Battery thermal management method, battery thermal management system, and battery

Abstract

This invention relates to the field of battery technology, providing a battery thermal management method, a battery thermal management system, and a battery. It introduces a real-time SOH (State of Health) assessment method based on the fusion of multi-source data on capacity and internal resistance, enabling precise perception of the battery's "health age." This transforms the thermal management system from simple control to personalized cooling, significantly extending battery life. The method dynamically updates the battery's electrothermal coupling model parameters using real-time cell data, ensuring a high degree of consistency between the simulation model and the physical battery state. By using future operating conditions as predictive input, the cooling system is upgraded from "post-event response" to "pre-event prevention," effectively avoiding temporary cell overheating caused by the lag in the temperature control system. A three-dimensional MAP (Potential Analysis Map) with SOH-heat generation rate curve-temperature as input is designed as the control basis, integrating the battery's long-term health, immediate load, and current state for more comprehensive decision-making, enabling synergistic optimization control between safety, lifespan, and energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of battery technology, specifically to a battery thermal management method, a battery thermal management system, and a battery. Background Technology

[0002] Currently, battery thermal management systems in the energy storage and electric vehicle fields generally adopt feedback control strategies based on fixed temperature thresholds. For example, when the battery temperature is higher than 45°C, a fan or water pump is activated, and when the temperature is lower than 10°C, heating is activated. The cooling / heating intensity is usually adjusted by a PID (Proportional Integral Derivative) algorithm based on the difference between the current temperature and the set threshold.

[0003] The response mechanism of the aforementioned thermal management methods suffers from lag. Pure feedback control only activates after the temperature has already risen, failing to prevent excessive temperature increases, which is detrimental to battery life and safety. Furthermore, these methods do not consider differences in battery health, applying the same strategy to all batteries in the same condition. Therefore, for new batteries, an overly conservative cooling strategy wastes energy; for aging batteries, with increased internal resistance and heat generation, a fixed cooling intensity may be insufficient to effectively suppress temperature rise, thus accelerating aging and creating a vicious cycle that hinders effective extension of product lifespan. Summary of the Invention

[0004] To address the aforementioned technical problems, a battery thermal management method is proposed in the first aspect of this invention.

[0005] A second aspect of the present invention provides a battery thermal management system.

[0006] The technical solution adopted in this invention is as follows:

[0007] An embodiment of the first aspect of the present invention proposes a battery thermal management method, comprising the following steps: real-time acquisition of battery voltage, current, and temperature; obtaining the battery SOH (State of Health) calculated by the BMS (Battery Management System). The battery's state of health (SOH) is calculated based on its capacity. The instantaneous internal resistance is calculated using the voltage jump value during charging and discharging, and temperature compensation is applied to this instantaneous internal resistance. The SOH based on the compensated instantaneous internal resistance is then calculated. The reliability of the capacity-based SOH and the internal resistance-based SOH is assessed based on the collected voltage, current, and temperature. The capacity-based SOH and the internal resistance-based SOH are then fused based on the reliability assessment results to generate the final SOH. The parameters of the corresponding electrothermal coupling model are periodically updated based on the collected voltage, current, and temperature. Future work plans and the timing characteristics of the current battery state are obtained and input into the updated electrothermal coupling model and the trained neural network model to obtain the battery's heat generation rate. The confidence level of the heat generation rate output by the electrothermal coupling model and the neural network model is assessed, and a heat generation rate curve for the battery is generated based on the confidence level assessment results. Finally, the control parameters of the battery cooling system are output based on the final SOH, the heat generation rate curve, and the temperature.

[0008] The battery thermal management method proposed above in this invention may also have the following additional technical features:

[0009] According to one embodiment of the present invention, the reliability assessment of the capacity-based State of Health (SOH) based on the acquired voltage, current, and temperature specifically includes: calculating the battery's cycle integrity C_cycle based on the SOC change ΔSOC during the test cycle; calculating the temperature standard deviation σ_T during the test cycle based on the acquired temperature, and calculating the battery's temperature stability C_temp based on the temperature standard deviation σ_T; calculating the battery's measurement consistency C_consist based on the currently detected battery capacity C_est and the historical average capacity C_avg; calculating the battery's current suitability C_rate based on the acquired current I_est and the optimal evaluation current I_opt; and calculating the reliability Conf_C of the capacity-based SOH based on the cycle integrity C_cycle, temperature stability C_temp, measurement consistency C_consist, and current suitability C_rate.

[0010] According to one embodiment of the present invention, the confidence level Conf_C of the capacity-based SOH is specifically calculated according to the following formula: Where w1 is the first capacity weighting coefficient, w2 is the second capacity weighting coefficient, w3 is the third capacity weighting coefficient, w4 is the fourth capacity weighting coefficient, C_cycle is cycle integrity, C_temp is temperature stability, C_consist is measurement consistency, and C_rate is current suitability.

[0011] According to one embodiment of the present invention, the reliability assessment of the internal resistance-based SOH is performed based on the acquired voltage, current, and temperature, specifically including: calculating the excitation intensity R_excite of the internal resistance based on the measured relative amplitude ΔI of the current jump; calculating the temperature suitability R_temp of the internal resistance based on the measured temperature T_meas; calculating the signal quality R_signa of the internal resistance based on the measured voltage signal signal-to-noise ratio; calculating the compensation reliability R_comp of the internal resistance based on the temperature compensation coefficient β; and calculating the reliability Conf_R of the internal resistance-based SOH based on the excitation intensity R_excite, the temperature suitability R_temp, the signal quality R_signa, and the compensation reliability R_comp.

[0012] According to one embodiment of the present invention, the confidence level Conf_R of SOH based on internal resistance is specifically calculated using the following formula:

[0013] Wherein, k1 is the first internal resistance weighting coefficient, k2 is the second internal resistance weighting coefficient, k3 is the third internal resistance weighting coefficient, k4 is the fourth internal resistance weighting coefficient, R_excite is the excitation intensity, R_temp is the temperature suitability, R_signal is the signal quality, and R_comp is the compensation reliability.

[0014] According to one embodiment of the present invention, the capacity-based SOH and the internal resistance-based SOH are fused and calculated based on the confidence assessment results to generate the final SOH, specifically including: calculating the absolute difference ΔSOH between the capacity-based SOH and the internal resistance-based SOH;

[0015] The contradiction diagnosis threshold T is determined based on the confidence level Conf_C of the capacity-based SOH and the confidence level Conf_R of the internal resistance-based SOH. Specifically, T = 15% is set when Conf_C > 0.7 and Conf_R > 0.7, T = 8% is set when at least one of Conf_C and Conf_R is less than 0.3, and T = 12% is set in all other cases.

