, a fault diagnosis module, and a hierarchical automatic recovery module; The BMS evaluation subsystem includes an aging scenario evaluation module, a multi-model evaluation module, a dynamic acquisition module, a driving habit quantification module, and a battery status parameter correction module.

2. The battery management system based on the AFE chip according to claim 1, characterized in that, The data interference immunity compensation subsystem includes an interference detection and identification module, a data optimization and processing module, a temperature compensation module, and a thermal management module. The interference detection and identification module includes an interference feature acquisition unit, an interference threshold determination unit, and an interference scene classification unit. The data optimization processing module includes a filtering algorithm execution unit and an interference data replacement unit; The temperature compensation module includes a temperature error calibration unit, a real-time temperature acquisition unit, and a sampling data dynamic compensation unit. The interference feature acquisition unit collects voltage and current data sampled by the AFE chip in real time, records the data acquisition timestamp synchronously, and extracts key feature parameters from the voltage and current data. The interference threshold determination unit presets multiple scene interference feature thresholds and determines whether interference exists. The interference scene classification unit classifies interference scenes according to the differences in interference features. The filtering algorithm execution unit adaptively selects the filtering algorithm based on the interference scene classification results. The interference data replacement unit triggers the data replacement logic when the interference detection module determines that there is transient pulse interference in the circuit. The temperature error calibration unit fits the linear compensation formulas for "temperature-voltage error" and "temperature-current error", generates a calibration table and stores it in the memory. The real-time temperature acquisition unit collects the temperature data of the AFE chip and the battery pack in real time. Before each AFE sampling, the sampling data dynamic compensation unit calls the real-time temperature data and the pre-stored calibration table to correct the current and voltage data collected by the AFE chip and outputs the compensated voltage and current data. The thermal management module performs active thermal management actions.

3. The battery management system based on the AFE chip according to claim 2, characterized in that, The current and voltage acquisition module includes an AFE data acquisition unit and an acquisition parameter configuration unit; The data accuracy calibration module includes a reference voltage calibration unit, a link attenuation calibration unit, and a multi-channel calibration unit. The asynchronous acquisition and analysis module includes an asynchronous acquisition unit, a single-chip fluctuation analysis unit, an alignment unit, and a synchronization analysis unit. The anomaly diagnosis and processing module includes a reference data feedback unit, a multi-dimensional diagnosis unit, an error correction processing unit, and a communication anomaly monitoring unit. The AFE data acquisition unit controls each AFE chip to transmit the acquired analog signal to the AFE chip for analog-to-digital conversion according to the acquisition command issued by the service terminal. The acquisition parameter configuration unit presets and configures the acquisition parameters of the AFE chip according to the battery pack type and acquisition requirements. The reference voltage calibration unit periodically sends reference calibration commands to the AFE chips of each subunit. After receiving the command, the AFE chip compares the collected battery voltage data with the output value of the reference voltage source. If the deviation exceeds the threshold, it automatically adjusts the gain and offset of the internal ADC. The link attenuation calibration unit controls the main controller to periodically send standard test signals to each subunit. After receiving the signal, the subcontroller feeds back the actual amplitude of the measured data, calculates the attenuation coefficient of each link according to the formula "standard amplitude / actual amplitude", and stores the attenuation coefficient of each link in the attenuation coefficient table. The attenuation coefficient table is used to perform reverse compensation on the collected current and voltage data. The multi-channel calibration unit selects one of the AFE chips as the reference chip, calculates the error value between each AFE chip and the reference chip, and generates a channel error compensation table. The channel error compensation table is used to correct the collected data of each channel. The asynchronous acquisition unit aggregates the data collected by multiple AFE chips in the same subunit into a data group arranged in time sequence. After the sub-controller receives the data group corresponding to a single AFE chip, the single-chip fluctuation analysis unit performs internal data fluctuation analysis on the data group. After the sub-controller receives the data groups of all AFE chips, the alignment unit first aligns the timestamps of different data groups one by one. The synchronization analysis unit uses data from different data groups with the same timestamp as the vertical reference series for synchronization data analysis, analyzes the vertical reference series, and checks whether there are fluctuation differences between the data collected by a single AFE chip and the data from other AFE chips. The reference data feedback unit inputs a set of timestamp-aligned vertical reference data into the evaluation model and obtains its output temporary reference evaluation data. The multi-dimensional diagnostic unit performs multi-dimensional analysis, including: ① whether there are abnormal fluctuations in the data within the same data group; ② whether there are fluctuation differences between the data collected by a single AFE chip and the data from other AFE chips; ③ judging whether the overall data is normal by combining the changes in the temporary reference evaluation data; if any of the above ①②③ is abnormal, it can be determined that the AFE chip is abnormal. The error correction processing unit performs local and overall data fluctuation analysis on the data group of the abnormal AFE chip according to the time series. If only the local data has a different fluctuation pattern than the overall data, it indicates that there is an error in the local data, and the data replacement logic is triggered to replace the erroneous data. If all the data in the data group fluctuates irregularly, it indicates that the AFE chip is faulty, and the operation of the AFE chip is cut off. The communication anomaly monitoring unit checks whether the received data is missing a timestamp or has missing key data; if so, it determines that the communication is abnormal.

