A novel output power control for variable-speed switched reluctance generators using artificial neural network

Abstract

This paper presents a novel output power control method for variable-speed switched reluctance generators (SRG). In the single pulse mode, the conduction angles dictate the waveform of phase current which directly influences the output power. With the proposed method, the conduction angles can be estimated from the output power profiles at several speeds using an artificial neural network (ANN). The output power profiles are simulated from a dynamic model of an 8/6 SRG and used as the training data of the network. The inputs of the network are the output power and rotor speed, and its outputs are the turn-on and turn-off angles. The proposed control scheme was developed by incorporating the trained network into the dynamic model before the simulation in MATLAB/Simulink. The experiments with the 8/6 SRG were carried out using TMS320F2812 DSP to control the SRG drive system. In addition, the optimal trained weights and biases of the network are utilized in the DSP program. The comparisons of the simulation and experimental results are made to verify the proposed method.