Fast algorithm for matrix inversion in generalized predictive control

New fast algorithm for the inverse of the matrix in generalized predictive control(GPC) is presented to reduce the computational load of matrix inverse in GPC, and improve the real-time ability of system. When maximum prediction horizon p is equal to control horizon m, the inverse matrix in control increment of GPC is transformed Toeplitz matrix, then the Trench-Zohar algorithm is applied to calculate the inverse ; when p is not equal to m, the matrix is decomposed based on Cholesky factorization firstly, the matrix inverses for lower triangular is used secondly, finally the result is gotten by matrix multiplication. The theory and the example indicate the effectiveness of the above algorithm. Moreover, the method is not only simple, but also easy to programme.