| With the increasingly serious problems of energy shortage and environmental pollution,the electric vehicles(EVs)have become an important direction for the development of the automotive industry.However,the endurance of EVs and the temperature adaptability of the battery restrict the application of EVs.On the one hand,charging and discharging in high and low temperature environments lead to heat accumulation and lithium dendrite generation of battery respectively,thus causing internal short circuit,thermal runaway and other accidents.On the other hand,large acceleration or deceleration during driving and excessive cooling or heating during battery thermal management consume a lot of energy,resulting in a decrease in the cruising range of EVs.Therefore,to ensure the safety and stability of EVs and alleviate the anxiety of mileage,it is necessary to maintain battery at a suitable temperature and reduce energy consumption(EC)of EVs.Therefore,the model of battery thermal management system(BTMS)and vehicle EC system is built.The collaborative optimization strategy for the battery pack temperature and EC of EVs are studied in this paper.The main research contents of this paper are as follows.First,to achieve effective cooling and heating of the battery pack in high and low temperature environments respectively,a control-oriented third-order model of BTMS is established based on the law of energy conservation and the principle of reverse Carnot cycle.In this model,the battery pack and heat exchangers are lumped separately,which simplifies the complexity of the model.At the same time,the model fully describes the heat transfer relationships among coolant,refrigerant,and battery temperature based on three state variables.In addition,the EC model of vehicle is established in this paper.The model describes the relationships between vehicle speed,driving EC,and heat exchange rate(HER)of BTMS,as well as the relationship between vehicle EC and battery heat generation rate(HGR).The accuracy of BTMS and vehicle EC model is verified based on AVL Cruise.Then,the HGR and HER of the battery is closely coupled with the vehicle EC,which makes the solve of optimization problem in a long-time domain too complex.Therefore,a vehicle speed planning and battery temperature optimization strategy based on Iterative Dynamic Programming-Deep Q Network(IDP-DQN)is proposed.Firstly,according to the traffic information in the forward-looking domain,IDP is used to adjust the acceleration and then plan the optimal speed of EVs,to improve the energy utilization rate.The IDP proposed in this paper can reduce the grid density while reducing the boundary of the state quantity and control quantity in the iterative process,thereby reducing the calculation burden and improving the optimization performance.Then,aiming at the problem that the power and exponential terms in BTMS model cause the system to have strong nonlinearity,DQN is used to train the off-line model based on the planned speed and the established model.Thus,the optimal rotating speed of pump and compressor under different state variables can be obtained,to solve the problem of temperature optimization in high dimensional continuous state.Compared with traditional strategy,the effectiveness of IDP-DQN is verified.Finally,the speed planned by IDP in the forward-looking domain is difficult to cope with emergency,and the driving conditions used in DQN cannot fully reflect the actual traffic conditions.Therefore,a collaborative optimization strategy of the battery temperature and vehicle EC based on Learning-based Model Predictive Control(LMPC)is proposed to comprehensively consider the influence of global and real-time traffic status on battery temperature and vehicle EC.The residual model based on Gaussian process regression is integrated in LMPC.Therefore,the state prediction model can learn and update itself continuously during prediction to improve estimation accuracy.In upper LMPC,global optimal speed obtained by IDP is taken as the reference speed,and the driving torque is adjusted based on the obtained real-time traffic,to optimize the vehicle speed.The lower LMPC receives the optimal vehicle speed obtained by upper LMPC in the prediction time domain.And the relationship between temperature deviation and EC is balanced according to the objective function.Therefore,the influence of real-time vehicle speed on the heat generation and battery HER is fully considered,and the adjustment accuracy of the control quantity is guaranteed.At the same time,the target temperature of the lower LMPC is calculated by DQN based on global optimal vehicle speed,which reflects the slow change characteristics of the heat transfer process in BTMS under global conditions.A traffic simulation model is built on VISSIM software to realize the speed planning simulation of EVs.On this basis,the economics of EVs under two-layer LMPC optimization strategy and the IDP-PI-DQN strategy are compared.The results show that,compared with IDP-PI-DQN strategy,the average aging rate of battery pack in high temperature and low temperature environment is reduced by 6.8% and 5.5% respectively,and the EC of EV per100 kilometers is reduced by 10.3% and 11.1% respectively,which proves the superiority of the coordinated optimization strategy of battery temperature and EC for EVs. |