[0016] The ΔSOH is evaluated. If ΔSOH > T, the smaller of the capacitance-based SOH and the internal resistance-based SOH is taken as the final SOH, and the diagnostic process is triggered. If ΔSOH ≤ T, the capacitance-based SOH and the internal resistance-based SOH are further evaluated. Specifically, if Conf_C > 0.7 and Conf_R < 0.3, the capacitance-based SOH is taken as the final SOH; if Conf_R > 0.7 and Conf_C < 0.3, the internal resistance-based SOH is taken as the final SOH; if Conf_C < 0.3 and Conf_R < 0.3, the internal resistance-based SOH is taken as the final SOH. If f_R < 0.3, the smaller of the SOH based on capacity and the SOH based on internal resistance will be used as the final SOH, and a low confidence warning will be issued. If neither of the above conditions is met, the final SOH will be calculated according to the formula SOH_fused = (Conf_C × SOH_C + Conf_R × SOH_R) / (Conf_C + Conf_R), where SOH_fused is the final SOH, SOH_C is the capacity-based SOH, and SOH_R is the internal resistance-based SOH. A moderate difference warning will be added when ΔSOH > 5%.

[0017] According to one embodiment of the present invention, updating the parameters of the electrothermal coupling model corresponding to the battery based on the collected voltage, current, and temperature includes: extracting voltage, current, and temperature data for the corresponding time period when a complete standard charge-discharge cycle is completed, or when the SOH value calculated by the BMS exceeds a set threshold; calculating the SOH and internal resistance of the battery based on the extracted data; converting the calculated SOH and internal resistance into variable names and values ​​that need to be updated for the corresponding components in the electrothermal coupling model, and generating a model update instruction file; loading the electrothermal coupling model so that the electrothermal coupling model executes the update instruction file to complete the parameter update of the electrothermal coupling model.

[0018] According to one embodiment of the present invention, a confidence assessment is performed on the heat generation rate output by the electrothermal coupling model and the neural network model, and a heat generation rate curve of the battery is generated based on the confidence assessment result. Specifically, this includes: determining whether the current state time-series feature of the input is within the high confidence interval of the neural network model; if it is within the high confidence interval of the neural network model, the heat generation rate curve is directly generated based on the heat generation rate Q_nn output by the neural network model; if it is not within the high confidence interval of the neural network model, the difference in heat generation rate ∆Q ​​between the electrothermal coupling model and the neural network model is calculated; and determining the heat generation rate curve... The system checks whether the heat rate difference ∆Q is less than or equal to a set difference threshold. If ∆Q is less than or equal to the set difference threshold, the heat rate curve is generated based on the heat rate Q_sim output by the electrothermal coupling model, or the heat rate curve is generated based on the weighted average of the heat rate Q_sim output by the electrothermal coupling model and the heat rate Q_nn output by the neural network model. If ∆Q is greater than the set difference threshold, the heat rate curve is generated based on the heat rate Q_sim output by the electrothermal coupling model, and the current sample is marked as a low-confidence sample and sent to the training module of the neural network model.

[0019] A second aspect of the present invention provides a battery thermal management system, comprising: a data acquisition layer, wherein the acquisition layer is used to acquire in real time the voltage, current, and temperature of each battery cell, and to obtain the battery state of equilibrium (SOH) calculated by the battery management system (BMS) as a capacity-based SOH; a control calculation layer, wherein the control calculation layer is used to calculate the instantaneous internal resistance of the battery based on the voltage jump value at the moment of charging and discharging, and to perform temperature compensation on the instantaneous internal resistance, and to calculate the SOH based on the internal resistance based on the compensated instantaneous internal resistance; to perform a reliability assessment on the capacity-based SOH and the internal resistance-based SOH based on the acquired voltage, current, and temperature; and to perform a reliability assessment on the capacity-based SOH and the internal resistance-based SOH based on the reliability assessment result. The system performs fusion calculations on the State of Health (SOH) to generate the final SOH; periodically updates the parameters of the corresponding electrothermal coupling model based on the collected voltage, current, and temperature; a prediction and decision layer is used to obtain the future work plan and the time-series characteristics of the current battery state, and inputs the future work plan and the current battery state time-series characteristics into the updated electrothermal coupling model and the trained neural network model to obtain the battery's heat generation rate; the confidence level of the heat generation rate output by the electrothermal coupling model and the neural network model is evaluated, and the battery's heat generation rate curve is generated based on the confidence level evaluation results; an execution layer is used to output the control parameters of the battery cooling system based on the final SOH, the heat generation rate curve, and the temperature.

[0020] A third aspect of the present invention provides a battery including the battery thermal management system described above.

[0021] The beneficial effects of this invention are:

[0022] This invention introduces a real-time SOH evaluation method based on the fusion of multi-source data on capacity and internal resistance, which enables accurate perception of the battery's "health age". This allows the thermal management system to change from single control to personalized cooling, saving energy for new batteries and slowing down the degradation of old batteries, thereby significantly extending the battery's lifespan throughout its entire life cycle.

[0023] An online calibration mechanism for the model was established, which dynamically updates the electrothermal coupling model parameters of the battery using real-time cell data, ensuring a high degree of consistency between the simulation model and the physical battery state, and providing a reliable model basis for high-precision short-term thermal behavior prediction.

[0024] A control method based on the corrected model for "prospective thermal behavior prediction" is proposed. The future operating conditions are used as the prediction input, and the cooling system is upgraded from "post-event response" to "pre-event prevention", which effectively avoids the phenomenon of temporary overheating of the battery cells due to the lag of the temperature control system.

[0025] A three-dimensional MAP diagram with SOH-heat generation rate curve-real-time temperature as input was designed as the control basis. It integrates the battery's long-term health, immediate load and current state, making more comprehensive decisions. Based on the MAP diagram, synergistic optimization control between safety, lifespan and energy efficiency can be achieved.

[0026] A hybrid prediction framework combining physical simulation models and data-driven neural networks was proposed, and a dynamic decision-making mechanism based on the matching degree between the current working conditions and historical experience was designed. While ensuring the physical reliability of the prediction results, the system’s real-time response capability to high-frequency and repetitive working conditions was greatly improved. Through the confidence arbitration mechanism, prediction tasks were intelligently allocated, enabling the system to have both the efficiency to handle known working conditions and the robustness to deal with unknown working conditions.

[0027] A closed-loop neural network for continuous learning, driven by low-confidence prediction samples, was established, enabling the system to continuously optimize itself by utilizing the "difficulties" and "new points" encountered in actual operation, thereby extending the effective service cycle. Attached Figure Description

[0028] Figure 1 This is a flowchart of a battery thermal management method according to an embodiment of the present invention;

[0029] Figure 2 This is a schematic diagram illustrating the final SOH generation principle according to an embodiment of the present invention;

[0030] Figure 3 This is a schematic diagram illustrating the principle of generating a heat production rate curve according to an embodiment of the present invention;

[0031] Figure 4 This is a block diagram of a battery thermal management system according to an embodiment of the present invention. Detailed Implementation

[0032] 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.

[0033] Figure 1 This is a flowchart of a battery thermal management method according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0034] S1 collects the battery's voltage, current, and temperature in real time.

