Tham khảo tài liệu 'frontiers in adaptive control part 6', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | 116 Frontiers in Adaptive Control numbers of fuzzy rules and tuning parameters. Furthermore the fuzzy consequent parts are adjusted by 33 where Y 50 0 104 f 10 and 0f 0 0 72x10. Furthermore the control parameters are chosen as n1 0-1 n2 20 kf 15 Am diag 10 5 5 p 5 and Q I6. Then after choosing Ts diag 40 20 10 Tse Te 2I3 TJ I3 for 37 the control gains are obtained as K 7-9 K diag and Pm diag by solving the LMI 39 . In this simulation the system begins at the oositĩon rrtol n 2 73 0 n 5n n n nlT And all hiivp zero initial velocities 1 p poson q 2 2 6326 3 O-AL I avc zero AL a. vetoc es .e. q 0 0 . The external disturbance is injected to the first joint of two robots as Td which is a square wave with amplitude and frequency Hz. According to Thm. 4 the simulation results of position and velocity tracking for the object are illustrated in Figs. 4 and 5 respectively. The internal force errors between the desired and actual internal force are shown in Fig. 6. The second joints of two robots are driven by torques illustrated in Fig. 7. For a comparison a traditional SFA-based controller is also constructed and applied to the cooperative robots where the Mamdani SFA takes q1e q 1t q1dl q 1dl and q as the premise variables. Furthermore the SFA-based control is set with the same initial conditions and feedback compensation part as the proposed controller but Ts 0 Te 0 Tse 0 and TJ 0 . Then the position tracking results for using Mamdani SFA and TS FFA based control are made as a comparison given in Fig. 8. Obviously the TS FFA-based controller leads to a smaller tracking error. Figure 4. The position tracking results of the held object. object - - reference Global Feed-forward Adaptive Fuzzy Control of Uncertain MIMO Nonlinear Systems 117 Figure 5. The velocity tracking results of the held object. object - - reference Figure 6. a-c Internal force tracking errors for Robot 1 and d-f internal force tracking errors for Robot 2 118 .