Multi-level intelligent control using three-step learning strategy: analysis and implementation

Combining fuzzy control, three-step learning strategy and supervisory control, the synthetic method of intelligent control (SMIC) based on system response characteristics is proposed and analyzed theoretically. The SMIC has been implemented in our low cost distributed computer control system (DCCS). A simple example of an electric furnace control system demonstrated the applicability of our low cost DCCS utilizing SMIC. The analyses and experiments show that the proposed SMIC has the capability of self-extracting fuzzy control rules or parameters from nonlinear plants with time delays, especially it can guarantee the learning convergence and the control stability.