[0035] In this invention, the voltage, current, and temperature of the battery can be collected by monitoring the voltage, current, and temperature of each cell. For example, the highest temperature among the cells can be taken as the battery temperature. The battery voltage and current can be calculated based on the voltage and current of each cell and the connection method between the cells. Thus, considering the inconsistency in the cell conditions, the "weakest link" cells can be given priority protection.

[0036] S2, obtain the battery SOH calculated by BMS, as the capacity-based SOH (SOH_C).

[0037] Specifically, the BMS can record the total ampere-hours (Ah) and the start and end SOC in each charge-discharge cycle, calculate the single-cycle capacity, and thus calculate the battery's SOH, using this SOH as the capacity-based SOH (SOH_C).

[0038] S3, calculate the instantaneous internal resistance of the battery based on the voltage jump value at the moment of charging and discharging, and perform temperature compensation on the instantaneous internal resistance. Calculate the SOH (SOH_R) based on the internal resistance based on the compensated instantaneous internal resistance.

[0039] Specifically, at the moment of charging and discharging, the voltage jump value is captured, the instantaneous internal resistance is calculated, and then temperature compensation is performed on the instantaneous internal resistance of the calculation table. The temperature compensation provides the following three schemes, which can be selected according to the required scenario and existing parameters:

[0040] Temperature compensation scheme 1: Use the temperature coefficient α (unit: % / ℃ or Ω / ℃) given in the battery supplier's datasheet, or set the temperature coefficient α in advance and use the formula: R_ref = R_test / [1 + α × (T_test - T_ref)] to calculate the internal resistance temperature compensation, where R_ref is the compensated instantaneous internal resistance, R_test is the collected battery temperature, and T_ref is the set standard temperature, such as 25℃.

[0041] Temperature Compensation Scheme 2: a. Laboratory Calibration: Perform HPPC (Hybrid Pulse Power Characteristic) testing on the battery cell across the entire temperature range to establish its RT (internal resistance-temperature) curve database. b. Curve Storage: Store the RT curves in the BMS or thermal management controller. c. Online Lookup: After the BMS measures the temperature online, calculate R_ref by looking up a table + linear interpolation or by directly substituting into the fitting formula.

[0042] Temperature compensation scheme 3: Continuously collect (T, R) data of the system at different temperatures during operation, update and optimize the RT curve model online through machine learning algorithms, and use the updated latest model for compensation calculation.

[0043] Based on the compensated instantaneous internal resistance, the SOH based on the internal resistance is calculated by the rate of change of internal resistance.

[0044] S4 assesses the reliability of both capacity-based and internal resistance-based SOHs based on the collected voltage, current, and temperature.

[0045] In one embodiment of the present invention, the reliability assessment of the capacity-based SOH is performed based on the collected voltage, current, and temperature, specifically including the following steps S401-S405:

[0046] S401, calculate the battery cycle integrity C_cycle based on the SOC change ΔSOC during the test cycle.

[0047] Specifically, C_cycle = ΔSOC / 100%. The high confidence condition for cycle integrity C_cycle is ΔSOC > 70%, and the low confidence condition for cycle integrity C_cycle is ΔSOC < 30%.

[0048] S402, calculate the temperature standard deviation σ_T within the test cycle based on the collected temperature, and calculate the battery temperature stability C_temp based on the temperature standard deviation σ_T.

[0049] Specifically, C_temp = (1 - σ_T) / 5. The high confidence condition for temperature stability C_temp is σ_T < 2℃, and the low confidence condition for temperature stability C_temp is σ_T > 5℃.

[0050] S403, calculate the measurement consistency C_consist of battery capacity based on the currently detected battery capacity C_est and the average capacity C_avg over a historical period.

[0051] Specifically, C_consist = 1 - (C_est - C_avg) / C_avg, where C_avg is the average capacity measured over a historical period, representing the average capacity measured some time prior to the current time. C_est is the capacity-based State of Health (SOH) calculated by the BMS. A high confidence level for measurement consistency C_consist is achieved when the deviation between the currently measured battery capacity C_est and the average capacity C_avg over a historical period is less than 2%, while a low confidence level is achieved when the deviation between the currently measured battery capacity C_est and the average capacity C_avg over a historical period is greater than 5%.

[0052] S404, calculate the battery current suitability C_rate based on the collected current I_est and the optimal evaluation current I_opt.

[0053] Specifically, C_rate = 1 - (I_est - I_opt) / I_opt. The high confidence condition for current suitability C_rate is that the collected current I_est is close to the optimal evaluation current I_opt of the battery, and the low confidence condition for current suitability C_rate is that the collected current I_est is too large or too small.

[0054] S405, calculate the confidence level Conf_C of capacity-based SOH based on cycle integrity C_cycle, temperature stability C_temp, measurement consistency C_consist, and current suitability C_rate.

[0055] Furthermore, in a specific embodiment of the present invention, the confidence level Conf_C based on capacity-based SOH is calculated according to the following formula:

[0056] Conf_C=w1×C_cycle+w2×C_temp+w3×C_consist+w4×C_rate;

[0057] Where w1 is the first capacity weighting coefficient, w2 is the second capacity weighting coefficient, w3 is the third capacity weighting coefficient, w4 is the fourth capacity weighting coefficient, C_cycle is cycle integrity, C_temp is temperature stability, C_consist is measurement consistency, and C_rate is current suitability.

[0058] It is understandable that w1, w2, w3, and w4 can be set according to the actual situation. For example, w1 = 0.4, w2 = 0.25, w3 = 0.2, and w4 = 0.15. w1, w2, w3, and w4 can be adjusted regularly according to the calculated cycle integrity C_cycle, temperature stability C_temp, measurement consistency C_consist, and current suitability C_rate. For example, if C_cycle is in the corresponding high-confidence condition, w1 can be increased; if C_temp is in the corresponding low-confidence condition, w2 can be decreased. When adjusting the weight coefficients, it is necessary to ensure that w1 + w2 + w3 + w4 = 1.

[0059] In an embodiment of the present invention, a credibility assessment of the SOH based on internal resistance is performed based on the collected voltage, current, and temperature, specifically including the following steps S406 - S410:

[0060] S406, calculate the excitation intensity R_excite of the internal resistance according to the relative amplitude of the measured current jump ΔI.

[0061] Specifically, R_excite = ΔI / I_rated, where I_rated is the preset rated current. The high-confidence condition for the excitation intensity R_excite is ΔI / I_rate > 0.5, and the low-confidence condition for the excitation intensity R_excite is ΔI / I_rate < 0.2.

[0062] S407, calculate the temperature suitability R_temp of the internal resistance according to the measured temperature T_meas.

[0063] Specifically, R_temp = 1 - |T_meas - 25| / 40. The high-confidence condition for the temperature suitability R_temp is 20°C < T_meas < 35°C, and the low-confidence condition for the temperature suitability R_temp is T_meas < 0°C or > 45°C.

[0064] S408, calculate the signal quality R_signa of the internal resistance according to the signal-to-noise ratio of the measured voltage signal.

