A method, apparatus and vehicle for controlling the charging and discharging of a low-voltage lithium battery

Abstract

This invention discloses a low-voltage lithium battery charging and discharging control method, device, and vehicle, comprising: real-time acquisition of the internal parameters, SOC error band, charging and discharging efficiency of the low-voltage lithium battery, as well as vehicle operating condition data, environmental data, and user behavior data; determining the vehicle's current scenario based on the vehicle operating condition data, environmental data, and user behavior data; in the default scenario, first acquiring the capacity decay rate and health status of the low-voltage lithium battery, correcting the internal resistance of the low-voltage lithium battery, and using the capacity decay rate as the SOH decay gradient; then, based on the SOC error band, vehicle operating condition data, environmental data, SOH decay gradient, and corrected internal resistance, performing reinforcement learning to output a charging and discharging control strategy; finally, executing the strategy. The capacity decay rate and health status are obtained through prediction; the reward function is determined based on the charging and discharging efficiency, health status, and reward function weights. This invention enables high-efficiency and long-life operation of low-voltage lithium batteries.

Description

Technical Field

[0001] This invention belongs to the field of low-voltage power management technology for new energy vehicles, specifically relating to a low-voltage lithium battery charging and discharging control method, device, and vehicle. Background Technology

[0002] Low-voltage lithium batteries (such as 12V lithium batteries) are widely used as low-voltage power sources for new energy vehicles. However, existing low-voltage lithium battery charging and discharging control methods have the following problems: (1) Based on fixed thresholds (such as voltage and temperature) or preset charging and discharging curves, charging and discharging control cannot adapt to lithium battery aging, environmental changes and load fluctuations, resulting in an increased risk of overcharging / over-discharging; (2) Insufficient scenario adaptability, generally only default scenarios, which cannot cover dynamic collaborative optimization of multi-modal working conditions. Summary of the Invention

[0003] The purpose of this invention is to provide a low-voltage lithium battery charging and discharging control method, device, and vehicle to achieve efficient, safe, and long-life operation of low-voltage lithium batteries.

[0004] In a first aspect, the present invention provides a low-voltage lithium battery charging and discharging control method, comprising:

[0005] Real-time acquisition of internal parameters, SOC error band, and charge / discharge efficiency η of low-voltage lithium batteries. eff In addition to vehicle operating data, environmental data, and user behavior data.

[0006] Based on vehicle operating condition data, environmental data, and user behavior data, the vehicle's current scenario is determined.

[0007] If the vehicle is in the default scenario, the charging and discharging control strategy corresponding to the default scenario will be executed.

[0008] If the vehicle is not in the default scenario, first obtain the capacity decay rate ΔC_pred and the health status SOH of the low-voltage lithium battery. cur The internal resistance of the low-voltage lithium battery is corrected, and the capacity decay rate ΔC_pred of the low-voltage lithium battery is used as the SOH decay gradient of the low-voltage lithium battery. Then, based on the SOC error band, vehicle operating data, environmental data, the SOH decay gradient, and the corrected internal resistance of the low-voltage lithium battery, reinforcement learning is performed to output a charging and discharging control strategy that corresponds to the vehicle's scenario and meets the reward objective. Finally, the charging and discharging control strategy is executed.

[0009] Among them, the capacity decay rate ΔC_pred and the state of health (SOH) of the low-voltage lithium battery curBased on the internal parameters of a low-voltage lithium battery from the past M charge-discharge cycles, the capacity decay rate and health status of the low-voltage lithium battery for the next N charge-discharge cycles are predicted; the reward function of reinforcement learning is based on the charge-discharge efficiency η. eff The health status SOH cur And the weights of the reward function are determined.

[0010] Preferably, the charge / discharge control strategy includes: adjusting at least one of the following: charging current, discharge cutoff voltage, balancing current, balancing trigger voltage difference threshold, and balancing duration of the low-voltage lithium battery, and / or hierarchical load management.

[0011] The charging and discharging control strategy corresponding to the default scenario is as follows: The charging current of the low-voltage lithium battery is maintained at a preset reference current; the discharge cutoff voltage of the low-voltage lithium battery is set to a preset discharge cutoff voltage reference value; the balancing current of the low-voltage lithium battery is set to a preset balancing current reference value; the balancing trigger voltage difference threshold of the low-voltage lithium battery is set to a preset balancing trigger voltage difference reference value; and the balancing duration of the low-voltage lithium battery is set to a preset balancing duration reference value. Balancing is triggered when the voltage difference between the two cells of the low-voltage lithium battery is greater than or equal to the balancing trigger voltage difference threshold.

[0012] Preferably, in executing the charge / discharge control strategy, the programmable DC-DC converter is controlled via SDPS instructions to adjust the charging current, discharge cutoff voltage, balancing current, balancing trigger voltage difference threshold, and balancing duration of the low-voltage lithium battery. The magnetically controlled solid-state relay is also controlled via SDPS instructions to manage the load in a tiered manner. SDPS (Software-Defined Power Supply) is an advanced power supply technology that dynamically reconfigures, intelligently manages, and flexibly controls the power functions that are traditionally fixed in hardware through software. Adopting SDPS technology reduces hardware costs.

[0013] Preferably, the internal parameters of the low-voltage lithium battery include the voltage, current, temperature, and internal resistance of the low-voltage lithium battery.

[0014] The voltage, current, temperature, and internal resistance of the low-voltage lithium battery over the past M charge-discharge cycles are input into an LSTM prediction model or an LSTM-Transformer hybrid prediction model to predict the capacity decay rate and health status of the low-voltage lithium battery over the next N charge-discharge cycles.

[0015] Preferably, the vehicle operating condition data includes navigation path gradient, load rate, and vehicle speed; the environmental data includes ambient temperature and ambient humidity; and the user behavior data includes the number of times the user accelerates rapidly and brakes rapidly per unit time.

[0016] Methods for determining the vehicle's current scenario based on vehicle operating condition data, environmental data, and user behavior data include:

[0017] Based on the ambient temperature and the preset ideal temperature, determine the temperature deviation coefficient ΔT; calculate the standard deviation of the load rate over the past preset time period and use this standard deviation as the load fluctuation rate Δload; calculate the user aggression index UAI based on the number of times the user accelerates and brakes over the past preset time period; and determine the humidity deviation coefficient ΔRH based on the ambient humidity and the preset ideal humidity.