4. The battery management system based on the AFE chip according to claim 3, characterized in that, Aligning timestamps from different data groups involves the following steps: Preset several time nodes, obtain the timestamp of each data point, and align the timestamp with the time nodes; If alignment is not possible, analyze the position of the timestamp between two adjacent time nodes: Divide the area between two adjacent time nodes into a front alignment area, a prediction area, and a back alignment area in chronological order to determine the timestamp region; If the timestamp is in the front alignment area, then the timestamp will be aligned to the front time node, and the data will remain unchanged; If the timestamp is in the back alignment area, then the timestamp will be aligned to the back-end time node, and the data will remain unchanged; If the timestamp is in the prediction zone, the data at the front-end time node is estimated based on the data corresponding to the timestamp in the previous prediction zone and the data corresponding to the timestamp in the current prediction zone. The time node is then replaced with a timestamp, and the estimated data is used as the data at the new timestamp to construct a new data set.

5. The battery management system based on the AFE chip according to claim 3, characterized in that, The current condition coordination module includes a current change monitoring unit and a vehicle condition scheduling unit. The wire breakage detection cycle module includes a basic wire breakage detection unit and a cycle adaptive adjustment unit; The multi-dimensional alarm module includes a disconnection level determination unit and an alarm signal transmission unit. The current change monitoring unit collects battery current data in real time, calculates the current change rate and compares it with a preset threshold. When the current change rate is less than the preset threshold, it marks the "current stability window period" and triggers the disconnection detection. When the current change rate is greater than the preset threshold, the detection is paused. The vehicle operating condition scheduling unit receives the operating condition warning signal based on the transient operating condition and immediately sends a pause detection command to the disconnection detection cycle module and records the operating condition duration. After the operating condition ends, it receives the resumption detection command and starts the detection. After receiving a current stabilization window signal or a resumption of detection command, the basic disconnection detection unit starts the disconnection detection process, calculates the voltage difference, and terminates the current detection when it receives a "pause detection command". The cycle adaptive adjustment unit dynamically adjusts the detection cycle based on the battery state evaluation parameters.

6. The battery management system based on the AFE chip according to claim 5, characterized in that, The disconnection level determination unit determines the disconnection level based on multi-feature fusion and feeds back the disconnection detection results to the modeling unit. The disconnection level determination process is as follows: The voltage difference threshold is dynamically adjusted based on the battery state assessment model results. If the voltage difference exceeds the voltage difference threshold, and the current change rate is less than the change rate threshold and the voltage difference between adjacent cells is less than the first adjacent voltage difference threshold, it is judged as "instantaneous interference" and no alarm is triggered. If the voltage difference exceeds the voltage difference threshold, and the current change rate is less than the change rate threshold and the voltage difference between adjacent cells is greater than the second adjacent voltage difference threshold, it is judged as "suspected disconnection" and a secondary detection is initiated. If the secondary detection result is consistent with the initial detection result, it is judged as "actual disconnection" and an alarm is triggered. After receiving the alarm signal from the disconnection level determination unit, the alarm signal transmission unit generates standard alarm information and transmits it to the service terminal.