[0065] Specifically, R_signal = SNR_V / SNR_th, where SNR_V is the voltage signal power and SNR_th is the noise power. The high-confidence condition for the signal quality R_signa is a high signal-to-noise ratio and a clear waveform, and the low-confidence condition for the signal quality R_signa is a large noise and a blurred voltage jump edge.

[0066] S409, calculate the compensation reliability R_comp of the internal resistance according to the temperature compensation coefficient β.

[0067] Specifically, R_comp = 1-β. Using a highly calibrated β value is a high-confidence condition for compensating for reliability R_comp, while using a default or coarsely estimated β value is a low-confidence condition for compensating for reliability R_comp.

[0068] S410, calculate the confidence level Conf_R of SOH based on internal resistance according to excitation intensity R_excite, temperature suitability R_temp, signal quality R_signa, and compensation reliability R_comp.

[0069] In one embodiment of the present invention, the confidence level Conf_R of SOH based on internal resistance is specifically calculated using the following formula:

[0070] Conf_R = k1×R_excite + k2×R_temp + k3×R_signal + k4×R_comp;

[0071] Where k1 is the first internal resistance weighting coefficient, k2 is the second internal resistance weighting coefficient, k3 is the third internal resistance weighting coefficient, k4 is the fourth internal resistance weighting coefficient, R_excite is the excitation intensity, R_temp is the temperature suitability, R_signal is the signal quality, and R_comp is the compensation reliability.

[0072] Understandably, k1, k2, k3, and k4 can be set according to the actual situation, for example, k1=0.35, k2=0.3, k3=0.2, and k4=0.15. k1, k2, k3, and k4 can be adjusted periodically based on the calculated excitation intensity R_excite, temperature suitability R_temp, signal quality R_signa, and compensation reliability R_comp. For example, if R_excite is at the corresponding high confidence level, k1 can be increased; if R_temp is at the corresponding low confidence level, k2 can be decreased. When adjusting the weighting coefficients, it is necessary to ensure that k1+k2+k3+k4=1.

[0073] S5, based on the credibility assessment results, performs a fusion calculation of the capacity-based SOH and the internal resistance-based SOH to generate the final SOH.

[0074] In one embodiment of the present invention, such as Figure 2 As shown, based on the reliability assessment results, the capacity-based SOH and the internal resistance-based SOH are fused and calculated to generate the final SOH, specifically including:

[0075] S501, calculate the absolute difference between SOH based on capacitance and SOH based on internal resistance, ΔSOH=|SOH_C-SOH_R|.

[0076] S502, determine the contradiction diagnosis threshold T based on the confidence level Conf_C of SOH based on capacity and the confidence level Conf_R of SOH based on internal resistance.

[0077] S503, when Conf_C > 0.7 and Conf_R > 0.7, set T = 15%; when at least one of Conf_C and Conf_R is less than 0.3, set T = 8%; otherwise, set T = 12%.

[0078] It should be noted that the above thresholds of 8%, 12%, and 15% are only examples. In actual applications, they can be adjusted according to the battery's chemical system, operating conditions, historical data statistics, or safety level requirements.

[0079] S504, determine ΔSOH.

[0080] S505, if ΔSOH > T, then the smaller value of SOH based on capacitance and SOH based on internal resistance will be used as the final SOH, and the diagnostic process will be triggered, i.e., SOH_fused = min(SOH_C, SOH_R).

[0081] Specifically, if ΔSOH > T, it indicates that there is a serious contradiction in the calculation results of SOH by the two methods. The smaller value between the two is directly taken as the final SOH, i.e., SOH_fused = min(SOH_C, SOH_R), and the diagnostic process is triggered to record the contradictory event and start a special diagnosis to check for sensor or battery cell failure.

[0082] S506, if ΔSOH≤T, then further determine the SOH based on capacitance and the SOH based on internal resistance.

[0083] S507, if Conf_C > 0.7 and Conf_R < 0.3, then the capacity-based SOH is taken as the final SOH, i.e., SOH_fused = SOH_C.

[0084] S508, if Conf_R > 0.7 and Conf_C < 0.3, then the SOH based on the internal resistance is taken as the final SOH, that is, SOH_fused = SOH_R.

[0085] S509, if Conf_C < 0.3 and Conf_R < 0.3, then the smaller value of the SOH based on capacity and the SOH based on internal resistance is taken as the final SOH, i.e., SOH_fused = min(SOH_C, SOH_R), and a low confidence warning is issued.

[0086] S510 If none of the conditions in S506, S507, and S508 are met (i.e., ΔSOH≤T, high confidence dominance is not met, and at least one confidence level is not less than 0.3), then the final SOH is calculated using the confidence-weighted average formula: SOH_fused=(Conf_C×SOH_C+Conf_R×SOH_R) / (Conf_C+Conf_R). Furthermore, when ΔSOH>5%, an additional moderate difference warning is added.

[0087] Specifically, when one condition is ideal (high confidence) while the other is harsh, the data from the ideal condition is directly taken as the final SOH. When the two results are close, a fine-grained weighting is performed using confidence levels to obtain the final SOH. If the SOH based on capacity and the SOH based on internal resistance are severely contradictory, the smaller value is output, triggering a diagnostic process. This contradictory event is recorded, and a dedicated diagnostic procedure is initiated to check for individual cell failures in the sensor or battery. For other cases not explicitly listed in the above steps, the smaller value between the SOH based on capacity and the SOH based on internal resistance is used by default as the final SOH to ensure the conservatism and safety of the system output.

[0088] S6 updates the electrothermal coupling model parameters corresponding to the battery based on the collected voltage, current and temperature.

[0089] In one embodiment of the present invention, updating the parameters of the electrothermal coupling model corresponding to the battery based on the collected voltage, current, and temperature includes: extracting voltage, current, and temperature data for the corresponding time period when a complete standard charge-discharge cycle is completed, or when the SOH value calculated by the BMS exceeds a set threshold; calculating the SOH and internal resistance of the battery based on the extracted data; converting the calculated SOH and internal resistance into variable names and values ​​that need to be updated for the corresponding components in the electrothermal coupling model, and generating a model update instruction file; loading the electrothermal coupling model so that the electrothermal coupling model executes the update instruction file to complete the parameter update of the electrothermal coupling model.

[0090] Specifically, the electrothermal coupling model uses Amesim for modeling and simulation. When a complete standard charge-discharge cycle is completed, or when the SOH value calculated by the BMS exceeds a set threshold, an electrothermal coupling model parameter update event is triggered. After the event is triggered, the BMS automatically extracts high-frequency raw data (battery voltage, current, and temperature) and calculation results for the relevant time period, packages the data in a predetermined format, and sends the data packet to the server via the energy storage system's internal network. The server, utilizing its powerful computing capabilities, can finely calculate parameters such as capacity and internal resistance. Based on a mapping table, it converts the calculated physical parameters such as capacity and internal resistance into the variable names and values ​​that need to be updated for the corresponding components in the Amesim model (such as battery cell components and internal resistance parameters). According to the methods supported by Amesim, it generates the model update instruction file. The automated script starts, loads the Amesim simulation model, executes the update instruction file, replaces the old file with the newly generated parameter file, and re-initializes the model, completing the electrothermal coupling model parameter update. Therefore, an online calibration mechanism for the model was established, which dynamically updates the electrothermal coupling model parameters of the battery using real-time cell data, ensuring a high degree of consistency between the simulation model and the physical battery state, and providing a reliable model basis for high-precision short-term thermal behavior prediction.