[0018] The scene priority is calculated using the formula: Priority = W1×ΔT + W2×ΔLoad + W3×UAI + W4×ΔRH; where W1 represents the preset first weight coefficient, W2 represents the preset second weight coefficient, W3 represents the preset third weight coefficient, W4 represents the preset fourth weight coefficient, and W1+W2+W3+W4=1.

[0019] Based on the scenario priority, vehicle operating condition data, environmental data, and user behavior data, a match is made in a preset scenario library to obtain the scenario in which the vehicle is located.

[0020] By unifying the processing of three types of heterogeneous data—vehicle operating conditions, environment, and user behavior—a comprehensive scenario evaluation system was constructed. The degree of environmental anomaly was quantified by temperature deviation coefficient and humidity deviation coefficient, the standard deviation of load rate was used to characterize operating condition fluctuations, and the user's aggressiveness index was used to quantify driving style, outputting a single priority score, which simplifies the horizontal comparison of complex scenarios. Finally, scenario priority scores, along with independent vehicle operating condition data, environmental data, and user behavior data, were combined to perform scenario matching, which significantly improved matching efficiency while ensuring matching accuracy and rationality.

[0021] Preferably, the preset scenario library includes low-temperature steep slope scenarios, high-humidity high-load scenarios, urban congestion scenarios, and default scenarios. The matching conditions for high-humidity high-load scenarios are: the scenario priority is greater than or equal to Pr1, the ambient humidity is greater than a preset humidity threshold, and the load rate is greater than a preset load rate threshold. The matching conditions for low-temperature steep slope scenarios are: the scenario priority is greater than or equal to Pr2, the ambient temperature is less than a preset temperature threshold, and the navigation path gradient is greater than a preset gradient threshold. The matching conditions for urban congestion scenarios are: the scenario priority is greater than or equal to Pr3, the average vehicle speed per unit time in the past is less than a preset vehicle speed threshold, and the sum of the number of times the user accelerates and brakes per unit time in the past is greater than a preset number threshold. Wherein, Pr1 represents the first preset priority threshold, Pr2 represents the second preset priority threshold, Pr3 represents the third preset priority threshold, and Pr3 > Pr2 > Pr1.

[0022] In high-humidity, high-load scenarios, condensation easily forms on the battery casing / connectors, leading to decreased insulation or partial short circuits. Furthermore, continuous high-load operation causes rapid temperature rise in lithium batteries, accelerating SEI film decomposition and electrolyte evaporation. Therefore, a different charge / discharge control strategy than the default scenario is required. In low-temperature, steep-slope scenarios, increased internal resistance and intensified cell polarization make low-voltage lithium batteries prone to localized over-discharge. Therefore, a different charge / discharge control strategy than the default scenario is needed. In urban congestion scenarios, traffic jams cause a surge in repeated start-stop charging and discharging cycles (mostly shallow charging and discharging). Shallow charging and discharging continuously thickens the negative electrode SEI film, accumulating irreversible lithium loss. Therefore, a different charge / discharge control strategy than the default scenario is required.

[0023] Preferably, if a vehicle only meets the matching conditions for a high humidity and high load scenario, then the vehicle is determined to be in a high humidity and high load scenario; if a vehicle only meets the matching conditions for a low temperature and steep slope scenario, then the vehicle is determined to be in a low temperature and steep slope scenario; if a vehicle only meets the matching conditions for an urban congestion scenario, then the vehicle is determined to be in an urban congestion scenario; if a vehicle meets both the matching conditions for a high humidity and high load scenario and the matching conditions for a low temperature and steep slope scenario, then the vehicle is determined to be in a low temperature and steep slope scenario; if a vehicle meets both the matching conditions for a high humidity and high load scenario and the matching conditions for an urban congestion scenario, then... The vehicle is determined to be in an urban congestion scenario. If the vehicle meets the matching conditions for both the low temperature steep slope scenario and the urban congestion scenario, then the vehicle is determined to be in an urban congestion scenario. If the vehicle meets the matching conditions for the high humidity high load scenario, the low temperature steep slope scenario, and the urban congestion scenario, then the vehicle is determined to be in an urban congestion scenario. Otherwise (i.e., if the vehicle does not meet the matching conditions for the high humidity high load scenario, the low temperature steep slope scenario, or the urban congestion scenario), the vehicle is determined to be in the default scenario.

[0024] If a vehicle meets the matching conditions of two or more scenarios, the scenario with the highest preset priority threshold is taken as the scenario in which the vehicle is located, thus eliminating ambiguity in scenario recognition.

[0025] Preferably, if the vehicle is in a high-humidity, high-load scenario, the charging and discharging control strategy output after reinforcement learning, corresponding to the high-humidity, high-load scenario and satisfying the reward objective, is as follows: increase the balancing current of the low-voltage lithium battery to X times the preset balancing current benchmark value, and extend the balancing time of the low-voltage lithium battery to a preset first time threshold; where X > 1. If the vehicle is in a low-temperature, steep-slope scenario, the charging and discharging control strategy output after reinforcement learning, corresponding to the low-temperature, steep-slope scenario and satisfying the reward objective, is as follows: increase the charging current of the low-voltage lithium battery to Y times the preset benchmark current, and reduce the balancing trigger voltage difference threshold of the low-voltage lithium battery to a first preset voltage difference threshold; where Y > 1. If the vehicle is in an urban congestion scenario, the charging and discharging control strategy output after reinforcement learning, corresponding to the urban congestion scenario and satisfying the reward objective, is as follows: increase the discharge cutoff voltage of the low-voltage lithium battery by a first preset voltage threshold based on the preset discharge cutoff voltage benchmark value, and shut down non-critical loads.

[0026] Preferably, the method for correcting the internal resistance of a low-voltage lithium battery is as follows:

[0027] Using the formula: R pred = k1×ΔC_pred+b, calculate the predicted internal resistance R of the low-voltage lithium battery. pred Where k1 represents the preset electrochemical calibration coefficient, and b represents the preset electrochemical calibration value.

[0028] Using the formula: R new = R old + k2×(R meas - R pred Calculate the internal resistance R of the current corrected low-voltage lithium battery. new Where k2 represents the preset asymptotic correction coefficient, R meas R represents the measured internal resistance of a current low-voltage lithium battery. old R represents the internal resistance of the low-voltage lithium battery after the last correction. old The initial value is the preset internal resistance reference value.

[0029] Preferably, the reward function weights include lifetime weights and efficiency weights.

