Charging and discharging control strategy based on performance degradation of power battery

Abstract

The invention provides a charging and discharging control strategy based on performance degradation of a power battery, and belongs to the technical field of charging piles, and the charging and discharging control strategy comprises the steps: training a deep reinforcement learning algorithm based on an electrochemical-thermal-aging coupling model, and obtaining an optimal charging strategy of the battery; and inputting the time-of-use electricity price of the power grid and the battery state into the trained deep reinforcement learning algorithm, and generating the optimal charging electric quantity by the deep reinforcement learning algorithm to charge the battery. The optimal charging current can be automatically generated according to the real-time information of the battery, complex calculation is not needed, the robustness is high, the method can adapt to battery charging under different environment temperatures and initial SOC, and the optimal charging strategy of battery charging is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of charging piles, and in particular relates to a charging and discharging control strategy based on power battery performance attenuation. Background Art

[0002] Electric vehicles have been rapidly developed due to their great potential in energy conservation and environmental protection. Lithium-ion batteries are the main energy storage source for electric vehicles due to their outstanding characteristics such as high energy density, no memory effect, and environmental friendliness. However, how to achieve fast and efficient charging is a key issue in energy conversion and management of electric vehicles. A well-designed charging strategy can protect the battery from overheating, extend the battery's cycle life, and improve energy conversion efficiency. In addition, by using a suitable battery charging protocol, the inevitable economic losses during the operation of electric vehicles can be reduced.

[0003] Existing solutions usually rely on optimizing economic charging from the perspective of the power grid, but pay little attention to the interests of electric vehicle owners, nor the economic losses caused to users by battery degradation during battery charging. Or, such as constant current and constant voltage charging algorithms and multi-stage constant current charging algorithms, they focus more on fast charging without violating physical constraints, but do not consider the battery damage and power loss caused by charging, thus affecting the interests of users. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, a charge and discharge control strategy based on power battery performance attenuation is provided.

[0005] The technical solution adopted by the present invention to solve its technical problem is: This technical solution proposes a charge and discharge control strategy based on power battery performance attenuation, including: Based on the electrochemical-thermal-aging coupling model, the deep reinforcement learning algorithm is trained to obtain the optimal battery charging strategy; The grid’s time-of-use electricity price and battery status are input into the trained deep reinforcement learning algorithm, which generates the optimal charging power to charge the battery.

[0006] Preferably, the battery status includes battery damage, battery temperature and battery state of charge.

[0007] Preferably, during the training process, a total reward function is constructed, and the formula is as follows: (1); In the formula, R ( k ) represents the total charging reward function, represents the reward function coefficient, Indicates the user's loss reward during the charging process, is a smooth reward function, is the voltage reward function, For charging speed bonus, For temperature rewards, Rewards for charging goals; An electrochemical-thermal-aging coupled model is used for training to maximize the total reward function.

[0008] Preferably, the total reward function is constrained based on battery charging safety, and battery charging safety includes voltage constraints and temperature constraints, and the formula is as follows: (2); (3); In the formula, and Respectively represent the upper and lower cut-off voltages of the battery. is the battery temperature reward function, is the upper battery cutoff core temperature.

[0009] Preferably, the total reward function is constrained based on the charging speed, and the formula is as follows: (4); (5); In the formula, is the maximum charging current that the battery can withstand, Indicates the battery charging target SOC, It is the SOC at the current moment.

[0010] Preferably, the total reward function is constrained based on battery degradation and minimum energy consumption, and the formula is as follows: (6); In the formula, Indicates the user's loss during the charging process, including the battery loss part and power loss ; The battery damage part is expressed as: (7); In the formula, is the price of a new battery, It is the price at which the battery is sold for cascade utilization after reaching the retirement standard. It is the retirement standard of different batteries; is the damage of the battery during charging, which can be obtained through formula (8): (8); In the formula, is the rated capacity of the battery, represents the duration of the current, It is expressed as: (9); The power loss part is expressed as: (10); (11); In the formula, , and represents the ohmic internal resistance of the battery, Indicates the battery charging current at the current moment.

