Neural networks controlled PWM inverters

A high performance PWM inverter using neural networks controller is presented in this paper. Owing to the learning ability and robust capability of neural networks, the switching pattern of the PWM inverter can be optimized under the linear and non-linear load conditions. Thus, a high quality and low harmonics sinusoidal output voltage is expected to be obtained. The proposed PWM inverter provides the feasible characteristics and the memory size needed for hardware implementation is small since the learned informations are parallel distributed in neural network's weights. The simulation results of PWM inverter presented are shown to demonstrate its excellent performance.