[0030] The reward function weights are based on the vehicle's location and the charging / discharging efficiency η of the low-voltage lithium battery. eff Confirmed, specifically:

[0031] Based on the vehicle's current scenario, a pre-defined table showing the correspondence between scenarios and lifespan weights and efficiency weights is consulted to obtain the initial lifespan weight α0 and the initial efficiency weight β0 corresponding to the vehicle's current scenario.

[0032] Set the lifespan weight α = α0 corresponding to the vehicle's scenario, and set the efficiency weight β = β0 corresponding to the vehicle's scenario.

[0033] After a certain reinforcement learning iteration, if the reward value (i.e., the reward value calculated based on the charging and discharging control strategy corresponding to the vehicle's scenario generated after this reinforcement learning iteration) is greater than or equal to a preset second reward threshold and less than a preset first reward threshold, then the formula is used: α=α0+k3×(1−η) eff ), β=1−α, update the lifespan weight α and the efficiency weight β corresponding to the vehicle's scenario; where k3 represents the preset weight adjustment coefficient.

[0034] The reward function for reinforcement learning is: R = α×[1–(SOH0-SOH)] cur ) / SOH0]+ β×η eff Where R represents the reward value and SOH0 represents the preset initial health state of the low-voltage lithium battery.

[0035] The reward objective is: the reward value is greater than or equal to a preset first reward threshold.

[0036] In a second aspect, the present invention provides a low-voltage lithium battery charging and discharging control device, which includes a controller configured to execute the above-described low-voltage lithium battery charging and discharging control method.

[0037] Thirdly, the present invention provides a vehicle that includes the aforementioned low-voltage lithium battery charging and discharging control device.

[0038] The present invention has the following effects:

[0039] (1) The vehicle's location is identified based on multimodal data (i.e., vehicle operating condition data, environmental data, and user behavior data), which improves the system's adaptability to complex and ever-changing real-world usage environments. In addition, the introduction of user behavior data allows the scenario judgment to adapt to the user's driving or usage habits to a certain extent, which is conducive to providing more personalized low-voltage lithium battery charging and discharging management in the future.

[0040] (2) A layered architecture of “default scenario preset strategy + non-default scenario reinforcement learning generation strategy” is adopted. For common and clear “default scenarios”, preset and verified optimization strategies are used to ensure the speed and stability of the response. For complex and variable “non-default scenarios”, reinforcement learning is used to generate strategies that best match the current battery state and optimization goals (efficiency + lifespan) in real time online, which has strong flexibility and adaptability.

[0041] (3) Using the SOH decay gradient (i.e. the predicted capacity decay rate ΔC_pred) as reinforcement learning input, the long-term low-voltage lithium battery degradation trend is incorporated into the short-term control decision, which improves the long-term optimality of the control and enables the control strategy to adjust the aggressiveness of charging and discharging according to the predicted decay rate. When the decay is predicted to be fast, the control strategy may be more conservative to slow down the decay. When the decay is predicted to be slow, the control strategy may allow slightly aggressive behavior to meet efficiency or performance requirements.

[0042] (4) Introducing the SOC error band in conjunction with vehicle operating conditions and environment enables reinforcement learning to generate the most suitable charging and discharging control strategy for the current conditions; the reward function is based on the charging and discharging efficiency η eff and health status SOH cur To balance the two core objectives of efficiency and lifespan, by adjusting the weights, we can perform optimizations that prioritize either efficiency or lifespan in different scenarios.

[0043] (5) Input the corrected internal resistance of the low-voltage lithium battery into reinforcement learning, so that the output charging and discharging control strategy after reinforcement learning is more accurate. Correcting the internal resistance of the low-voltage lithium battery can optimize the equalization current, shorten the convergence time of the voltage difference between cells in the low-voltage lithium battery, and improve the SOC estimation accuracy, thereby avoiding overcharging, over-discharging or ineffective control caused by inaccurate internal resistance. Attached Figure Description

[0044] Figure 1 This is a flowchart of the low-voltage lithium battery charging and discharging control method in an embodiment of the present invention.

[0045] Figure 2 This is a flowchart illustrating the method for determining the scene in which a vehicle is located, as described in an embodiment of the present invention.

[0046] Figure 3 This is a flowchart illustrating a method for correcting the internal resistance of a low-voltage lithium battery in an embodiment of the present invention.

[0047] Figure 4 This is a flowchart of reinforcement learning in an embodiment of the present invention. Detailed Implementation

[0048] To gain a more detailed understanding of the features and technical content of the embodiments of the present invention, the implementation of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for reference and illustration only and are not intended to limit the embodiments of the present invention.

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0050] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0051] like Figure 1 As shown, the low-voltage lithium battery charging and discharging control method in this embodiment of the invention includes the following steps:

[0052] S1. Real-time acquisition of internal parameters, SOC error band, and charge / discharge efficiency η of low-voltage lithium batteries. eff In addition to vehicle operating data, environmental data, and user behavior data, S2 is then executed.

[0053] In some embodiments, the internal parameters of the low-voltage lithium battery include its voltage, current, temperature, and internal resistance. The voltage is detected by a voltage sensor, the current by a current sensor, the temperature by a battery temperature sensor, and the internal resistance by measurement (e.g., DC pulse method or AC impedance method).

[0054] The SOC error band of a low-voltage lithium battery refers to the deviation range between the actual SOC value of the low-voltage lithium battery and the nominal SOC value of the BMS. It is calculated through real-time monitoring by the BMS and belongs to existing technology. The SOC error band is usually between -5% and 5%.

[0055] The charge / discharge efficiency η of low-voltage lithium batteries eff It refers to the ratio of the actual output energy to the input energy of a low-voltage lithium battery. The BMS calculates this in real time, and the calculation method is existing technology.

[0056] In some embodiments, vehicle operating data includes navigation path gradient, load rate, and vehicle speed. Vehicle speed is obtained from the CAN bus, and the average speed over a past unit of time can be calculated. Navigation path gradient refers to the gradient of the road the vehicle travels on, obtained through navigation (e.g., Gaode API). Load rate refers to the ratio of vehicle load power to rated power, as resolved in real-time via the CAN bus (range 0–100%, where 0 represents no load and 100% represents full load). The load rate is divided into discrete levels as load demand levels for use in subsequent reinforcement learning. As an example, the division rules are as follows:

[0057] Level 1: Load rate [0, 20%] → Low load

[0058] Level 2: Load factor (20%, 40%) → Low to medium load

[0059] Level 3: Load factor (40%, 60%) → Medium load

[0060] Level 4: Load Rate (60%, 80%) → High Load

[0061] Level 5: Load factor (80%, 100%) → Emergency High Load.

