engleski

Neural control of boost converters' input current

An application of neural networks in the cascade control structure of a boost converter is investigated. Inverse model of process realized using two-layered MLP neural network is used as the input current controller. A Levenberg–Marquardt learning algorithm is employed for off-line adjustment of the network weights and biases. This control structure ensures good performance in all operating regions and compensation of ripples in converter's input current caused by variations of the input voltage. Advantages of the proposed control structure are demonstrated through experimental comparison to the PI controller (with and without manually adjusted feed forward compensator).

power converters; current control; harmonics; neural networks; boost converters; inverse control

nije evidentirano