Time-Varying Optimal Terminal Control For Autonomous Soft Landing Of Helicopter

This dissertation concerns the design of linear control strategies for autonomous landing of helicopter with less shock force upon ground.A helicopter is an aerial vehicle that posses several flying qualities which makes it an interesting subject for development of an autonomous platform. Helicopter is a multiinput multioutput (MIMO) systems in terms of modern control theroy, or linear quadratic (LQ) control,which provide graceful,coordinated controls for MIMO systems;such controllers are, in general, more difficult to obtain by other synthesis methods such as pole placement or successive loop closure. The helicopter soft-landing control can be regarded as a finite-horizon LQ problem with terminal constraints, in which the initial state is manually assigned and the terminal state is weighted quadratically in the performance index, is traditionally solved by soft terminal control and hard terminal control.In the first part of dissertation,two efficient algorithms are developed for continuous and discrete finite-horizon LQ terminal control problem. It is shown by numerical example that, under the circumstance of the deterministic and stochastic control, the efficient algorithms have less computation time and less feedback gains than existing hard terminal control while achieve the same terminal accuracy at cost of less control expenditure. It also discovered that the transit-matrix method is crucial to reduce terminal error greatly. The problem in realizing continuous gaussian white noise in MATLAB is also disscussed deeply.Two robust control strategies, linear worst-case (LQW) control and parameters-robust control, are introduced and appreciated in the second part of the dissertation. The two control are results of combination of LQ and minimax method in differential game.Linear worst-case control determine the worst possible disturbances and corresponding best controls with intergral-quadratic penalties on bothof them, using minimax methods from differential game theory. Parameter-Robust LQ control method finds the worst combinations of parameter deviations and design the best controller that handles of them. Meanwhile, a robust measure, based on the expected deviations of plant parameters from their nominal values is presemted,like the gain margin or the phase margin of classical controllers.Finally, outlook and conclusions for the research work are presented.