[0062] In some embodiments, environmental data includes ambient temperature and ambient humidity, with the ambient temperature detected by an ambient temperature sensor and the ambient humidity detected by an ambient humidity sensor. User behavior data includes the number of times the user accelerates and brakes suddenly per unit time, obtained from the CAN bus. As an example, the unit time is 1 minute.

[0063] S2. Based on vehicle operating condition data, environmental data, and user behavior data, determine the vehicle's current scenario, and then execute S3.

[0064] In some embodiments, a method for determining the scene in which the vehicle is located (see...) Figure 2 )include:

[0065] S21. Determine the temperature deviation coefficient ΔT based on the ambient temperature and the preset ideal temperature; calculate the standard deviation of the load rate over the past preset time period and use this standard deviation as the load fluctuation rate Δload; calculate the user aggression index UAI based on the number of times the user accelerates and brakes over the past preset time period; determine the humidity deviation coefficient ΔRH based on the ambient humidity and the preset ideal humidity.

[0066] As an example, the preset ideal temperature is 25℃, and the temperature deviation coefficient ΔT =|T amb -25|,T amb Indicates ambient temperature. Load fluctuation rate Δload reflects the drastic degree of load change. User aggression index UAI = 0.6 × N accel / T obs + 0.4 × N brake / T obs N accel N represents the number of times the user has accelerated within a preset time period. brake T represents the number of times the user has braked suddenly within a preset time period. obs T represents the preset time. obs =10 minutes. The past 10 minutes refers to the 10 minutes that have elapsed relative to the current time. The preset ideal humidity is 60%RH, and the humidity deviation coefficient ΔRH = |RH real - 60|, RH real Indicates ambient humidity.

[0067] S22. Using the formula: Priority = W1×ΔT + W2×ΔLoad + W3×UAI + W4×ΔRH, the scene priority is calculated. Where W1 represents the preset first weight coefficient, W2 represents the preset second weight coefficient, W3 represents the preset third weight coefficient, and W4 represents the preset fourth weight coefficient, W1 + W2 + W3 + W4 = 1. Multiple parameters jointly determine the scene priority, which is more conducive to improving the accuracy of scene judgment.

[0068] In some embodiments, W1=0.35, W2=0.25, W3=0.25, and W4=0.15. Temperature has a significant impact on the performance of low-voltage lithium batteries (e.g., low temperatures cause a sharp drop in capacity, while high temperatures accelerate aging), therefore it is given the highest weight. Experimental data shows that temperature fluctuations contribute approximately 35% of the capacity decay risk. Sudden load changes (e.g., rapid acceleration, start-up and shutdown of high-power devices) directly affect the stability of the charge-discharge strategy; historical data analysis shows that load fluctuations contribute approximately 25% to scenario matching. User driving behavior (frequency of rapid acceleration / braking) reflects the urgency of real-time operating conditions, and its weight is equal to that of load fluctuation rate, reflecting equal emphasis on human operational factors. High humidity exacerbates cell polarization, but its impact is weaker than that of temperature and load, therefore it has the lowest weight; experiments show that humidity contributes approximately 15% to the equalization efficiency.

[0069] S23. Based on the scenario priority, vehicle operating condition data, environmental data, and user behavior data, match the scenario in the preset scenario library to obtain the scenario in which the vehicle is located.

[0070] As an example, the preset scenario library includes low temperature steep slope scenario, high humidity and high load scenario, urban congestion scenario, and default scenario.

[0071] The matching conditions for high humidity and high load scenarios are: the scenario priority is greater than or equal to Pr1, the ambient humidity is greater than the preset humidity threshold, and the load rate is greater than the preset load rate threshold.

[0072] The matching conditions for low temperature steep slope scenarios are: the scenario priority is greater than or equal to Pr2, the ambient temperature is less than the preset temperature threshold, and the slope of the navigation path is greater than the preset slope threshold.

[0073] The matching conditions for urban congestion scenarios are: the scenario priority is greater than or equal to Pr3, the average vehicle speed in the past unit time is less than the preset vehicle speed threshold, and the sum of the number of times the user accelerates and brakes in the past unit time is greater than the preset number threshold.

[0074] Wherein, Pr1 represents the first preset priority threshold, Pr2 represents the second preset priority threshold, and Pr3 represents the third preset priority threshold, with Pr3 > Pr2 > Pr1. Scene priority is considered in the scene matching conditions to facilitate rapid scene switching and avoid conflicts between multiple scenes.

[0075] As an example, Pr1=14, Pr2=15, Pr3=16, the preset humidity threshold is 90%RH, the preset load rate threshold is 85%, the preset temperature threshold is -10℃, the preset slope threshold is 15%, the preset vehicle speed threshold is 20km / h, and the preset number of times threshold is 5.

[0076] In some embodiments, if a vehicle only meets the matching conditions for a high humidity and high load scenario, then the vehicle is determined to be in a high humidity and high load scenario. If a vehicle only meets the matching conditions for a low temperature and steep slope scenario, then the vehicle is determined to be in a low temperature and steep slope scenario. If a vehicle only meets the matching conditions for an urban congestion scenario, then the vehicle is determined to be in an urban congestion scenario. If a vehicle meets both the matching conditions for a high humidity and high load scenario and the matching conditions for a low temperature and steep slope scenario, then the vehicle is determined to be in a low temperature and steep slope scenario. If a vehicle meets both the matching conditions for a high humidity and high load scenario and the matching conditions for an urban congestion scenario, then the vehicle is determined to be in an urban congestion scenario. If a vehicle meets all the matching conditions for a high humidity and high load scenario, the matching conditions for a low temperature and steep slope scenario, and the matching conditions for an urban congestion scenario, then the vehicle is determined to be in an urban congestion scenario. If the vehicle does not meet the matching conditions for high humidity and high load scenarios, low temperature and steep slope scenarios, and urban congestion scenarios, then the vehicle is determined to be in the default scenario.

[0077] S3. Determine whether the vehicle is in the default scenario. If so, execute S4; otherwise, execute S5.

[0078] S4. Execute the charge / discharge control strategy corresponding to the default scenario, and then end.

[0079] In some embodiments, the charge / discharge control strategy includes: adjusting at least one of the following: charging current, discharge cutoff voltage, balancing current, balancing trigger voltage difference threshold, and balancing duration of the low-voltage lithium battery, and / or hierarchical load management.