[0091] S7: Obtain the future work plan and the timing features of the current battery state. Input the future work plan and the timing features of the current battery state into the updated electrothermal coupling model and the trained neural network model to obtain the heat generation rate of the battery.

[0092] S8 performs a confidence assessment on the heat generation rate output by the electrothermal coupling model and the neural network model, and generates the heat generation rate curve of the battery based on the confidence assessment results.

[0093] Specifically, in the early stages of system operation, an electrothermal coupling model can be used to simulate and generate heat generation rate simulation data Q_sim that strictly corresponds to the measured operating conditions, forming a sample library. Once the sample library reaches a preset threshold (e.g., 100 sets of data), offline training of the neural network can be initiated, using the heat generation rate as the training objective, to train a time-series neural network prediction model.

[0094] Upon receiving future work plans (e.g., charging and discharging plans) from the EMS (Energy Management System), a dual prediction pathway is initiated in parallel. The future work plans and the current state are input into the latest corrected electrothermal coupling model, and simulation is run to obtain the heat generation rate Q_sim. The same input features are then input into a pre-loaded, trained neural network model to obtain a rapid prediction result Q_nn for the heat generation rate. Confidence assessments are performed on the heat generation rates Q_sim and Q_nn output by the electrothermal coupling model and the neural network model, and the battery's heat generation rate curve is generated based on the confidence assessment results.

[0095] In one specific embodiment of the present invention, such as Figure 3 As shown, the confidence levels of the heat generation rates output by the electrothermal coupling model and the neural network model are evaluated. Based on the confidence evaluation results, the heat generation rate curve of the battery is generated, specifically including:

[0096] S801, determine whether the current state temporal features of the input are within the high confidence interval of the neural network model.

[0097] S802, if it is in the high confidence interval of the neural network model, the heat production rate curve is directly generated based on the heat production rate Q_nn output by the neural network model.

[0098] S803, if not within the high confidence interval of the neural network model, calculate the difference in heat generation rate ∆Q ​​between the outputs of the electrothermal coupling model and the neural network model.

[0099] S804, determine whether the difference in heat production rate ∆Q ​​is less than or equal to the set difference threshold.

[0100] S805, if ∆Q is less than or equal to the set difference threshold, then generate a heat production rate curve based on the heat production rate Q_sim output by the electrothermal coupling model, or generate a heat production rate curve based on the weighted average of the heat production rate Q_sim output by the electrothermal coupling model and the heat production rate Q_nn output by the neural network model.

[0101] S806, if ∆Q is greater than the set difference threshold, then generate a heat production rate curve based on the heat production rate Q_sim output by the electrothermal coupling model, and mark the current sample as a low-confidence sample and send it to the training module of the neural network model.

[0102] Specifically, the high-confidence interval of the neural network model is defined based on the training sample data of the neural network, corresponding to time-series features with sufficient training samples and low errors. The system continuously determines whether the current input battery state time-series features fall within the "high-confidence interval" of the neural network model (i.e., a feature space with sufficient training and low historical prediction errors). If it does, the heat generation rate Q_nn output by the neural network model is preferentially adopted to generate the heat generation rate curve Q_final, maximizing response speed. Simultaneously, such predictions are sampled at a lower frequency and compared with Q_sim in the background for verification, continuously monitoring the reliability of the neural network predictions.

[0103] If the input battery current state time-series features are not in a high-confidence interval or a new operating condition, a forced decision is initiated: the difference between the two model predictions ΔQ = |Q_nn - Q_sim| is calculated. If ΔQ is less than a set difference threshold, it indicates that the two model predictions are consistent, and Q_sim or the weighted average of the two can be adopted as Q_final. If ΔQ is greater than the set difference threshold, it indicates that there is a significant discrepancy in the predictions. Based on the safety principle, the simulation result Q_sim of the electrothermal coupling model is adopted first as Q_final, and this set of "input features and Q_sim labels" is automatically labeled as high-value learning samples and stored in the learning buffer.

[0104] Therefore, this invention is based on a hybrid prediction framework that combines physical simulation models and data-driven neural networks, and designs a dynamic decision-making mechanism based on the matching degree between the current working conditions and historical experience. While ensuring the physical reliability of the prediction results, it greatly improves the system's real-time response capability to high-frequency and repetitive working conditions. Through a confidence arbitration mechanism, prediction tasks are intelligently allocated, which enables both the efficiency of handling known working conditions and the robustness of dealing with unknown working conditions.

[0105] The neural network model can continuously learn and evolve, triggered by two sources: ① periodic retraining from the sample database; ② incremental learning from low-confidence, high-value samples received from the decision-maker. This learning module runs in the background, generating a new network weight file after training and seamlessly replacing the online prediction neural network model. This allows its predictive capabilities to continuously evolve with battery aging and new operating conditions, forming a self-optimizing closed loop of "prediction-verification-learning-update." It can continuously self-optimize by utilizing the "difficulties" and "new points" encountered in actual operation, extending the effective service life.

[0106] S9 outputs the control parameters of the battery cooling system based on the final SOH, heat generation rate curve, and temperature.

[0107] Specifically, the system uses real-time final SOH, heat generation rate curve, and real-time temperature as common inputs to query a preset 3D MAP. The MAP outputs the corresponding target flow rate and target temperature of the coolant, converting the target values ​​into control parameters for the pumps, valves, and compressors of the cooling system, thereby achieving battery thermal management control.

[0108] In summary, the battery thermal management method according to embodiments of the present invention introduces a real-time SOH evaluation method based on the fusion of multi-source data on capacity and internal resistance, achieving accurate perception of the battery's "healthy age." This transforms the thermal management system from simple control to personalized cooling, saving energy for new batteries and slowing down the degradation of older batteries, thereby significantly extending the battery's lifespan. An online model calibration mechanism is established, dynamically updating the battery's electrothermal coupling model parameters using real-time cell data, ensuring a high degree of consistency between the simulation model and the physical battery state, providing a reliable model foundation for high-precision short-term thermal behavior prediction. A control method based on the calibrated model for "proactive thermal behavior prediction" is proposed, using future operating conditions as predictive input, upgrading the cooling system from "post-event response" to "pre-event prevention," effectively avoiding temporary cell overheating caused by temperature control system lag. A design based on... A three-dimensional MAP diagram with SOH-heat generation rate curve-real-time temperature as input is used as the control basis, which integrates the battery's long-term health, immediate load, and current state, resulting in more comprehensive decision-making. Based on the MAP diagram, synergistic optimization control between safety, lifespan, and energy efficiency can be achieved. A hybrid prediction framework combining physical simulation model and data-driven neural network is proposed, and a dynamic decision-making mechanism based on the matching degree between current operating conditions and historical experience is designed. While ensuring the physical reliability of the prediction results, the system's real-time response capability to high-frequency and repetitive operating conditions is greatly improved. Through a confidence arbitration mechanism, prediction tasks are intelligently allocated, enabling the system to have both the efficiency to handle known operating conditions and the robustness to cope with unknown operating conditions. A neural network continuous learning closed loop driven by low-confidence prediction samples is established, enabling the system to continuously self-optimize by utilizing the "difficulties" and "new points" encountered in actual operation, thereby extending the effective service cycle.

