Accession Number : ADA564788


Title :   Intelligent Hybrid Vehicle Power Control - Part 1: Machine Learning of Optimal Vehicle Power


Descriptive Note : Journal


Corporate Author : MICHIGAN UNIV DEARBORN


Personal Author(s) : Murphey, Yi L ; Park, Jungme ; Chen, ZhiHang ; Kuang, Ming ; Masrur, Abul ; Phillips, Anthony


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/a564788.pdf


Report Date : 30 Jun 2012


Pagination or Media Count : 24


Abstract : Energy management in Hybrid Electric Vehicles (HEV) has been actively studied recently because of its potential to significantly improve fuel economy and emission control. Because of the dual-power-source nature and the complex configuration and operation modes in a HEV, energy management is more complicated and important than in a conventional vehicle. Most of the existing vehicle power optimization approaches do not incorporate knowledge about driving patterns into their vehicle energy management strategies. Our approach is to use machine learning technology combined with roadway type and traffic congestion level specific optimization to achieve quasi-optimal energy management in hybrid vehicles. In this series of two papers, we present a machine learning framework that combines Dynamic Programming with machine learning to learn about roadway type and traffic congestion level specific energy optimization, and an integrated online intelligent power controller to achieve quasi-optimal energy management in hybrid vehicles. These two papers cover the modeling of power flow in HEVs, mathematical background of optimization in energy management in HEV, machine learning algorithms and real-time optimal control of energy flow in a HEV. This first paper presents our research in machine learning for optimal energy management in HEVs. We will present a machine learning framework, ML_EMO_HEV, developed for the optimization of energy management in a HEV, machine learning algorithms for predicting driving environments and generating optimal power split for a given driving environment. Experiments are conducted based on a simulated Ford Escape Hybrid vehicle model provided by Argonne National Laboratory's PSAT (Powertrain Systems Analysis Toolkit). Based on the experimental results on the test data, we can conclude that the neural networks trained under the ML_EMO_HEV framework are effective in predicting roadway type and traffic congestion levels, in predicting driving trend and


Descriptors :   *ELECTRIC PROPULSION , *FUEL CONSUMPTION , *HYBRID SYSTEMS , *VEHICLES , ALGORITHMS , CONGESTION , COST EFFECTIVENESS , DYNAMIC PROGRAMMING , EMISSION CONTROL , ENERGY MANAGEMENT , ENVIRONMENTS , ESCAPE SYSTEMS , EXPERIMENTAL DATA , LEARNING MACHINES , NEURAL NETS , REAL TIME , ROADS


Subject Categories : Surface Transportation and Equipment
      Electric and Ion Propulsion
      Fuels
      Rocket Engines


Distribution Statement : APPROVED FOR PUBLIC RELEASE