Energy-management system for a hybrid electric vehicle, using ultracapacitors and neural networks

A very efficient energy-management system for hybrid electric vehicles (HEVs), using neural networks (NNs), was developed and tested. The system minimizes the energy requirement of the vehicle and can work with different primary power sources like fuel cells, microturbines, zinc-air batteries, or other power supplies with a poor ability to recover energy from a regenerative braking, or with a scarce power capacity for a fast acceleration. The experimental HEV uses lead-acid batteries, an ultracapacitor (UCAP) bank, and a brushless dc motor with nominal power of 32 kW, and a peak power of 53 kW. The digital signal processor (DSP) control system measures and stores the following parameters: primary-source voltage, car speed, instantaneous currents in both terminals (primary source and UCAP), and actual voltage of the UCAP. When UCAPs were installed on the vehicle, the increase in range was around 5.3% in city tests. However, when optimal control with NN was used, this figure increased to 8.9%. The car used for this experiment is a Chevrolet light utility vehicle (LUV) truck, similar in shape and size to Chevrolet S-10, which was converted to an electric vehicle (EV) at the Universidad Catolica de Chile. Numerous experimental tests under different conditions are compared and discussed.