[0109] Corresponding to the battery thermal management method described above, this invention also proposes a battery thermal management system. Since the system embodiments of this invention correspond to the method embodiments described above, details not disclosed in the system embodiments can be found in the method embodiments described above, and will not be repeated here.

[0110] Figure 4 This is a block diagram of a battery thermal management system according to an embodiment of the present invention, as shown below. Figure 4 As shown, the battery thermal management system includes: a data acquisition layer 100, a control calculation layer 200, a prediction and decision layer 300, and an execution layer 400.

[0111] The data acquisition layer 100 is used to collect the voltage, current, and temperature of each battery cell in real time, and to obtain the battery SOH calculated by the BMS as the capacity-based SOH. The control calculation layer 200 is used to calculate the instantaneous internal resistance of the battery based on the voltage jump value at the moment of charging and discharging, and to perform temperature compensation on the instantaneous internal resistance. Based on the compensated instantaneous internal resistance, the control calculation layer 200 is used to calculate the SOH based on the internal resistance; to evaluate the reliability of the capacity-based SOH and the internal resistance-based SOH based on the collected voltage, current, and temperature; to perform a fusion calculation of the capacity-based SOH and the internal resistance-based SOH based on the reliability evaluation results to generate the final SOH; and to periodically update the electrothermal coupling model parameters corresponding to the battery based on the collected voltage, current, and temperature. The prediction and decision layer 300 is used to obtain the future work plan and the time-series characteristics of the current battery state. These characteristics are then input into the updated electrothermal coupling model and the trained neural network model to obtain the battery's heat generation rate. The confidence level of the heat generation rate output by the electrothermal coupling model and the neural network model is evaluated, and a heat generation rate curve for the battery is generated based on the confidence evaluation results. The execution layer 400 is used to output the control parameters of the battery cooling system based on the final state of equilibrium (SOH), the heat generation rate curve, and the temperature.

[0112] According to one embodiment of the present invention, the control calculation layer 200 is specifically used to: calculate the cycle integrity C_cycle of the battery based on the SOC change ΔSOC during the test cycle; calculate the temperature standard deviation σ_T during the test cycle based on the collected temperature, and calculate the temperature stability C_temp of the battery based on the temperature standard deviation σ_T; calculate the measurement consistency C_consist of the battery capacity based on the currently detected battery capacity C_est and the average capacity C_avg over a historical time period; calculate the current suitability C_rate of the battery based on the collected current I_est and the optimal evaluation current I_opt; and calculate the confidence level Conf_C based on the capacity-based SOH based on the cycle integrity C_cycle, temperature stability C_temp, measurement consistency C_consist, and current suitability C_rate.

[0113] According to one embodiment of the present invention, the control computing layer 200 specifically calculates the confidence level Conf_C of the capacity-based SOH according to the following formula:

[0114] Conf_C=w1×C_cycle+w2×C_temp+w3×C_consist+w4×C_rate;

[0115] Where w1 is the first capacity weighting coefficient, w2 is the second capacity weighting coefficient, w3 is the third capacity weighting coefficient, w4 is the fourth capacity weighting coefficient, C_cycle is cycle integrity, C_temp is temperature stability, C_consist is measurement consistency, and C_rate is current suitability.

[0116] According to one embodiment of the present invention, the control calculation layer 200 is specifically configured to: calculate the excitation intensity R_excite of the internal resistance based on the measured relative amplitude ΔI of the current jump; calculate the temperature suitability R_temp of the internal resistance based on the measured temperature T_meas; calculate the signal quality R_signa of the internal resistance based on the measured voltage signal signal-to-noise ratio; calculate the compensation reliability R_comp of the internal resistance based on the temperature compensation coefficient β; and calculate the confidence level Conf_R of the SOH based on the internal resistance based on the excitation intensity R_excite, temperature suitability R_temp, signal quality R_signa, and compensation reliability R_comp.

[0117] According to one embodiment of the present invention, the control calculation layer 200 specifically uses the following formula to calculate the confidence level Conf_R of SOH based on internal resistance:

[0118] Conf_R = k1×R_excite + k2×R_temp + k3×R_signal + k4×R_comp;

[0119] Where k1 is the first internal resistance weighting coefficient, k2 is the second internal resistance weighting coefficient, k3 is the third internal resistance weighting coefficient, k4 is the fourth internal resistance weighting coefficient, R_excite is the excitation intensity, R_temp is the temperature suitability, R_signal is the signal quality, and R_comp is the compensation reliability.

[0120] According to one embodiment of the present invention, the control calculation layer 200 is specifically used to: firstly calculate the absolute difference ΔSOH = |SOH_C - SOH_R| between the SOH based on capacity and the SOH based on internal resistance; then determine the contradiction diagnosis threshold T based on the confidence levels Conf_C and Conf_R, wherein T = 15% is set when Conf_C > 0.7 and Conf_R > 0.7, T = 8% is set when at least one of Conf_C and Conf_R is less than 0.3, and T = 12% is set in other cases; if ΔSOH > T, the smaller value of the SOH based on capacity and the SOH based on internal resistance is taken as the final SOH, i.e., SOH_fused = min(SOH_C, SOH_R), and the diagnosis process is triggered; if ΔSOH ≤ T, the confidence level is further determined: if Conf_C > 0.7 and Conf_R < If Conf_R > 0.3, then the SOH based on capacity is taken as the final SOH, i.e., SOH_fused = SOH_C; if Conf_R > 0.7 and Conf_C < 0.3, then the SOH based on internal resistance is taken as the final SOH, i.e., SOH_fused = SOH_R; if the above high confidence dominant condition is not met, and Conf_C < 0.3 and Conf_R < 0.3, then the smaller of the two values ​​is taken as the final SOH, i.e., SOH_fused = min(SOH_C, SOH_R), and a low confidence warning is issued; if none of the above conditions are met, then the confidence-weighted average formula is used to calculate the final SOH: SOH_fused = (Conf_C × SOH_C + Conf_R × SOH_R) / (Conf_C + Conf_R), and a moderate difference warning is added when ΔSOH > 5%.