[0080] In some embodiments, the charge / discharge control strategy corresponding to the default scenario is as follows: the charging current of the low-voltage lithium battery is set to a preset reference current; the discharge cutoff voltage of the low-voltage lithium battery is set to a preset discharge cutoff voltage reference value; the balancing current of the low-voltage lithium battery is set to a preset balancing current reference value; the balancing trigger voltage difference threshold of the low-voltage lithium battery is set to a preset balancing trigger voltage difference reference value; and the balancing duration of the low-voltage lithium battery is set to a preset balancing duration reference value. Balancing is triggered when the voltage difference between the two cells of the low-voltage lithium battery is greater than or equal to the balancing trigger voltage difference threshold of the low-voltage lithium battery.

[0081] In some embodiments, during specific execution, the controller issues SDPS commands to the programmable DC-DC converter, controlling the converter to set the charging current of the low-voltage lithium battery to a preset reference current, the discharge cutoff voltage of the low-voltage lithium battery to a preset discharge cutoff voltage reference value, the balancing current of the low-voltage lithium battery to a preset balancing current reference value, the balancing trigger voltage difference threshold of the low-voltage lithium battery to a preset balancing trigger voltage difference reference value, and the balancing duration of the low-voltage lithium battery to a preset balancing duration reference value. As an example, the programmable DC-DC converter uses the TPS548D22, supporting 0.1A step adjustment.

[0082] As an example, the preset reference current is 10A, the preset discharge cutoff voltage reference value is 10.5V, the preset equalization current reference value is 0.5A, the preset equalization trigger voltage difference reference value is 30mV, and the preset equalization duration reference value is 20 minutes.

[0083] S5. Obtain the capacity decay rate ΔC_pred and the state of health (SOH) of the low-voltage lithium battery. cur The internal resistance of the low-voltage lithium battery is corrected, and ΔC_pred is used as the SOH decay gradient of the low-voltage lithium battery. Then, S6 is executed.

[0084] Among them, the capacity decay rate ΔC_pred and state of health (SOH) of low-voltage lithium batteries cur Based on the internal parameters of the low-voltage lithium battery after M charge-discharge cycles, the capacity decay rate and health status of the low-voltage lithium battery for the next N charge-discharge cycles are predicted. The prediction process is continuously performed in the controller.

[0085] The specific method is as follows: The voltage, current, temperature, and internal resistance (i.e., a four-dimensional time series with a sliding window length of M cycles) of the low-voltage lithium battery over the past M charge-discharge cycles are input into an LSTM prediction model for prediction. This predicts the capacity decay rate (unit: % / cycle) and health status of the low-voltage lithium battery over the next N charge-discharge cycles. M is much larger than N. For example, M=1000, N=50. The initial voltage, current, temperature, and internal resistance of the low-voltage lithium battery over 1000 charge-discharge cycles can be obtained through experimental testing. Subsequently, as the low-voltage lithium battery undergoes charge-discharge cycles in the vehicle, the voltage, current, temperature, and internal resistance are detected and recorded by corresponding sensors.

[0086] In some embodiments, the voltage, current, temperature, and internal resistance of the low-voltage lithium battery over the past M charge-discharge cycles can be input into a (lightweight) LSTM-Transformer hybrid prediction model for prediction, thereby obtaining the capacity decay rate and health status of the low-voltage lithium battery over the next N charge-discharge cycles.

[0087] The lightweight LSTM-Transformer hybrid prediction model comprises an input layer, an LSTM layer, a Transformer encoder layer, a physical constraint layer, and an output layer, enabling collaborative optimization of local and global data. The input layer takes into account the voltage, current, temperature, and internal resistance of the low-voltage lithium battery over the past M charge-discharge cycles. The LSTM layer extracts local features and captures capacity degradation patterns. The Transformer encoder layer models long-term aging trends and identifies cross-cycle degradation patterns. The physical constraint layer corrects the output to ensure the prediction results meet requirements. The output layer outputs the capacity degradation rate and state of health of the low-voltage lithium battery for the next N charge-discharge cycles. It achieves higher prediction accuracy for both capacity degradation rate and state of health.

[0088] In some embodiments, a method for correcting the internal resistance of a low-voltage lithium battery (see...) Figure 3 ),include:

[0089] S51. Using the formula: R pred = k1×ΔC_pred+b, calculate the predicted internal resistance R of the low-voltage lithium battery. pred Where k1 represents the preset electrochemical calibration coefficient, and b represents the preset electrochemical calibration value.

[0090] S52. Using the formula: R new = R old + k2×(R meas - R pred Calculate the internal resistance R of the current corrected low-voltage lithium battery. new Where k2 represents the preset asymptotic correction coefficient, R measR represents the measured internal resistance of the current low-voltage lithium battery (measured using the DC pulse method or AC impedance method, which is also the internal resistance of the low-voltage lithium battery obtained in real time in S1). old R represents the internal resistance of the low-voltage lithium battery after the last correction. old The initial value is a preset internal resistance reference value (i.e., the nominal value at the time of manufacture). As an example, k2=0.2, which balances the correction magnitude with the stability of the internal resistance model and avoids overfitting.

[0091] S6. Based on the SOC error band of low-voltage lithium batteries, vehicle operating condition data, environmental data, SOH decay gradient of low-voltage lithium batteries, and the corrected internal resistance (i.e., R) of low-voltage lithium batteries. new The system performs reinforcement learning to output a charging and discharging control strategy that corresponds to the vehicle's current scenario and satisfies the reward objective, and then executes step S7. The reward function for reinforcement learning is based on the charging and discharging efficiency η of the low-voltage lithium battery. eff State of Health (SOH) of low-voltage lithium batteries cur The reward function weights are determined based on the vehicle's location and the charging / discharging efficiency η of the low-voltage lithium battery. eff Sure.

[0092] In some embodiments, reinforcement learning employs a proximal policy optimization (PPO) algorithm, suitable for continuous action spaces, balancing exploration and exploitation, and ensuring the stability of the control strategy. The input parameters (i.e., the state space) of reinforcement learning include the SOC error band of the low-voltage lithium battery, the SOH decay gradient of the low-voltage lithium battery, the load demand level (obtained based on the load rate in the vehicle operating data, as described above), ambient temperature, ambient humidity, and the corrected internal resistance of the low-voltage lithium battery. The output parameters (i.e., the action space) include: adjusting at least one of the following: charging current, discharge cutoff voltage, balancing current, balancing trigger voltage difference threshold, and balancing duration of the low-voltage lithium battery, and / or hierarchical load management. As an example, the charging current range of the low-voltage lithium battery is 1A to 20A, with a step accuracy of 1A; the discharge cutoff voltage range is 10V to 12V, with a step accuracy of 0.1V; the balancing current range is 0.5A to 1.5A; and the balancing trigger voltage difference threshold range is 10mV to 40mV.

