Optimal Design Of LCL Filter Parameters Based On Evolutionary Algorithm

With the development of the technology of clean and renewable energy,more and more scholars are contributed to the research of grid-connected technology of clean energy such as wind power generation.Among them,how to optimize the filter to effectively filter the harmonics in the grid-connected process is essential.In this thesis,we select the LCL type filter as the research object.And we design the filter parameters by optimizing the harmonic attenuation,ripple current,and power loss to improve the overall performance of the filter.This thesis introduces the design and optimization methods of mainstream filters firstly.In the condition of selecting the LCL topology structure as the analysis object,we discuss the access method of the damping resistor and the system constraints,and introduce an engineering experience method for tuning the parameters of the filter.For the problem that the filter cannot arrive the optimum performance by using the traditional method to tune the parameters of the filter,the evolutionary algorithm is introduced to optimize the parameters of the filter.The single-objective evolutionary algorithm such as the genetic algorithm and the multi-objective optimization genetic algorithm such as the non-dominated sorting genetic algorithm are introduced.For the problem that the original non-dominated sorting genetic algorithm has large computational complexity and more redundant operations,we propose a novel non-dominated sorting method and apply it to multi-objective optimization algorithms,and we improve the related operations of the original algorithm.Through the above improvements,the computational complexity of the algorithm is reduced and the operating efficiency of the algorithm is improved.We analyze and prove that when dealing with multi-objective optimization problems,the solution obtained by the weighted singleobjective optimization algorithm belongs to the Pareto solution set obtained by the multiobjective optimization algorithm.A mathematical model for parameter optimization of LCL-type filters with maximum harmonic attenuation,minimum ripple current,and minimum power loss is established.For the limitation of the traditional method in dealing with the constraints of this thesis,a constraint processing method for transforming constraints into decision variables is proposed.This method transforms the model from constrained multi-objective optimization problems into unconstrained multi-objective optimization problems,which greatly simplifies the difficulty of evolutionary algorithms in solving multi-objective optimization problems.In order to verify the performance of the LCL filter designed by each parameter tuning method,the experiments were designed based on the NI experimental platform.The experimental results show that the filter with the evolutionary algorithm design parameters has larger harmonics attenuation,smaller ripple current,and smaller power loss than the filter with the engineering empirical design parameters;The multi-objective optimization algorithm based on the novel non-dominated sorting method proposed in this thesis is more suitable than the single-objective optimization algorithm in solving this model.