[0011] Preferably, the total reward function is constrained based on the charging current fluctuation, and the formula is as follows: (12); In the formula, and Represent the battery charging current in the current and previous control cycles respectively. The smooth reward function is expressed by the difference between two control cycles of the battery. The larger the difference between the two charging cycles, the smaller the reward function value.

[0012] Preferably, the time-of-use electricity price of the power grid includes peak moments with high electricity prices and valley moments with low electricity prices. During peak moments, a charging strategy of a high rate when the battery is low in power and a low rate when the battery is medium or high in power is adopted. During valley moments, a strategy of a low rate when the battery is low in power and a high rate when the battery is medium or high in power is adopted.

[0013] Compared with the prior art, the present invention has the following advantages: This application inputs the time-of-use electricity price and battery status of the power grid into the trained deep reinforcement learning algorithm. The deep reinforcement learning algorithm is combined with the battery degradation mechanism, comprehensively considers parameters such as charging damage, battery attenuation, and safety constraints, and autonomously generates the optimal charging current without the need for complex calculations. It has strong robustness and can adapt to battery charging under different ambient temperatures and initial SOCs. Without violating the charging constraints, from the perspective of electric vehicles and battery owners, it fully considers the economic cost of power loss and battery degradation, and can achieve the optimal battery charging strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which: Figure 1 It is a schematic diagram of the model training process in the present invention; Figure 2 It is a schematic diagram of the battery execution process in the present invention; Figure 3 is the battery charging strategy at peak times in the present invention; Figure 4 This is the battery charging strategy during the valley moment in the present invention. DETAILED DESCRIPTION

[0015] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0016] like Figure 1-Figure 4 As shown, this embodiment proposes a charge and discharge control strategy based on power battery performance attenuation, including: Based on the electrochemical-thermal-aging coupling model, the deep reinforcement learning algorithm is trained to obtain the optimal battery charging strategy; Compared with ordinary electrochemical models, the electrochemical-thermal-aging coupling model can capture the battery degradation characteristics and simulate the changes in parameters such as internal resistance during the battery degradation process. Therefore, the trained model is more accurate.

[0017] The time-of-use electricity price of the power grid and the battery status are input into the trained deep reinforcement learning algorithm, which generates the optimal charging power to charge the battery. The battery status includes battery damage, battery temperature and battery charge state.

[0018] For the deep reinforcement learning algorithm training, during the reinforcement learning training process, the electrochemical-thermal-aging coupling model acts as the external environment, namely the battery. During the charging process, different time-of-use electricity prices are randomly generated externally and input into the deep reinforcement learning algorithm. The deep reinforcement learning algorithm generates a corresponding charging strategy based on the electricity price and battery status. The battery model generates the corresponding status based on the charging strategy. The deep reinforcement learning algorithm evaluates the pros and cons of the charging strategy based on the reward function combined with the battery status, and updates the network parameters to maximize the reward obtained by reinforcement learning, thereby obtaining the best battery charging strategy.

[0019] The deep reinforcement learning algorithm is combined with the battery degradation mechanism, comprehensively considers parameters such as charging damage, battery degradation, safety constraints, and the time-of-use electricity price of the power grid, provides the optimal charging strategy, and autonomously generates the optimal charging current. The charging algorithm based on the degradation mechanism can achieve the least battery degradation during the battery charging process and the lowest user-side cost.

[0020] It does not require complex calculations, has strong robustness, and can adapt to battery charging under different ambient temperatures and initial SOCs. Without violating charging constraints, it can From the perspective of the battery owner, the optimal charging strategy for the battery can be achieved by fully considering the economic costs of power loss and battery degradation.

[0021] During the training process, the total reward function is constructed as follows: (1); In the formula, R ( k ) represents the total charging reward function, Represents the reward function coefficient, whose value is allocated according to the importance of the reward. For example, voltage and temperature are safety constraints, and their reward functions are more important, so and It is numerically greater than other rewards. Indicates the user's loss reward during the charging process, is a smooth reward function, is the voltage reward function, is the charging speed reward, For temperature rewards, Rewards for charging goals; An electrochemical-thermal-aging coupled model is used for training to maximize the total reward function.

