Research On Control Strategy Of Electric Vehicle Heat Pump Air Conditioning System Based On Machine Learning

In this paper,the electric vehicle heat pump air conditioning system as research background.The system is a complex system with large lag,non-linearity and variable operating conditions.Conventional control methods are difficult to accurately meet the comfort requirements of the passenger compartment.Limited by the battery energy density,electric vehicle subsystem should minimize power consumption,particularly power consumption of the thermal management system.Therefore,the in-depth study of the heat pump air-conditioning system and the heat exchange mechanism of components,the relevant algorithms in machine learning and the current popular control methods are used as the starting point,and the system control strategy is studied from the passenger cabin temperature and system power.Based on MATLAB/SIMULINK development environment,the heat pump air conditioning system simulation platform model is established.Based on historical experimental data and BP neural network,the compressor machine learning model is established to accurately predict compressor operating status.The remaining parts of the system is modeled based on thermodynamic principles and the establishment of various components coupled system simulation platform.Select experimental equipment,build the heat pump air conditioning system experimental bench,and verify the platform according to the heating/cooling test conditions.In compressor control strategies proposed neural network fuzzy control strategy,to make up for the lack of expert fuzzy control relies heavily on experience.Input of the current difference between the set temperature and the temperature difference between the rate of change in the passenger compartment as a controller to DC motor speed control of the compressor PWM(Pulse Width Modulation)duty ratio as an output,the compressor construct a two-dimensional neural network-fuzzy controller.Based trained neural network,adjusting the amount of the corresponding input membership function.In the heating/cooling conditions,the temperature of the passenger compartment under the control of neural network-fuzzy,fuzzy,and PID are compared and analyzed.The control accuracy of neural network-fuzzy is better than the other two control strategies.Meet the needs of the crew cabin.Selection neural network-fuzzy controller as the compressor control strategy.Combined with the electric vehicle heat pump air conditioning system and the passenger cabin model,the factors affecting the performance of the system are analyzed.According to the wind speed of the external heat exchanger 2~20m / s,the air volume of the internal heat exchanger 150~550 m3/h,the ambient temperature of-5,0,5 ?,the ambient temperature of 30,35,42 ? Simulations are performed for mass flow,compressor power consumption,system heating/cooling capacity,COP/EER performance parameter changes,and fuzzy neural networks under different operating conditions The temperature change of the passenger compartment under control is studied,and provide an optimization direction for the system control strategy.The system power is selected as the optimization target,and the PWM duty cycle of the motor speed driving the compressor,the external heat exchanger fan,and the internal heat exchanger fan is used as the control variable.Constraints are given based on ensuring the comfort of the passenger compartment and the characteristics of the compressor,the external heat exchanger fan,and the internal heat exchanger fan.An improved particle swarm algorithm is used as an optimization algorithm to optimize the system control strategy.After optimization,the maximum energy saving effect is 41.31%,which achieves the purpose of energy saving while satisfying the comfort of the passenger cabin of the heat pump air conditioning system.