Robust Optimization for A Nonlinear Switched Time-Delay System with Noisy Output Measurements Using Hybrid Optimization Algorithm

This paper considers the problem of using noisy output data to estimate unknown switching times and unknown system parameters arising in 1,3-PD batch production process of glycerol induced by Klebsiella pneumonia. We formulate the problem as a robust optimization problem in which the unknown quantities are decision variables to be chosen optimally, with the cost function penalizing the mean and variance of the relative error between the output of the system and the measured actual noisy system output. This problem is governed by a switched time-delay system subject to continuous state inequality constraints arising from engineering specifications. A new time-scaling transformation and constraint transcription technique are used then to convert this resulting problem into a sequence of approximate subproblems. A hybrid optimization algorithm combined genetic algorithm with gradient-based method is developed to solve these resulting subproblems. Finally, we explore the correctness of the optimal switching times and system parameters obtained, as well as the effectiveness of the proposed algorithm.