Artificial neural network based soft-starter for induction motor

The necessity of soft-starter is increasing day by day to reduce the starting current & to maintain the torque smoothly according to the load requirement. Now intelligent soft-starter is developed to improve the performance of conventional starter. This paper focused on the designing of an artificial neural network controlled soft-starter. Back-propagation algorithm is used as learning algorithm in the artificial neural network. Error correcting capability of this learning algorithm makes it more suitable to use in neural network. For comparative performance analysis two different types of back propagation algorithms are used in the neural network learning process. According to condition of learning rate parameter gradient descent with momentum back-propagation algorithm provide better response. A comparative study between conventional starting method (Direct on Line)and proposed soft-starter. Artificial Neural Network controlled soft-starter is able to reduce starting current compared with DOL method & able to accelerate the load at starting period efficiently compared with star-delta starting method.