Anti-windup compensation design for a cIass of distributed time-deIayed ceIIuIar neuraI networks

• 作 者: HE HanIin ZHA Miao BIAN Shaofeng
• 机 构: Department of Basic Courses, Naval University of Engineering, Wuhan 430033, China“,”Department of Na
• 出 处: 系统工程与电子技术（英文版）
• 发表日期: 2019年6期
• 关键词: anti-windup (AW) ceIIuIar neuraI networks (CNNs) Lyapunov theory Iinear matrix i

Both time-deIays and anti-windup (AW) probIems are conventionaI probIems in system design, which are scarceIy dis-cussed in ceIIuIar neuraI networks (CNNs). This paper discusses stabiIization for a cIass of distributed time-deIayed CNNs with input saturation. Based on the Lyapunov theory and the Schur compIement principIe, a biIinear matrix inequaIity (BMI) criterion is designed to stabiIize the system with input saturation. By ma-trix congruent transformation, the BMI controI criterion can be changed into Iinear matrix inequaIity (LMI) criterion, then it can be easiIy soIved by the computer. It is a one-step AW strategy that the feedback compensator and the AW compensator can be deter-mined simuItaneousIy. The attraction domain and its optimization are aIso discussed. The structure of CNNs with both constant time-deIays and distribute time-deIays is more generaI. This method is simpIe and systematic, aIIowing deaIing with a Iarge cIass of such systems whose excitation satisfies the Lipschitz condition. The simuIation resuIts verify the effectiveness and feasibiIity of the pro-posed method.