[0093] like Figure 4 As shown, the specific steps of reinforcement learning include:

[0094] S61. Query the preset correspondence table between the scenario and the lifespan weight and efficiency weight according to the scenario in which the vehicle is located, obtain the initial value of lifespan weight α0 and the initial value of efficiency weight β0 corresponding to the scenario in which the vehicle is located, and then execute S62.

[0095] As an example, when the vehicle operates in a high-humidity, high-load environment, with the goal of balancing lifespan and efficiency, the table shows α0=0.4 and β0=0.6, which can accelerate cell consistency management. When the vehicle operates in a low-temperature, steep-slope environment, with the goal of prioritizing extended lifespan, the table shows α0=0.7 and β0=0.3, which can reduce the risk of lithium battery aging at low temperatures. When the vehicle operates in a congested urban environment, with the goal of prioritizing the suppression of capacity decay, the table shows α0=0.8 and β0=0.2, which can ensure long-term operation of the lithium battery.

[0096] S62. Set the lifetime weight α = α0 and the efficiency weight β = β0 corresponding to the vehicle's scenario, then execute S63. Here, the lifetime weight α and the efficiency weight β corresponding to the vehicle's scenario are reward function weights. The lifetime weight α corresponding to the vehicle's scenario is used to control the rate of degradation of the low-voltage lithium battery's health state. The larger α is, the more the strategy tends to reduce charging and discharging power to extend the low-voltage lithium battery's lifespan. The efficiency weight β corresponding to the vehicle's scenario is used to optimize charging and discharging efficiency. The larger β is, the more the strategy tends to increase charging and discharging speed to meet real-time requirements.

[0097] S63. Input the SOC error band of the low-voltage lithium battery, vehicle operating condition data, environmental data, SOH decay gradient of the low-voltage lithium battery, and the corrected internal resistance of the low-voltage lithium battery into the reinforcement learning model, perform reinforcement learning, generate a charging and discharging control strategy corresponding to the vehicle's scenario, and then execute S64.

[0098] The reward function for reinforcement learning is: R = α×[1–(SOH0-SOH)] cur ) / SOH0]+ β×η eff R represents the reward value, and SOH0 represents the preset initial health state of the low-voltage lithium battery, with SOH0=100%. The reward function of this reinforcement learning integrates indicators such as lifespan extension, energy efficiency improvement, and safety.

[0099] S64. Execute the charging and discharging control strategy corresponding to the scenario in which the vehicle is located, and then execute S65.

[0100] S65. Calculate the reward value corresponding to the scene where the vehicle is located, and then execute S66.

[0101] S66. Determine whether the reward value corresponding to the vehicle's current scenario is greater than or equal to R. th1 If so (indicating that the reward objective is met), then execute S69; otherwise, execute S67. Where R... th1 This represents the preset first reward threshold. As an example, R... th1 =0.85.

[0102] S67. Determine whether the reward value corresponding to the vehicle's current scenario is greater than or equal to R. th2 And less than R th1 If so, execute S68; otherwise, return to execute S63. Where R... th2 This represents the preset second reward threshold. As an example, R... th2 =0.6.

[0103] S68. Using the formula: α=α0+k3×(1−η) eff ), β=1−α, update the lifespan weight α and efficiency weight β corresponding to the vehicle's scenario, and then return to execute S65. Here, k3 represents the preset weight adjustment coefficient. As an example, k3=0.1.

[0104] S69. Output a charging and discharging control strategy that corresponds to the vehicle's current scenario and satisfies the reward objective, and then end.

[0105] In some embodiments, if the vehicle is in a high-humidity, high-load scenario, the charge-discharge control strategy output after reinforcement learning, which corresponds to the high-humidity, high-load scenario and satisfies the reward objective, is as follows: The equalization current of the low-voltage lithium battery is increased to X times the preset equalization current benchmark value, and the equalization time of the low-voltage lithium battery is extended to a preset first time threshold; where X > 1. For example, X = 1.5, and the preset first time threshold is 30 minutes, meaning the equalization current of the low-voltage lithium battery is increased to 0.75A, and the equalization time of the low-voltage lithium battery is extended to 30 minutes. The above charge-discharge control strategy in the high-humidity, high-load scenario aims to enhance the equalization current to address the polarization effect under high humidity conditions. Compared with the charge-discharge control strategy in the default scenario, it shortens the convergence time of the voltage difference between cells in the low-voltage lithium battery, reduces the standard deviation of capacity consistency, and reduces equalization energy consumption.

[0106] In some embodiments, if the vehicle is in a low-temperature steep slope scenario, the charge / discharge control strategy output after reinforcement learning, which corresponds to the low-temperature steep slope scenario and satisfies the reward objective, is as follows: The charging current of the low-voltage lithium battery is increased to Y times a preset reference current, and the equalization trigger voltage difference threshold of the low-voltage lithium battery is reduced to a first preset voltage difference threshold; where Y > 1. For example, Y = 1.2, and the first preset voltage difference threshold is 20mV, meaning the charging current of the low-voltage lithium battery is increased to 12A, and the equalization trigger voltage difference threshold is reduced to 20mV. The above charge / discharge control strategy in the low-temperature steep slope scenario primarily aims to increase the charging current to cope with the low-temperature steep slope. At low temperatures, cell polarization intensifies, and reducing the equalization trigger voltage difference threshold can accelerate equalization and prevent localized over-discharge of the low-voltage lithium battery. Compared to the charge / discharge control strategy in the default scenario, it reduces charging time, lowers the capacity decay rate, provides controllable temperature rise, and also shortens the convergence time of the voltage difference between cells within the low-voltage lithium battery, reducing the standard deviation of capacity consistency.

