A Rotor-Flux-Oriented Vector Control of an AC Drive with an Induction Motor using the Progressive Learning Neural Network

This paper presents a rotor flux estimator in a rotor-flux-oriented vector control of an ac drive with induction motor based on the use of the progressive learning neural network (PLN). This neural network has been employed for its versatility, quick learning phase and adaptive capability. By properly choosing the training set, the neural estimator is made insensitive to the variations of the rotor time constants. This neural network is particularly suitable for this task as a result of its inherent clustering capability. Various tests under conditions different from those presented in the training phase have shown that the PLN gives a satisfactory generalisation capabilty, although with fewer training data than those generally required by other supervised neural networks.