Energy management is a fundamental task and challenge of plug-in split hybrid electric vehicle (PSHEV) research field because of the complicated powertrain and variable driving conditions. Motivated by the foresight of intelligent vehicle and the breakthroughs of deep reinforcement learning framework, an energy management strategy of intelligent plug-in split hybrid electric vehicle (IPSHEV) based on optimized Dijkstra’s path planning algorithm (ODA) and reinforcement learning Deep-Q-Network (DQN) is proposed to cope with the challenge. Firstly, a gray model is used to predict the traffic congestion of each road and the length of each road calculated in the traditional Dijkstra’s algorithm (DA) is modified for path planning. Secondly, on the basis of the predicted velocity of each road, the planned velocity is constrained by the vehicle dynamics to ensure the driving security. Finally, the planning information is inputted to DQN to control the working mode of IPSHEV, so as to achieve energy saving of the vehicle. The simulation results show the optimized path planning algorithm and proposed energy management strategy is feasible and effective.

In the high latitudes, the icy patches on the road are frequently generated and have a wide distribution, which are difficult to remove and obviously affect the normal usage of the highways, bridges and airport runways. Physical deicing, such as microwave (MW) deicing, help the ice melt completely through heating mode and then the ice layer can be swept away. Though it is no pollution and no damage to the ground, the low efficiency hinders the development of MW deicing vehicle equipped without sufficient speed. In this work, the standard evaluation of deicing is put forward firstly. The intensive MW deicing is simplified to ice melting process characterized by one-dimensional slab with uniform volumetric energy generation, which results in phase transformation and interface motion between ice and water. The heating process is split into the superposition of three parts -non-heterogeneous heating for ground without phase change, heat transfer with phase change and the heat convection between top surface of ice layer and flow air. Based on the transient heat conduction theory, a mathematical model, combining electromagnetic and two-phase thermal conduction, is proposed in this work, which is able to reveal the relationship between the deicing efficiency and ambient conditions, as well as energy generation and material parameters. Using finite difference time-domain, this comprehensive model is developed to solve the moving boundary heat transfer problem in a one-dimensional structured gird. As a result, the stimulation shows the longitudinal temperature distributions in all circumstances and quantitative validation is obtained by comparing simulated temperature distributions under different conditions. In view of the best economy and fast deicing, these analytic solutions referring to the complex influence factors of deicing efficiency demonstrate the optimal matching for the new deicing design.

No abstract

No abstract