APPLICATION OF CERTAIN OPTIMIZATION TECHNIQUES FOR CAGE INDUCTION MOTOR DESIGN

Abstract

Two approaches are used for the design of cage induction motors. In the first approach the design of both the magnetic and electric circuits is carried out whereas in the second approach only the electric circuit is designed for a given fixed magnetic circuit. In the present work these approaches are termed as fresh and frame designs. In fresh design cost of active materials is taken as the objective function and performance items as constraints. The design variables which are most sensitive to the cost of active materials and performance have been judiciously chosen. They are '. specific electric loading, ratio core length to pole pitch, stator slot depth to width ratio, depth of stator core, specific magnetic loading, stator windine; current den sity and rotor winding current density. The objective func tion and constraints have been expressed in taring of the above variables. The design problem is stated as a nonlinear multivariable constrained optimization problem. Direct and indirect methods of optimization i.e. incremental search, Rosenbrock' g, Box's complex and interior penalty function methods have been employed to solve the problem. Optimized results obtained for several motor ratings are given. Salient features of the optimized design, and merits and demerits of the methods used have been discussed. Graphs showing convergence progress of each method are given

to support the discussion. Interior penalty function method has been attempted with several strategies for selecting initial values of penalty parameter and its effect on convergence is also discussed. In frame designed induction motors selection of design variables is made after testing their effective ness in controlling the cost of active materials and per formance. The variables are stack length, number of turns per phase and the winding wire cross section area. The mathematical programming problem formulated has been solved using incremental search, Box's complex and interior penalty function method. A new fast converging method consisting of two direct search methods, complex and incremental search, has been developed which can respond to continuous as well as discrete variables. Results obtained using all these methods have been discussed.