An evolutionary approach for path following optimal control of multibody systems

An approach to the determination of approximate solutions of path following optimal control problems by exploiting evolutionary optimization techniques is proposed. According to the methodology developed, controls are represented by discrete vectors and substituted in the system equations. Alternatively, when it is possible to exclude the controls from the system equations, the parameterization function of the path is represented by a discrete vector. In both cases the components of these vectors are regarded as variables of a performance index based goal function that is to be minimized with respect to the control and phase constraints. Such an approach enables modeling and solution of a wide class of path following optimal control problems, arising in engineering practice, within a unified framework of constrained optimization techniques, including: implementation of genetic algorithms for global optimization and multiobjective control, and utilization of parallel processing to alleviate the computational burden in high dimensional optimal control problems. An optimal path following algorithm for control of multibody systems is developed and a test example for minimum time control of a robot is presented. Computer realization of the proposed method is mainly based on MATLAB simulation programs for constrained optimization and genetic algorithms. The results obtained can be implemented to solve path following optimal control problems in the field of computer aided control engineering, computer integrated manufacturing, mechatronics and robotics