In this paper, a novel and generic multi-objective design paradigm isproposed which utilizes quantum-behaved PSO(QPSO) for deciding the optimalconfiguration of the LQR controller for a given problem considering a set ofcompeting objectives. There are three main contributions introduced in thispaper as follows. (1) The standard QPSO algorithm is reinforced with aninformed initialization scheme based on the simulated annealing algorithm andGaussian neighborhood selection mechanism. (2) It is also augmented with alocal search strategy which integrates the advantages of memetic algorithm intoconventional QPSO. (3) An aggregated dynamic weighting criterion is introducedthat dynamically combines the soft and hard constraints with control objectivesto provide the designer with a set of Pareto optimal solutions and lets her todecide the target solution based on practical preferences. The proposed methodis compared against a gradient-based method, seven meta-heuristics, and thetrial-and-error method on two control benchmarks using sensitivity analysis andfull factorial parameter selection and the results are validated usingone-tailed T-test. The experimental results suggest that the proposed methodoutperforms opponent methods in terms of controller effort, measures associatedwith transient response and criteria related to steady-state.
展开▼
