Prediction-Based Game-Theoretic Strategy for Energy Management of Hybrid Electric Vehicles

Abstract

This article studies a prediction-based energy management for onboard hybrid energy storage system (HESS), combining engine-generator, battery, and ultracapacitor. Each of these energy sources has a specific utility function to represent its unique preference. Thus, a game-theoretic strategy is presented to model the different preferences of these energy sources and their interactions, and hence to properly dispatch the power load demand among them. To further improve this power dispatch, i.e., the energy management, that may be influenced by the fluctuation of the uncertain power load demand, a prediction is included in the basic game-theoretic strategy to form a prediction-based game-theoretic strategy. The power load demand can be derived from the velocity in HESS and the velocity prediction is implemented by a long short-term memory network. An improvement on the accuracy of this prediction is achieved by utilizing feature extraction and time-series analysis. A multiple time-series method is newly applied to group the input features according to the target prediction horizon. The solution, i.e., Nash equilibrium, of this proposed strategy is reached based on the best response functions of the energy sources and its performance is quantified by four criteria. Short-distance and long-distance driving in a broader scope are analyzed in simulation. Both the simulation and experiment results demonstrate the efficiency of the proposed strategy to smoothen the battery power with decreasing 0.01% in <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\sigma _{pb}$</tex-math></inline-formula> (i.e., prolong battery life), to reduce the engine-generator power with reducing 0.01% in <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\mu _{Eg}$</tex-math></inline-formula> (i.e., deplete fossil fuels), and to lower the driving costs. Moreover, the robustness and sensitivity of the proposed strategy are validated through case studies with increasing velocity prediction error.