Differential Evolution Algorithm And Its Application Study In Parameter Optimization Of Control System

This paper is divided into three parts. The first part is the improvement of the basic differential evolution algorithm and corresponding simulation. The second part combines the improved differential evolution algorithm with multi-objective optimization algorithm and improves some of the algorithm strategy. The superiority of the algorithm is validated through simulation. The third part applies the improved multi-objective differential evolution algorithm to the level controller parameter tuning of steam generator in nuclear power plant and the optimization results are studied and validated on simulation model.The first part of the article improves the way that the control parameters of the basic differential evolution algorithm form so that it can adapt with evolution. Object parameters are identified from the simulation model data and actual plant operating data respectively using the improved differential evolution algorithm to prove the effectiveness of the new algorithm.The second part of the article combines the above improved differential evolutionary algorithm with multi-objective optimization. To avoide the shortcomings of the universally applied crowding distance algorithm, an improved method for calculation of crowding distance is proposed whihe for the problem of the basic differential evolutionary algorithm, an self-adjusted strategy is introduced both of which largely improve the speed of the multi-objective optimization algorithm.The third part of the article applies the improved multi-objective differential evolution algorithm to parameter setting of steam generator level control system in nuclear power plant. Firstly water level transfer function models under different conditions are established. Simulation is conducted to prove its dynamic features and usability. Secondly, the feed-forward-cascade water level control system of steam generator is established. Then, the water level controller parameters are encoded and are optimized and tuned using the improved multi-objective differential evolution algorithm and NSGA-Ⅱ respectively in five typical operating conditions. Compare the controlling effects of the controllers tuned using the above methods when the water level, water flow and steam flow are imposed step disturbances to verify the superiority of the improved multi-objective differential evolution algorithm.