[0107] In some embodiments, if the vehicle is in an urban congestion scenario, the charging and discharging control strategy output after reinforcement learning, which corresponds to the urban congestion scenario and satisfies the reward objective, is as follows: The discharge cutoff voltage of the low-voltage lithium battery is increased by a first preset voltage threshold based on a preset discharge cutoff voltage benchmark value, and non-critical loads are turned off (e.g., turning off seat heating, ambient lighting, and the entertainment system). For example, the first preset voltage threshold is 0.2V, meaning the discharge cutoff voltage of the low-voltage lithium battery is increased to 10.7V. In urban congestion scenarios, frequent vehicle starts and stops lead to a surge in charging and discharging cycles, necessitating priority to suppress capacity decay and sacrifice short-term efficiency to extend the lifespan of the low-voltage lithium battery. Therefore, the above charging and discharging control strategy primarily aims to optimize energy distribution by adjusting the discharge cutoff voltage. Compared to the charging and discharging control strategy in the default scenario, although the discharge cutoff voltage is increased, the number of deep discharge cycles is reduced, thereby reducing the capacity decay rate. Furthermore, turning off non-critical loads saves energy, prioritizes power supply to the ADAS and braking systems, and improves the vehicle's driving range.

[0108] S7. Execute the charging and discharging control strategy that corresponds to the scenario in which the vehicle is located and meets the reward objective, and then end.

[0109] In some embodiments, when the vehicle is in a high humidity and high load scenario, the controller sends an SDPS command to the programmable DC-DC converter to control the programmable DC-DC converter to increase the balancing current of the low-voltage lithium battery to X times the preset balancing current reference value and extend the balancing time of the low-voltage lithium battery to a preset first time threshold.

[0110] In some embodiments, when the vehicle is in a low-temperature steep slope scenario, the controller sends an SDPS command to the programmable DC-DC converter to control the programmable DC-DC converter to increase the charging current of the low-voltage lithium battery to Y times the preset reference current and reduce the equalization trigger voltage difference threshold of the low-voltage lithium battery to a first preset voltage difference threshold.

[0111] In some embodiments, when the vehicle is in an urban congestion scenario, the controller sends an SDPS command to the programmable DC-DC converter and the magnetically controlled solid-state relay. This command controls the programmable DC-DC converter to increase the discharge cutoff voltage of the low-voltage lithium battery by a first preset voltage threshold based on a preset discharge cutoff voltage reference value. The controller then controls the magnetically controlled solid-state relay to shut down non-critical loads (such as turning off seat heating, ambient lighting, and the entertainment system). As an example, the magnetically controlled solid-state relay uses a CPC1984Y with a response time ≤1ms.

[0112] In addition, embodiments of the present invention also provide a low-voltage lithium battery charging and discharging control device, which includes a controller configured to execute the aforementioned low-voltage lithium battery charging and discharging control method. As an example, the controller uses an NXP S32G2.

[0113] In addition, embodiments of the present invention also provide a vehicle that includes the aforementioned low-voltage lithium battery charging and discharging control device.

[0114] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for controlling the charging and discharging of a low-voltage lithium battery, characterized in that, include: Real-time acquisition of internal parameters of low-voltage lithium battery, SOC error band, charge and discharge efficiency η eff And vehicle working condition data, environmental data, user behavior data; Based on vehicle operating condition data, environmental data, and user behavior data, determine the vehicle's current scenario. If the vehicle is in the default scenario, then the charging and discharging control strategy corresponding to the default scenario will be executed. If the scenario in which the vehicle is located is not the default scenario, first obtain the capacity attenuation rate ΔC_pred and the state of health SOH of the low-voltage lithium battery cur , correct the internal resistance of the low-voltage lithium battery, and take ΔC_pred as the SOH attenuation gradient of the low-voltage lithium battery; then, based on the SOC error band, vehicle working condition data, environmental data, the SOH attenuation gradient, and the corrected internal resistance of the low-voltage lithium battery, perform reinforcement learning to output a charging and discharging control strategy corresponding to the scenario in which the vehicle is located and meeting the reward target; finally, execute the charging and discharging control strategy; Among them, the capacity decay rate ΔC_pred and the state of health (SOH) of the low-voltage lithium battery cur Based on the internal parameters of a low-voltage lithium battery from the past M charge-discharge cycles, the capacity decay rate and health status of the low-voltage lithium battery for the next N charge-discharge cycles are predicted; the reward function of reinforcement learning is based on the charge-discharge efficiency η. eff The health status SOH cur And the weights of the reward function are determined.

2. The low-voltage lithium battery charging and discharging control method according to claim 1, characterized in that: The charge and discharge control strategy includes at least one of the following: adjusting the charging current, discharge cutoff voltage, balancing current, balancing trigger voltage difference threshold, and balancing duration of the low-voltage lithium battery, and / or hierarchical management of the load. The charging and discharging control strategy corresponding to the default scenario is as follows: the charging current of the low-voltage lithium battery is set to a preset reference current, the discharge cutoff voltage of the low-voltage lithium battery is set to a preset discharge cutoff voltage reference value, the balancing current of the low-voltage lithium battery is set to a preset balancing current reference value, the balancing trigger voltage difference threshold of the low-voltage lithium battery is set to a preset balancing trigger voltage difference reference value, and the balancing time of the low-voltage lithium battery is set to a preset balancing time reference value.

3. The low-voltage lithium battery charging and discharging control method according to claim 2, characterized in that: In executing the charge and discharge control strategy, the programmable DC-DC converter is controlled by SDPS instructions to adjust the charging current, discharge cutoff voltage, equalization current, equalization trigger voltage difference threshold, and equalization duration of the low-voltage lithium battery. The magnetically controlled solid-state relay is also controlled by SDPS instructions to manage the load in a hierarchical manner.

4. The low-voltage lithium battery charging and discharging control method according to claim 1, characterized in that: The internal parameters of the low-voltage lithium battery include its voltage, current, temperature, and internal resistance. The voltage, current, temperature, and internal resistance of the low-voltage lithium battery over the past M charge-discharge cycles are input into an LSTM prediction model or an LSTM-Transformer hybrid prediction model to predict the capacity decay rate and health status of the low-voltage lithium battery over the next N charge-discharge cycles.

