A multi-time scales model and control for hybrid stochastic production systems with quadratic cost

A hybrid stochastic production system corresponding to a situation where a basically deterministic production system, the fast subsystem, is subject to infrequent model disruptions occurring randomly, the slow subsystem is formulated with multi-time scales. One time scale is 1, the processing time scale, which is frequently and represents fast model. Another time scale is /spl epsiv//sup -1/, system failure time scale, which is infrequently and stands for slow model of the system. On different time scales, the long-run average cost function is decomposed into two sub-objective functions. Based on the two sub-objective functions, the optimal control policy of the system is obtained by using approximation numerical technique. Numerical experiments show its merit.