Abstract
Essential decision-making tasks such as power management in future vehicles will benefit from the development of artificial intelligence technology for safe and energy-efficient operations. To develop the technique of using neural network and deep learning in energy management of the plug-in hybrid vehicle and evaluate its advantage, this article proposes a new adaptive learning network that incorporates a deep deterministic policy gradient (DDPG) network with an adaptive neuro-fuzzy inference system (ANFIS) network. First, the ANFIS network is built using a new global K-fold fuzzy learning (GKFL) method for real-time implementation of the offline dynamic programming result. Then, the DDPG network is developed to regulate the input of the ANFIS network with the real-world reinforcement signal. The ANFIS and DDPG networks are integrated to maximize the control utility (CU), which is a function of the vehicle's energy efficiency and the battery state-of-charge. Experimental studies are conducted to testify the performance and robustness of the DDPG-ANFIS network. It has shown that the studied vehicle with the DDPG-ANFIS network achieves 8% higher CU than using the MATLAB ANFIS toolbox on the studied vehicle. In five simulated real-world driving conditions, the DDPG-ANFIS network increased the maximum mean CU value by 138% over the ANFIS-only network and 5% over the DDPG-only network.
Highlights
R ECENT advances in artificial intelligence (AI) and informatics have significantly promoted the development of connected and autonomous vehicles [1]
The proposed deep deterministic policy gradient (DDPG)-adaptive neuro-fuzzy inference system (ANFIS) network was implement in the prototype controller with two steps: 1) the network was built with MATLAB/Simulink with its inputs/outputs connected to the CAN interface blocks that are provided in the Simulink real-time block set and 2) the Simulink model was compiled into executive code with the Simulink code generator for real-time control
By implementing the control models in the HiL testing platform, the real-time performance of the fuzzy model obtained by global K-fold fuzzy learning (GKFL)-9-genetic algorithms (GAs) (κ = 9, using GA for learning) is compared in Fig. 8 with both benchmark strategy and the model obtained by conventional Conv-particle swarm optimization (PSO) (ANFIS toolbox with PSO)
Summary
R ECENT advances in artificial intelligence (AI) and informatics have significantly promoted the development of connected and autonomous vehicles [1]. Most research on RL-based power management control focus on learning from scratch [39], [40] This approach requires a long time to develop a proper control policy, so it is not practical in real-world applications [41]. To enable knowledge implementation and transfer in power management of the plug-in hybrid electric vehicle (PHEV), this article proposes an adaptive learning network, which incorporates a deep deterministic policy gradient (DDPG) network [50] with an ANFIS network. Experimental evaluations are conducted to show how advanced artificial neural networks and learning systems help improve control performance in real-world driving. This work has two main contributions: 1) A new method named global K-fold fuzzy learning (GKFL) is proposed to build the ANFIS network that is used to implement the knowledge learned from offline DP to real-time control.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
