Rancang Bangun Sistem Kontrol Paralel PD-SMCNN Pada Mini-Quadcopter Otonom Robust

Mini-quadcopter merupakan salah satu jenis quadcopter yang memiliki panjang lengan antara 10-20 cm (dari pusat massa) dengan bobot berkisar ±63 gram sehingga sangat sensitif terhadap gangguan eksternal. Beberapa sistem kontrol telah dikembangkan untuk sistem dinamik quadcopter, mulai dari sistem kontrol linier (PID, LQR, dan LQG) hingga sistem kontrol nonlinier (backstepping, sliding mode control, kontrol cerdas, dan lain-lain). Akan tetapi, sistem kontrol linier kurang mampu mengatasi gangguan eksternal untuk sistem dinamik quadcopter. Pada tesis ini, telah berhasil dirancang sebuah sistem kontrol kombinasi antara sistem kontrol linier dan nonliner, yaitu proportional derivative-sliding mode control neural network (PD-SMCNN) yang disusun secara paralel. Sistem kontrol PD berfungsi memberikan gain thrust, dan SMCNN berfungsi sebagai kontrol robust di dalam sistem dinamik quadcopter secara bersamaan. Mini-quadcopter yang digunakan adalah jenis Parrot Mambo Minidrone dengan panjang lengan 18 cm dan bobot 73 gram. Sistem kontrol bawaan pabrik (PD) dari quadcopter ini dimodifikasi dengan sistem kontrol yang diusulkan (PD-SMCNN) menggunakan MATLAB-Simulink. Sistem kontrol yang diusulkan divalidasi melalui simulasi software dan eksperimen dengan masing-masing diberikan tiga macam input. Dari hasil simulasi software dan eksperimen yang dilakukan, didapatkan hasil sistem kontrol PD-SMCNN memiliki performa lebih unggul daripada sistem kontrol PD dalam hal robustness. Hal ini dibuktikan dengan nilai overshoot sebesar 26,7% untuk kontrol PD dan 0% untuk kontrol PD-SMCNN. Kemudian, melalui tahap eksperimen didapatkan tingkat robustness sebesar 46% untuk kontrol PD dan 100% untuk kontrol PD-SMCNN pada kondisi tanpa gangguan eksternal, serta pada kondisi dengan gangguan eksternal tingkat robustness kontrol PD sebesar 25% dan kontrol PD-SMCNN sebesar 55,7%. ================================================================================================ Mini-Quadcopter is a small size quadcopter with 10-20 cm of arms-length (from its center of mass) and ±63 gram total weight, which causes high sensitivity to external disturbances and uncertainties. Several control strategies have been developed to overcome those parameters, beginning with the linear control approach (PID, LQR, and LQG) to the nonlinear control approach (backstepping, sliding mode control, intelligent control, etc.). In this study, has been developed a combination of linear and nonlinear control systems named proportional derivative-sliding mode control neural network (PD-SMCNN) in parallel. The PD controller functions to obtain the thrust gain during take-off and SMCNN functions as a robust control. The objectives of this thesis are to generate a robust control subject to external disturbances or any uncertainties and allow the quadcopter able to fly autonomously. The mini-quadcopter used for this study is the Parrot Mambo Mini Drone which has a 17 cm arm length and 73 grams total weight. Furthermore, the default controller (PD) has been removed and modified with the proposed controller (PD-SMCNN) through MATLAB-Simulink. The software simulation and experiment have been conducted to validate the proposed controller with three input trajectories respectively. From the software simulations and experiment results, it is found that the PD-SMCNN controller has superior performance to the PD controller in terms of robustness. This is validated by the overshoot value of 26.7% for the PD controller and 0% for the PD-SMCNN controller. Then, through the experimental results, the robustness value is about 46% for the PD controller and 100% for the PD-SMCNN controller in the absence of the external disturbance, also in the presence of the external disturbance, the robustness value of the PD controller is 25% and the PD-SMCNN controller is 55,7%