[0121] According to one embodiment of the present invention, the control calculation layer 200 is specifically used for: extracting voltage, current and temperature data for the corresponding time period when a complete standard charge-discharge cycle is completed, or when the SOH value calculated by the BMS exceeds a set threshold; calculating the SOH and internal resistance of the battery based on the extracted data; converting the calculated SOH and internal resistance into variable names and values ​​that need to be updated for the corresponding components in the electrothermal coupling model, and generating a model update instruction file; loading the electrothermal coupling model so that the electrothermal coupling model executes the update instruction file and completes the parameter update of the electrothermal coupling model.

[0122] According to one embodiment of the present invention, the prediction decision layer 300 is specifically used to determine whether the temporal features of the current state of the input are within the high confidence interval of the neural network model; if they are within the high confidence interval of the neural network model, a heat generation rate curve is directly generated based on the heat generation rate Q_nn output by the neural network model; if they are not within the high confidence interval of the neural network model, the difference ∆Q between the heat generation rates output by the electrothermal coupling model and the neural network model is calculated; it is determined whether the difference ∆Q is less than or equal to a set difference threshold; if ∆Q is less than or equal to the set difference threshold, a heat generation rate curve is generated based on the heat generation rate Q_sim output by the electrothermal coupling model, or a heat generation rate curve is generated based on the weighted average of the heat generation rate Q_sim output by the electrothermal coupling model and the heat generation rate Q_nn output by the neural network model; if ∆Q is greater than the set difference threshold, a heat generation rate curve is generated based on the heat generation rate Q_sim output by the electrothermal coupling model, and the current sample is marked as a low confidence sample and sent to the training module of the neural network model.

[0123] According to embodiments of the present invention, a battery thermal management system is introduced, employing a real-time SOH (State of Health) assessment method based on the fusion of multi-source data on capacity and internal resistance. This enables precise perception of the battery's "health age," transforming the thermal management system from simple control to personalized cooling. This saves energy for new batteries and slows down the degradation of older batteries, significantly extending the battery's lifespan. An online model calibration mechanism is established, dynamically updating the battery's electrothermal coupling model parameters using real-time cell data. This ensures a high degree of consistency between the simulation model and the physical battery state, providing a reliable model foundation for high-precision short-term thermal behavior prediction. A control method based on the calibrated model for "proactive thermal behavior prediction" is proposed, using future operating conditions as predictive input. This upgrades the cooling system from "post-event response" to "pre-event prevention," effectively avoiding temporary cell overheating caused by temperature control system lag. A system based on SO₂ is designed... A three-dimensional MAP diagram with H-heat generation rate curve and real-time temperature as input is used as the control basis, integrating the battery's long-term health, immediate load, and current state for more comprehensive decision-making. Based on the MAP diagram, synergistic optimization control between safety, lifespan, and energy efficiency can be achieved. A hybrid prediction framework combining physical simulation model and data-driven neural network is proposed, and a dynamic decision-making mechanism based on the matching degree between current operating conditions and historical experience is designed. While ensuring the physical reliability of prediction results, the system's real-time response capability to high-frequency and repetitive operating conditions is significantly improved. Through a confidence arbitration mechanism, prediction tasks are intelligently allocated, enabling the system to have both the efficiency to handle known operating conditions and the robustness to cope with unknown operating conditions. A neural network continuous learning closed loop driven by low-confidence prediction samples is established, enabling the system to continuously self-optimize by utilizing the "difficulties" and "new points" encountered in actual operation, thus extending the effective service cycle.

[0124] Furthermore, the present invention also proposes a battery including the battery thermal management system described above.

[0125] According to embodiments of the present invention, the battery thermal management system described above introduces a real-time SOH evaluation method based on the fusion of multi-source data on capacity and internal resistance, achieving accurate perception of the battery's "healthy age." This transforms the thermal management system from simple control to personalized cooling, saving energy for new batteries and slowing down the degradation of older batteries, thereby significantly extending the battery's lifespan. An online calibration mechanism for the model is established, dynamically updating the battery's electrothermal coupling model parameters using real-time cell data. This ensures a high degree of consistency between the simulation model and the physical battery state, providing a reliable model foundation for high-precision short-term thermal behavior prediction. A control method based on the calibrated model for "proactive thermal behavior prediction" is proposed, using future operating conditions as predictive input. This upgrades the cooling system from "post-event response" to "pre-event prevention," effectively avoiding temporary cell overheating caused by lag in the temperature control system. The design... A three-dimensional MAP diagram with SOH-heat generation rate curve-real-time temperature as input is used as the control basis, which integrates the battery's long-term health, immediate load, and current state, resulting in more comprehensive decision-making. Based on the MAP diagram, synergistic optimization control between safety, lifespan, and energy efficiency can be achieved. A hybrid prediction framework combining physical simulation model and data-driven neural network is proposed, and a dynamic decision-making mechanism based on the matching degree between current operating conditions and historical experience is designed. While ensuring the physical reliability of the prediction results, the system's real-time response capability to high-frequency and repetitive operating conditions is greatly improved. Through a confidence arbitration mechanism, prediction tasks are intelligently allocated, enabling the system to have both the efficiency to handle known operating conditions and the robustness to cope with unknown operating conditions. A neural network continuous learning closed loop driven by low-confidence prediction samples is established, enabling the system to continuously self-optimize by utilizing the "difficulties" and "new points" encountered in actual operation, thereby extending the effective service cycle.

[0126] In the description of this invention, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

[0127] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A battery thermal management method, characterized in that, Includes the following steps: Real-time monitoring of battery voltage, current, and temperature; Obtain the battery SOH calculated by BMS as the capacity-based SOH; The instantaneous internal resistance of the battery is calculated based on the voltage jump value at the moment of charging and discharging, and temperature compensation is performed on the instantaneous internal resistance. The SOH based on the internal resistance is then calculated based on the compensated instantaneous internal resistance. The reliability of the capacity-based SOH and the internal resistance-based SOH is assessed based on the collected voltage, current and temperature. The capacity-based SOH and the internal resistance-based SOH are fused and calculated based on the credibility assessment results to generate the final SOH. The parameters of the electrothermal coupling model corresponding to the battery are updated based on the collected voltage, current and temperature. The future work plan and the current state time-series features of the battery are obtained, and the future work plan and the current state time-series features of the battery are input into the updated electrothermal coupling model and the trained neural network model to obtain the heat generation rate of the battery. The confidence level of the heat generation rate output by the electrothermal coupling model and the neural network model is evaluated, and the heat generation rate curve of the battery is generated based on the confidence level evaluation results. The control parameters of the battery cooling system are output based on the final SOH, the heat generation rate curve, and the temperature.

2. The battery thermal management method according to claim 1, characterized in that, The reliability assessment of the capacity-based SOH is performed based on the collected voltage, current, and temperature, specifically including: The cycle integrity C_cycle of the battery is calculated based on the change in SOC ΔSOC during the test cycle. Calculate the temperature standard deviation σ_T within the test period based on the collected temperature, and calculate the battery temperature stability C_temp based on the temperature standard deviation σ_T. The measurement consistency C_consist of battery capacity is calculated based on the current measured battery capacity C_est and the average capacity over a historical period C_avg. The battery current suitability C_rate is calculated based on the collected current I_est and the optimal evaluation current I_opt. The confidence level Conf_C based on capacity-based SOH is calculated based on the cycle integrity C_cycle, temperature stability C_temp, measurement consistency C_consist, and current suitability C_rate.

