Development of electric vehicle battery capacity estimation using neuro-fuzzy systems
Development of Electric Vehicle Battery Capacity Estimation Using Neuro-Fuzzy Systems
for the Degree of Master of Philosophy at The University of Hong Kong
in September 2003
The growing concern with environmental protection and energy conservation has
encouraged the development of electric vehicle (BV) technologies, including battery charging methods and propulsion technologies. Ideally, EV batteries should use state of available capacity (SOAC) indicators instead of state-of-charge (SOC) indicators, but EV batteries are highly nonlinear, and this characteristic makes SOAC estimation difficult.
The objective of this study was to develop EV battery capacity estimation using neuro-fuzzy systems. The study adopted two main approaches: firstly, to develop battery capacity estimation for lead-acid battery using neural network (NN); and secondly, to develop battery capacity estimation for lead-acid battery, nickel metal hydride (Ni-MH) battery and lithium-ion (Li-Ion) battery using adaptive neuro-fuzzy inference system
The first approach involved usmg neural network (NN) to elucidate the relationship between the SOAC and the terminal voltage, discharge current, battery
surface temperature and discharged capacity. This enabled the SOAC to be estimated by mapping between the inputs and the output ofNN.
The second approach exploited the adaptive neuro-fuzzy inference system (ANFIS). In this approach, the discharged and regenerative capacity distribution was used to describe the discharge current profile in EV s. The capacity distribution and the battery surface temperature were selected as the inputs of the ANFIS while the output of the ANFIS was the SOAC. Consequently, the SOAC could be estimated by mapping between the inputs and the output of this ANFIS. For further improvement, the system inputs were modified and the results were given.
Experiments using the battery testing and evaluation system were performed to test the lead-acid battery, the Ni-MH battery and the Li-Ion battery. Various discharge current profiles at different battery surface temperatures were designed in order to replicate the battery operating conditions in EV s. Experimental data used for the development and verification of the proposed approaches were recorded, and comparisons between the estimated SOAC from the proposed approaches and recorded SOAC from the experimental data were made.
School:The University of Hong Kong
School Location:China - Hong Kong SAR
Source Type:Master's Thesis
Keywords:university of hong kong dissertations automobiles electric batteries neural networks computer science fuzzy systems
Date of Publication:01/01/2004