5. The low-voltage lithium battery charging and discharging control method according to claim 1, characterized in that: The vehicle operating data includes navigation path gradient, load rate, and vehicle speed; the environmental data includes ambient temperature and ambient humidity; and the user behavior data includes the number of times the user accelerates rapidly and brakes rapidly per unit time. Methods for determining the vehicle's current scenario based on vehicle operating condition data, environmental data, and user behavior data include: Based on the ambient temperature and the preset ideal temperature, determine the temperature deviation coefficient ΔT; calculate the standard deviation of the load rate over the past preset time period and use this standard deviation as the load fluctuation rate Δload; calculate the user aggression index UAI based on the number of times the user accelerates and brakes suddenly over the past preset time period; determine the humidity deviation coefficient ΔRH based on the ambient humidity and the preset ideal humidity. The scene priority is calculated using the formula: Priority = W1×ΔT + W2×ΔLoad + W3×UAI + W4×ΔRH; where W1 represents the preset first weight coefficient, W2 represents the preset second weight coefficient, W3 represents the preset third weight coefficient, W4 represents the preset fourth weight coefficient, and W1+W2+W3+W4=1. Based on the scenario priority, vehicle operating condition data, environmental data, and user behavior data, a match is made in a preset scenario library to obtain the scenario in which the vehicle is located.

6. The low-voltage lithium battery charging and discharging control method according to claim 5, characterized in that: The preset scenario library includes low temperature steep slope scenario, high humidity high load scenario, urban congestion scenario and default scenario; The matching conditions for high humidity and high load scenarios are: the scenario priority is greater than or equal to Pr1, the ambient humidity is greater than the preset humidity threshold, and the load rate is greater than the preset load rate threshold. The matching conditions for low temperature steep slope scenarios are: the scenario priority is greater than or equal to Pr2, the ambient temperature is less than the preset temperature threshold, and the slope of the navigation path is greater than the preset slope threshold. The matching conditions for urban congestion scenarios are: the scenario priority is greater than or equal to Pr3, the average vehicle speed in the past unit time is less than the preset vehicle speed threshold, and the sum of the number of times the user accelerates and brakes in the past unit time is greater than the preset number threshold. Wherein, Pr1 represents the first preset priority threshold, Pr2 represents the second preset priority threshold, Pr3 represents the third preset priority threshold, and Pr3 > Pr2 > Pr1.

7. The low-voltage lithium battery charging and discharging control method according to claim 6, characterized in that: If a vehicle only meets the matching conditions for a high humidity and high load scenario, then the vehicle is determined to be in a high humidity and high load scenario. If a vehicle only meets the matching conditions for a low-temperature steep slope scenario, then the scenario in which the vehicle is located is determined to be a low-temperature steep slope scenario. If a vehicle only meets the matching criteria for an urban congestion scenario, then the vehicle is determined to be in an urban congestion scenario. If a vehicle meets both the matching conditions for a high humidity and high load scenario and the matching conditions for a low temperature and steep slope scenario, then the scenario in which the vehicle is located is determined to be a low temperature and steep slope scenario. If a vehicle meets both the matching conditions for a high humidity and high load scenario and the matching conditions for an urban congestion scenario, then the scenario in which the vehicle is located is determined to be an urban congestion scenario. If a vehicle meets the matching conditions for both the low-temperature steep slope scenario and the urban congestion scenario, then the scenario in which the vehicle is located is determined to be the urban congestion scenario. If a vehicle simultaneously meets the matching conditions for high humidity and high load scenarios, low temperature and steep slope scenarios, and urban congestion scenarios, then the scenario in which the vehicle is located is determined to be an urban congestion scenario. Otherwise, the vehicle is determined to be in the default scenario.

8. The low-voltage lithium battery charging and discharging control method according to claim 6, characterized in that: If the vehicle is in a high humidity and high load scenario, the charging and discharging control strategy output after reinforcement learning that corresponds to it and meets the reward objective is: to increase the equalization current of the low-voltage lithium battery to X times the preset equalization current benchmark value, and to extend the equalization time of the low-voltage lithium battery to the preset first time threshold; where X>1. If the vehicle is in a low-temperature steep slope scenario, the charging and discharging control strategy output after reinforcement learning that corresponds to it and meets the reward objective is: to increase the charging current of the low-voltage lithium battery to Y times the preset reference current, and to reduce the equalization trigger voltage difference threshold of the low-voltage lithium battery to the first preset voltage difference threshold; where Y>1. If the vehicle is in an urban congestion scenario, the charging and discharging control strategy output after reinforcement learning that corresponds to and meets the reward objective is: increase the discharge cutoff voltage of the low-voltage lithium battery by a first preset voltage threshold based on the preset discharge cutoff voltage benchmark value, and shut down non-critical loads.

9. The low-voltage lithium battery charging and discharging control method according to any one of claims 1 to 8, characterized in that, The method for correcting the internal resistance of low-voltage lithium batteries is as follows: Using the formula: R pred = k1×ΔC_pred+b, calculate the predicted internal resistance R of the low-voltage lithium battery. pred Where k1 represents the preset electrochemical calibration coefficient, and b represents the preset electrochemical calibration value; Using the formula: R new = R old + k2×(R meas - R pred Calculate the internal resistance R of the current corrected low-voltage lithium battery. new Where k2 represents the preset asymptotic correction coefficient, R meas R represents the measured internal resistance of a current low-voltage lithium battery. old R represents the internal resistance of the low-voltage lithium battery after the last correction. old The initial value is the preset internal resistance reference value.

10. The low-voltage lithium battery charging and discharging control method according to any one of claims 1 to 8, characterized in that: The reward function weights include lifetime weights and efficiency weights, which are determined based on the vehicle's operating environment and the charging / discharging efficiency η. eff Confirmed, specifically: Based on the vehicle's current scenario, query the preset correspondence table between scenario and lifespan weight and efficiency weight to obtain the initial value α0 of lifespan weight and the initial value β0 of efficiency weight corresponding to the vehicle's current scenario. Set the lifetime weight α=α0 and the efficiency weight β=β0 to correspond to the scenario in which the vehicle is located. After a certain reinforcement learning iteration, if the reward value is greater than or equal to a preset second reward threshold and less than a preset first reward threshold, then the formula is used: α = α0 + k3 × (1 − η) eff ), β=1−α, update the lifespan weight α and efficiency weight β corresponding to the scenario in which the vehicle is located; where k3 represents the preset weight adjustment coefficient; The reward function for reinforcement learning is: R = α×[1–(SOH0-SOH)] cur ) / SOH0] + β×η eff Where R represents the reward value, and SOH0 represents the preset initial health state of the low-voltage lithium battery; The reward objective is: the reward value is greater than or equal to a preset first reward threshold.

11. A low-voltage lithium battery charging and discharging control device, comprising a controller, characterized in that: The controller is configured to perform the low-voltage lithium battery charge and discharge control method as described in any one of claims 1 to 10.

12. A vehicle, characterized in that: Includes the low-voltage lithium battery charge and discharge control device as described in claim 11.

Images