Model Identification and Autonomous Flight Control Design of Helicopter

This thesis proposes a model identification and control design for a manned helicopter. We expect this process forming as a basis for appropriate implementing an unmanned helicopter in the near future. According to the limited outfit, we install sensors in a helicopter R22 record and analyze the flight data for model identification. Control design of a helicopter is quite different from the design of a fixed-wing airplane, because the helicopter exhibits more complicated dynamic behavior than that of a fixed-wing airplane. To realize the unmanned function, this research collected flight data from the sensors when a professional pilot is operating the helicopter for a specific flight course. We construct the helicopter plant model by NARX black box system identification technique. And we adopt model predictive control strategy to control the reference model. Finally, a model predictive control unmanned helicopter system is built. The model predictive control is a kind of optimal control methods for this nonlinear plant. When synthesizing the engine vibrations and wind noise disturbance are considered. The experiment result shows that the model predictive control strategy ensures a smooth flight for the unmanned helicopter.此論文提出將原先載人直升機進行無人化控制的方法，盼做為現今軍事工業上對於舊有預淘汰之軍事載具之最佳的妥善方法，且將直升機無人化過程在學術單位的研究成果是極為稀有且較為難以實現的，在此依有限配備條件之下提出一套藉由在直升機體上架設量測器且由數據模擬分析方法涵蓋了無人控制以及實用性能考量。 直升機的控制方法非常不同於一般定翼機的設計，直升機在飛行的過程中具備比定翼機更加複雜且多重耦合地動態行為，亦有更多樣的飛行姿態可加以利用。為實現無人化飛行控制功能，本論文藉著載人直升機(Robinson R22 beta ii)的實際飛行過程搭配裝載感測器測量其飛行間數據，由蒐集而來的數據透過後續軟體中(NARX)黑盒辨識方法初步得到多近多出(MIMO)之直升機控制模型，而後藉此模型進行控制律之研發。 最終經由直升機引擎震動及風力考量，以及複雜的耦合效應、在NARX多進多出數學模型的條件之下，選擇使用一套可方便控制且兼具操控性及強健性的模型預測控制(Model Predictive Control, MPC)在多個旋翼機可行之動作達到將一實際飛行直升機進入模擬系統中無人化之功效。誌謝 i 中文摘要 ii Abstract iii Contents iv List of Figures vii List of Tables xiii Introduction 1 Chapter.1 Helicopter Analysis 3 1.1 Helicopter Structure 3 1.1.1 Aerodynamic Effects on Rotor Blades 5 1.1.2 R22 Beta II Physical parameters 7 1.1.3 Analysis of Helicopter Dynamics 9 1.1.4 Helicopter Flight Control 10 Chapter.2 Nonlinear ARX System Identification & Flight Simulation 13 2.1.1 Definition of nonlinear ARX model 14 2.1.2 System identification by using FSX 17 2.1.3 Experimental results in FSX 19 Chapter.3 Data Collection and System Identification 22 3.1.1 Sensors installed before flight testing 24 3.2 Flight experiments to collect data for system identification 31 3.2.1 The flight courses 32 3.2.2 Measuring Signal Similarities 34 Chapter.4 Model Predictive Control 41 4.1 MPC Mathematical Formulation 43 4.1.1 Flight Experiment 1: Tracking the velocity signal 49 4.2 Velocity control for unmanned helicopters based on MPC 55 4.2.1 Flight Experiment 2: MPC for Hovering Control 58 4.2.2 Flight Experiment 3: Helicopter flights in different direction 60 4.3 Robust MPC for Helicopter Hovering in Wind Noise 65 4.3.1 Tuning the Weight for Robustness in MPC 65 4.3.2 Robust Flight Experiment: R22 Helicopter Hover test in strong wind disturbance 67 Chapter.5 Conclusions 71 References 72 Appendix 7