Identification and Control Using Iterative Broad Learning for Nonlinear Digital Dynamic Systems

• Main Authors: Chien-Cheng Yu, 余建政
• Other Authors: 蔡清池
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi/login?o=dnclcdr&s=id=%22107NCHU5441100%22.&searchmode=basic

博士 === 國立中興大學 === 電機工程學系所 === 107 === This dissertation presents design methodologies and techniques for system identification and intelligent control of digital nonlinear dynamic systems. A new BLS identifier with a set of iterative learning algorithms is proposed to on-line learn the dynamic input-output relationships of a class of digital SISO nonlinear system models. Based on this BLS model, three different types of intelligent adaptive predictive controllers, which are respectively BLS-APPID, BLS-PIDLC and BLS-NMPC, are well proposed to achieve superior control performance in presence of exogenous disturbances. The learning algorithms for each adaptive BLS-based controller have been rigorously derived, and the overall adaptive real-time identification and control algorithm are also well presented to achieve superior control performance. The asymptotical convergence and/or asymptotical stability of the BLS-based identifiers and intelligent adaptive controllers are stringently proven via the well-known discrete-time Lyapunov stability theory. The effectiveness and superiority of these three constructed control approaches are well demonstrated by performing numerical simulations on two well-known digital nonlinear time-delay processes, in order to achieve step-like disturbance rejection and set-point tracking. The developed methods and system techniques may provide solid and useful references for researchers and engineers working in the area of intelligent control.