7. The battery management system based on an AFE chip according to claim 6, characterized in that, The hierarchical communication architecture module includes a sub-unit partitioning unit and a data frame segmentation unit; The fault diagnosis module includes a bidirectional verification unit and a bypass channel unit; The hierarchical automatic recovery module includes a feature classification unit, a real-time self-healing unit, a redundancy switching unit, and a security degradation unit; The sub-unit partitioning unit divides several AFE chips into a sub-unit, defines the sub-unit boundaries and hardware configuration, and connects each sub-unit to the main controller via a high-speed differential bus. The data frame segmentation unit splits the complete data frame containing all AFE chip acquisition information into multiple subframes and sends them to the main controller according to priority. The bidirectional verification unit enables the main controller to periodically send test data frames to each AFE chip. If the AFE chip does not respond, the main controller triggers reverse positioning to lock the abnormal chip. The bypass channel unit is immediately activated when the AFE chip fails. The feature classification unit establishes a fault feature database, collects data in real time, compares it with the fault feature database, and classifies each fault as mild, moderate, or severe. The real-time self-healing unit triggers real-time self-healing for mild faults. The redundancy switching unit starts switching the bypass channel when the fault continues to time out. The safety degradation unit enters safety degradation when the fault cannot self-heal or switch.

8. The battery management system based on the AFE chip according to claim 7, characterized in that, The multi-model evaluation module includes a modeling unit and a model optimization unit; The dynamic acquisition module includes a frequency linkage adjustment unit, a sampling and filtering unit, and a data preprocessing unit. The driving habit quantification module includes an operation intensity unit, an energy consumption unit, an operating condition stability unit, and a scenario classification unit; The aging scenario assessment module constructs a "scenario-aging feature" mapping library, calls the mapping library, and automatically identifies scenarios; The modeling unit builds an electrochemical sub-model and a data-driven sub-model and merges them to obtain a battery state assessment model. The battery state assessment model is used to obtain battery state assessment parameters, and the model optimization unit dynamically adjusts the weights according to the aging scenario.

9. The battery management system based on an AFE chip according to claim 8, characterized in that, The frequency linkage adjustment unit receives driving characteristics, analyzes the operating conditions based on driving characteristics, and switches the acquisition frequency according to the operating conditions. The sampling and filtering unit samples in layers according to the core layer, auxiliary layer, and redundant layer. The data preprocessing unit determines different frequency modes according to the frequency and performs redundant removal on the stable segments of data generated in the high-frequency mode where the deviation of multiple consecutive sampling data is less than the deviation threshold. The operation intensity unit collects driving characteristics in real time, calculates the average acceleration intensity and instantaneous peak intensity, distinguishes different intensities of braking, and calculates the proportion of different intensities of braking to quantify the operation intensity characteristics. The energy consumption unit analyzes the energy consumption deviation coefficient in real time based on current integral and driving mileage, and counts the vehicle idling time and battery discharge current during idling to quantify the contribution of idling conditions to total energy consumption, thereby constructing energy consumption characteristics. The operating condition stability unit calculates the vehicle speed fluctuation coefficient and counts the switching frequency of discharge-energy recovery-discharge to construct operating condition stability characteristics. The scenario classification unit combines driving characteristics and environmental data to divide the vehicle scenario into several typical dynamic scenarios and clarifies the driving habit characteristics and patterns under each scenario. The battery state parameter correction module dynamically calibrates the battery state assessment parameters based on the driving characteristic data provided by the driving habit quantification module and the current scene determined by the dynamic scene recognition module.

10. A method for applying the battery management system based on an AFE chip as described in any one of claims 1 to 9, characterized in that, Includes the following steps: The system is initialized and parameters are configured. Each AFE chip collects voltage and current data from the battery pack, acquires driving characteristics, and performs operating condition analysis. The sampling frequency of the data is adjusted according to the operating condition. The AFE chip's disconnection detection is enabled or disabled based on the current stability and operating conditions, the disconnection detection cycle is dynamically optimized, and an alarm is issued based on the disconnection detection results. The voltage and current data are subjected to anti-interference, parameter compensation, attenuation calibration and multi-channel calibration data correction processing. The corrected data is combined into several data groups according to different AFE chips, and the timestamps of different data groups are aligned. Multi-dimensional anomaly analysis is performed based on the aligned data groups, and data replacement or chip cutting is performed based on the anomaly analysis results. Based on the adaptive optimization of the battery state assessment model for battery aging scenarios, the battery state assessment model is input with aligned timestamp data to output battery state assessment parameters, and the battery state assessment parameters are calibrated according to driving characteristics and the current scenario. Different faults generated by the system are classified, and the fault status is fed back in real time. For minor faults, a self-healing program is initiated.

Images