A hybrid DE-SQP algorithm with switching procedure for dynamic optimization

A hybrid optimization algorithm DE-SQP is presented here to solve dynamic optimization problems. In this hybrid algorithm, differential evolution (DE) is firstly introduced to find the vicinity of the solution, and then sequential quadratic programming (SQP) is used to find the final solution. In particular, a new switching procedure is proposed to determine an efficient switching point at which the optimization is switched from the DE optimizer to the SQP optimizer. The proposed algorithm combines the two optimizers efficiently and avoids time consuming tests in advance. Performance of the DE-SQP algorithm is examined by solving two challenging multimodal optimal control problems. Simulation results show that the hybrid algorithm is insensitive to the initial point, and can find the final solution faster than the DE.