[0022] The total reward function is constrained based on battery charging safety. Battery charging safety includes voltage constraints and temperature constraints. The formula is as follows: (2); (3); In the formula, and They represent the upper and lower cut-off voltages of the battery, respectively. When the battery exceeds the voltage range, it will cause overcharging or over-discharging. In order to allow the battery to fully explore and learn the boundary conditions during the battery charging process, the voltage reward function is also set to a soft constraint mode; It is a battery temperature reward function, which aims to prevent the battery temperature from exceeding the safety constraint, causing irreversible damage to the battery, or even causing safety accidents such as thermal runaway. It is the upper limit cutoff core temperature of the battery. When the battery core temperature exceeds this limit, it may cause irreversible damage to the battery or even safety risks such as thermal runaway.

[0023] Voltage constraints and temperature constraints are designed to ensure battery charging safety. When the battery exceeds the required constraint range, certain penalties are imposed to prevent safety accidents such as overcharging and thermal runaway.

[0024] The total reward function is constrained based on the charging speed, and the formula is as follows: (4); (5); In the formula, is the maximum charging current that the battery can withstand, Indicates the battery charging target SOC, which is set to 80% in this experiment. It is the SOC at the current moment.

[0025] The goal of the charging speed constraint is to maximize the battery charging speed without violating the maximum charging current or maximum charging power to ensure the fastest charging speed. The reward function makes the battery charging current / power close to the maximum allowed charging current / power, and when the charging SOC reaches the target SOC, the charging algorithm is guided to charge quickly based on a larger reward.

[0026] The total reward function is constrained based on battery degradation and minimum energy consumption, and the formula is as follows: (6); In the formula, Indicates the user's loss during the charging process, including the battery loss part and power loss ; The battery damage part is expressed as: (7); In the formula, is the price of a new battery. According to market research, the price of an A123 26650 M1B is 35 RMB. It is the price at which the battery is sold for cascade utilization after reaching the retirement standard. It is the retirement standard for different batteries. Generally speaking, power batteries are retired when they reach 80% SOH, but considering the differences between different batteries, retirement index parameters are set.

[0027] is the damage of the battery during charging, which can be obtained through formula (8): (8); In the formula, is the rated capacity of the battery, represents the duration of the current, It is expressed as: (9); The power loss part is expressed as: (10); (11); In the formula, , and represents the ohmic internal resistance of the battery, Indicates the battery charging current at the current moment.

[0028] The total reward function is constrained based on the charging current fluctuation, and the formula is as follows: (12); In the formula, and Represent the battery charging current in the current and previous control cycles respectively. The smooth reward function can be expressed by the difference between two control cycles of the battery. The larger the difference between the two charging cycles, the smaller the reward function value.

[0029] During the charging process, the current needs to be kept as stable as possible to prevent excessive current fluctuations from affecting the power grid and causing significant damage to the battery. The current fluctuation reward function is set as the difference between the currents of adjacent control cycles. The larger the difference between the currents of adjacent cycles, the greater the current fluctuation.

[0030] The economic losses during the charging process mainly come from two parts: one part is the electric energy consumed during the charging process, which is billed through the power grid; the second part is the loss at the vehicle battery end. The battery will inevitably be damaged during the charging process. Although this part is not directly charged, for electric vehicles, when the available capacity of the battery decays to about 80%, the battery needs to be replaced.

[0031] Batteries account for a large proportion of the components of electric vehicles, so for users, battery damage during charging cannot be ignored. This part is invisible consumption, and most charging strategies ignore this part of economic losses. The grid price varies greatly when charging at different time periods. The peak electricity price is almost 4 times that of the battery at the valley time. Therefore, the battery charging strategy is different at different charging times.

[0032] When the battery is in the low power range, the battery can withstand large current charging and the battery damage is relatively small, but the battery internal resistance is large and the power loss is relatively large. When the battery is in the high power range, although the battery internal resistance is small, at this time, when charging at the same rate or power, the battery damage is much higher than that in the low power range.

[0033] The time-of-use electricity price of the power grid includes peak hours with high electricity prices and valley hours with low electricity prices. At peak hours, a high rate charging strategy is adopted when the battery is low in power and a low rate charging strategy is adopted when the battery is medium or high in power, so as to increase the energy loss of the battery and reduce the energy loss of the battery. At valley hours, a low rate charging strategy is adopted when the battery is low in power and a high rate charging strategy is adopted when the battery is medium or high in power, so as to reduce energy loss and minimize the total economic price.

