The L_∞ Parameter Identification Based On Linear Programming Recursive Algorithm

In the paper, we discuss the problem of optimal solution for the increase of constraints to the linear programming; the problem has a wide practical background in linear programming. In many modeling problems of control and system, such as the identification in L₁ and L_∞, the fitting of curves, parameter identification for uncertain system and bound identification can transfer the optimal solution for the increase of constraints to the linear programming.This paper discusses and analyses the changes of optimal solution for the increase of constraints to the linear programming from the view of convex region; Then one new algorithm based on convex region to solve this problem is presented, the algorithm cans quick judge contradiction constraining, redundant constraining and optimization of the problem; At last, its recursive algorithm is also presented.Research on the relations between identification and control is one of the most challenging fields in the control problem. The least squares algorithms require that error follow zero mean valued and homo-skedastcity normal distribution, but it is not applicable in some practical situation. So the L_∞ identification method is presented in the paper, which can transfer the optimal solution for the increase of constraints to the linear programming. In order to delete the redundant data of identification and the application of on-line identification, the paper also give the L_∞ parameter identification based on linear programming recursive algorithm. Finally experimental simulation and experimental study of algorithms are presented.