應用自適應模糊滑動模式控制策略與陀螺儀平衡器於自動導引自行車之平衡控制

Abstract

本論文旨在建立一自動導引自行車平台，應用陀螺儀平衡器 (Gyroscopic balancer) 與自適應模糊滑動模式控制策略 (Adaptive fuzzy sliding mode control, AFSMC) 於平衡控制，並比較使用AFSMC時與使用模糊滑動模式控制策略 (Fuzzy sliding mode control, FSMC) 時之差異。 首先應用FSMC控制策略與陀螺儀平衡器於自動導引自行車平台。陀螺儀平衡器具有下列幾項優點，如響應時間相對於其他平衡器較快、相對於其他平衡器能產生較大的力矩及具有較低的質量比(平衡器重量/自動導引自行車系統重量)。而FSMC控制策略相較於PID對非線性系統有較好的控制效果，跟SMC相比也能降低切跳現象與減少高頻響應，更能降低非線性系統在線性化時的影響。為了驗證上述所說的各項特質，本研究設計且進行了許多模擬及實驗。實驗結果說明本平台能克服崎嶇的地面，並有效的抵抗衝擊行的干擾，而實驗結果與模擬結果的一致性也驗證了先前所推導的動力學模型。 然而陀螺儀平衡器因為機構的限制，在遭受一連續性干擾時，飛輪勢必得往固定一方向旋轉以產生足夠的力矩去抵抗干擾，故當飛輪達到旋轉上下限時，平台即會失去自主平衡的能力。因此本論文旨在發展AFSMC控制器取代FSMC控制器，在遭受連續干擾的時候，控制器能調整本身的輸出命令，去避免飛輪牴觸到旋轉角上下限，進而有效的提升自行車平台抵抗連續性干擾的能力。從模擬的結果看來，AFSMC控制策略在任何情況下平衡能力與抵抗干擾能力皆不輸FSMC，更重要的是AFSMC控制策略抵抗連續性干擾的能力遠優於FSMC控制策略，說明了使用AFMSC控制策略能有效地的提升平台的穩定度。

A riderless bicycle has been developed with a gyroscopic balancer controller by a Fuzzy Sliding Mode Controller (FSMC) and an Adaptive Fuzzy Sliding Mode Controller (AFSMC). The FSMC controller has first been implemented because it has better performance at controlling nonlinear systems than the one with PID. The FSMC can also reduce the chattering phenomenon caused by SMC and the effect of linearizing a nonlinear system. Compared with other balancers, the gyroscopic balancer has a couple of advantages, such as faster system response, lower mass ratio of balancer to bicycle and relatively larger moment. To demonstrate the attributes stated above, we designed and conducted experiments, including the balancing of unmoving bicycle, unmoving bicycle with external impacts, as well as the bicycle moving forward and turning. The experimental results shows that the bicycle can overcome jolts, uneven terrain and external disturbances. Furthermore, since the results of experiments are consistent with the ones of the simulation, it validates the derived bicycle dynamics model with the gyroscopic balancer and proves its robustness. However, the system’s ability to resist continuous disturbance is not strong enough because of the limitation on the tilt angle of the gyroscopic balancer. Hence, we modified the control strategy by using AFSMC despite the fact that the combination of FSMC and gyroscopic balancer performed well than others. From the simulations in chapter IV, it shows that the AFSMC has better performance at resisting continuous disturbances than FSMC does. Furthermore, the abilities to balance the unmoving bicycle or moving bicycle in any case are no less than FSMC. Thus, the AFSMC is employed to replace the FSMC. The designs of adaptive law and estimation law of AFSMC are based on the Lyapunov function to ensure the stability of the system.