Integrated Structural/Control Design via Multiobjective Optimization.

摘要

A minimum correction homotopy approach is used to obtain the simultaneous/integrated optimal design of a large flexible structure and its active control system. Instead of the usual method of weighting and summing all desired objectives to form a constrained scalar optimization problem, a vector of objective functions is dealt with directly. The Draper/RPL configuration (a central hub with four symmetric, identical arms) is the design structure. The design seeks to minimize the mass of the arms. Using simple feedback of arm displacements and velocities, the control system seeks to achieve specified closed-loop eigenvalues (frequencies and damping ratios) and control effort. Design variables are the arm dimensions, control system gains, and sensor and actuator locations. Not only can the structural design be accomplished while placing the closed-loop eigenvalues, but a simultaneous 50% reduction in mass and/or control effort can be obtained. Since reduced-order models were used for the structural/control design, the resultant configurations are easily driven unstable by spillover from higher-order unmodeled modes. A modal suppression technique is applied to eliminate observation spillover and provide a decade of deadband above the controller bandwidth. Keywords: Control theory, Combinational analysis. (kr)