This text study to have memory to be non- fragile to control fuzzy system theoretically. The fuzzy system has high, stupid and excellent of good performance characteristic of precision; Only if controlling fragily can make the controller overcome one's own uncertainty, the stability of the security system; There is feedback control of memory that can be solved and lagged behind to control the unfavourable factor brought to the system. Because of these characteristics, this text utilizes Lyapunov stability theory, the method to use linear matrix inequality (LMI) at first, to T-S fuzzy model, have analyzed that stability not fragile and systematic of memory and stupid excellent state feedback the question that the controller is designed. On this basis, to to have memory to be non- fragile to * control, have memory to be non- fragile to assuare cost control, have memory to be non- fragile to * - * control research system, have obtained abundant condition and design method that these controllers exist. Analytic process adopt variable vary with Schur mending lemmaing parametering to prove LMI linear. Meanwhile, this literary grace carries on programming with LMI toolbox in MATLAB, ask and solve the controller to gain matrix; Regard as through SIMULINK toolbox to number value example go on, ask solve and emulation, for prove subject study algorithm and result received, further prove the resulting result has possessed certain validity and feasibility. Finally, summarize this research content, method and means, point out one existing while studying Some insufficient and has needed to be further improved questions, have planned the developing direction of the fuzzy control system.