Variable structure-based covariance assignment for stochastic multivariable model reference systems

Based on the concept of variable structure control, this paper studies the error state covariance assignment problem for stochastic multivariable model reference systems. Using the invariance property of variable structure systems, the reference model input and the plant error term disappear on the sliding mode. With the aid of Ito formula, the controller is proposed. Combining the sliding phase and hitting phase of the system, the gain matrix G is derived to achieve the error state covariance assignment such that the system has good tracking response.