Abstract:

For discrete-time Markov jump linear systems, an optimal control problem is studied. First, a quadratic cost function is determined, and then the optimal controller and Riccati difference equation are designed under complete states information，which are based on Bellman stochastic dynamic programming and the characters of Markov jump systems. Then the method is further extended to incomplete states information. The optimal control algorithm is presented based on a posterior probability density function deciding by observation vector. Finally, an example with incomplete states information is given to verify the effectiveness of the presented optimal controller.