3. The battery thermal management method according to claim 2, characterized in that, Specifically, the confidence level Conf_C based on capacity-based SOH is calculated using the following formula: ； Where w1 is the first capacity weighting coefficient, w2 is the second capacity weighting coefficient, w3 is the third capacity weighting coefficient, w4 is the fourth capacity weighting coefficient, C_cycle is cycle integrity, C_temp is temperature stability, C_consist is measurement consistency, and C_rate is current suitability.

4. The battery thermal management method according to claim 3, characterized in that, The reliability assessment of the SOH based on internal resistance is performed based on the collected voltage, current, and temperature, specifically including: The excitation intensity R_excite of the internal resistance is calculated based on the measured relative amplitude ΔI of the current jump. Calculate the temperature suitability R_temp of the internal resistance based on the measured temperature T_meas; The signal quality R_signa of the internal resistance is calculated based on the measured voltage signal signal-to-noise ratio. The reliability of internal resistance compensation, R_comp, is calculated based on the temperature compensation coefficient β. The confidence level Conf_R based on the internal resistance of the SOH is calculated based on the excitation intensity R_excite, the temperature suitability R_temp, the signal quality R_signa, and the compensation reliability R_comp.

5. The battery thermal management method according to claim 4, characterized in that, The confidence level Conf_R based on internal resistance SOH is calculated using the following formula: ； Wherein, k1 is the first internal resistance weighting coefficient, k2 is the second internal resistance weighting coefficient, k3 is the third internal resistance weighting coefficient, k4 is the fourth internal resistance weighting coefficient, R_excite is the excitation intensity, R_temp is the temperature suitability, R_signal is the signal quality, and R_comp is the compensation reliability.

6. The battery thermal management method according to claim 5, characterized in that, Based on the reliability assessment results, the capacity-based SOH and the internal resistance-based SOH are fused and calculated to generate the final SOH, specifically including: Calculate the absolute difference ΔSOH between SOH based on capacitance and SOH based on internal resistance; The contradiction diagnosis threshold T is determined based on the confidence level Conf_C of the capacity-based SOH and the confidence level Conf_R of the internal resistance-based SOH. Specifically, T = 15% is set when Conf_C > 0.7 and Conf_R > 0.7, T = 8% is set when at least one of Conf_C and Conf_R is less than 0.3, and T = 12% is set in all other cases. The ΔSOH is evaluated. If ΔSOH > T, the smaller of the capacitance-based SOH and the internal resistance-based SOH is taken as the final SOH, and the diagnostic process is triggered. If ΔSOH ≤ T, the capacitance-based SOH and the internal resistance-based SOH are further evaluated. Specifically, if Conf_C > 0.7 and Conf_R < 0.3, the capacitance-based SOH is taken as the final SOH; if Conf_R > 0.7 and Conf_C < 0.3, the internal resistance-based SOH is taken as the final SOH; if Conf_C > 0.7 and Conf_R < 0.3, the internal resistance-based SOH is taken as the final SOH. If _C < 0.3 and Conf_R < 0.3, then the smaller value of the SOH based on capacitance and the SOH based on internal resistance will be used as the final SOH, and a low confidence warning will be issued. If neither of the above conditions is met, then the final SOH will be calculated according to the formula SOH_fused = (Conf_C × SOH_C + Conf_R × SOH_R) / (Conf_C + Conf_R), where SOH_fused is the final SOH, SOH_C is the capacitance-based SOH, and SOH_R is the internal resistance-based SOH.

7. The battery thermal management method according to claim 5, characterized in that, The electrothermal coupling model parameters corresponding to the battery are updated based on the collected voltage, current, and temperature, including: Once a complete standard charge-discharge cycle is completed, or the change in SOH value calculated by the BMS exceeds the set threshold, the voltage, current, and temperature data for the corresponding time period are extracted. Calculate the battery's state of harm (SOH) and internal resistance based on the extracted data; The calculated SOH and internal resistance are converted into the variable names and values ​​that need to be updated for the corresponding components in the electrothermal coupling model, and a model update instruction file is generated. Load the electrothermal coupling model so that it can execute the update instruction file and complete the parameter update of the electrothermal coupling model.

8. The battery thermal management method according to claim 1, characterized in that, The confidence levels of the heat generation rates output by the electrothermal coupling model and the neural network model are evaluated, and the heat generation rate curve of the battery is generated based on the confidence evaluation results. Specifically, this includes: Determine whether the current state temporal features of the input are within the high confidence interval of the neural network model; If it is in the high confidence interval of the neural network model, the heat production rate curve is directly generated based on the heat production rate Q_nn output by the neural network model. If it is not in the high confidence interval of the neural network model, calculate the difference in heat production rate ∆Q ​​between the outputs of the electrothermal coupling model and the neural network model; Determine whether the heat production rate difference ∆Q is less than or equal to a set difference threshold; If ∆Q is less than or equal to the set difference threshold, the heat production rate curve is generated based on the heat production rate Q_sim output by the electrothermal coupling model, or the heat production rate curve is generated based on the weighted average of the heat production rate Q_sim output by the electrothermal coupling model and the heat production rate Q_nn output by the neural network model. If ∆Q is greater than the set difference threshold, the heat generation rate curve is generated based on the heat generation rate Q_sim output by the electrothermal coupling model, and the current sample is marked as a low-confidence sample and sent to the training module of the neural network model.

9. A battery thermal management system, characterized in that, include: The data acquisition layer is used to collect the voltage, current and temperature of each cell of the battery in real time, and to obtain the battery SOH calculated by the BMS as the capacity-based SOH; The control calculation layer is used to calculate the instantaneous internal resistance of the battery based on the voltage jump value at the moment of charging and discharging, and to perform temperature compensation on the instantaneous internal resistance. Based on the compensated instantaneous internal resistance, the SOH is calculated based on the internal resistance. The reliability of the capacity-based SOH and the internal resistance-based SOH is assessed based on the collected voltage, current and temperature. The capacity-based SOH and the internal resistance-based SOH are fused and calculated based on the credibility assessment results to generate the final SOH. The parameters of the electrothermal coupling model corresponding to the battery are periodically updated based on the collected voltage, current and temperature. A predictive decision layer is used to obtain the future work plan and the time-series characteristics of the current state of the battery, and input the future work plan and the time-series characteristics of the current state of the battery into the updated electrothermal coupling model and the trained neural network model to obtain the heat generation rate of the battery. The confidence level of the heat generation rate output by the electrothermal coupling model and the neural network model is evaluated, and the heat generation rate curve of the battery is generated based on the confidence level evaluation results. An execution layer is used to output control parameters for the battery cooling system based on the final SOH, the heat generation rate curve, and the temperature.

10. A battery, characterized in that, Includes the battery thermal management system according to claim 9.

Images