Research on energy management strategy of hybrid electric mining trucks based on SAC

In this paper, a 150-ton series hybrid mining truck is taken as the research object, and an energy management control strategy based on the Soft Actor-Critic (SAC) deep reinforcement learning algorithm is proposed by taking into account the effects of road gradient and load on the dynamic performance of the truck. The SAC-based energy management control strategy is built to solve the coordinated optimization problem of the mutual constraints between the fuel consumption of the whole truck and the cyclic lifespan of the power battery. The speed, acceleration, mass and gradient, power battery SOC, and charging/discharging multiplier that characterize the driving state of the truck are taken as the state variables in the simulation model. The engine power and rotational speed that characterize the working performance of the power components of the truck are taken as the action variables. The reward variables are the fuel consumption of the whole truck, the power battery SOC, and the cyclic lifespan that characterizes the power source's use performance. The simulation results show that the SAC-based energy management strategy improves fuel consumption by 5.26% and increases power battery loss by 5.30%, compared with the dynamic programming-based energy management strategy.