[0034] The input of the charging strategy is the battery status in the electric vehicle, which mainly includes state variables such as voltage, current, temperature, SOC, etc. The charging strategy outputs control instructions to the charging pile, which charges the car according to the instructions, monitors the state parameters of the electric vehicle, and feeds back to the charging strategy.

[0035] Figure 3 and Figure 4 These are the battery charging strategies for peak time and valley time in the present invention. It can be seen that the proposed charging strategy can realize charging the battery according to the time-sharing electricity price of the electric vehicle charging pile at different times. In addition, the decrease in the charging rate of the battery in the later stage is the result of the regulation of the battery temperature and voltage limit. In the later stage of charging, the battery temperature rises and the battery is close to the upper limit cut-off voltage, so at this time, the battery charging rate should be reduced to prevent internal short circuit and other faults, causing battery damage, and even fire and combustion accidents in severe cases.

[0036] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A charge and discharge control strategy based on power battery performance attenuation, characterized in that: include: Obtain the optimal battery charging strategy based on the electrochemical-thermal-aging coupling model; The grid’s time-of-use electricity price and battery status are input into the trained deep reinforcement learning algorithm, which generates the optimal charging power to charge the battery.

2. A charge and discharge control strategy based on power battery performance attenuation according to claim 1, characterized in that: The battery status includes battery damage, battery temperature and battery state of charge.

3. A charge and discharge control strategy based on power battery performance attenuation according to claim 1, characterized in that: During the training process, the total reward function is constructed as follows: （1）； In the formula, R ( k ) represents the total charging reward function, represents the reward function coefficient, Indicates the user's loss reward during the charging process, is a smooth reward function, is the voltage reward function, For charging speed bonus, For temperature rewards, Rewards for charging goals; An electrochemical-thermal-aging coupled model is used for training to maximize the total reward function.

4. A charge and discharge control strategy based on power battery performance attenuation according to claim 3, characterized in that: The total reward function is constrained based on battery charging safety. Battery charging safety includes voltage constraints and temperature constraints. The formula is as follows: （2）； （3）； In the formula, and Respectively represent the upper and lower cut-off voltages of the battery. is the battery temperature reward function, is the upper battery cutoff core temperature.

5. A charge and discharge control strategy based on power battery performance attenuation according to claim 3, characterized in that: The total reward function is constrained based on the charging speed, and the formula is as follows: （4）； （5）； In the formula, is the maximum charging current that the battery can withstand, Indicates the battery charging target SOC, which is set to 80% in this experiment. It is the SOC at the current moment.

6. A charge and discharge control strategy based on power battery performance attenuation according to claim 3, characterized in that: The total reward function is constrained based on battery degradation and minimum energy consumption, and the formula is as follows: （6）； In the formula, Indicates the user's loss during the charging process, including the battery loss part and power loss ; The battery damage part is expressed as: （7）； In the formula, is the price of a new battery, It is the price at which the battery is sold for cascade utilization after reaching the retirement standard. It is the retirement standard of different batteries; is the damage of the battery during charging, which can be obtained through formula (8): （8）； In the formula, is the rated capacity of the battery, represents the duration of the current, It is expressed as: （9）； The power loss part is expressed as: （10）； （11）； In the formula, , and represents the ohmic internal resistance of the battery, Indicates the battery charging current at the current moment.

7. A charge and discharge control strategy based on power battery performance attenuation according to claim 3, characterized in that: The total reward function is constrained based on the charging current fluctuation, and the formula is as follows: （12）； In the formula, and Represent the battery charging current in the current and previous control cycles respectively. The smooth reward function is expressed by the difference between two control cycles of the battery. The larger the difference between the two charging cycles, the smaller the reward function value.

8. The charge and discharge control strategy based on power battery performance attenuation according to claim 1 is characterized in that: The power grid's time-of-use electricity prices include peak hours with high electricity prices and valley hours with low electricity prices. During peak hours, a charging strategy is adopted that uses a large rate when the battery's power is low, and a small rate when the battery's power is medium or high. During valley hours, a strategy is adopted that uses a small rate when the battery's power is low, and a large rate when the battery's power is medium or high.

Images