Many-objective Evolutionary Algorithm And Its Application On LLCL Filter For PV Inverter

Most distributed generation systems are connected to the grid via grid-tied inverters.In order to reduce the injection of high-frequency harmonic into the grid,a low-pass filter like LLCL filter is usually inserted between the inverter and the grid.In order to solve the problems of high computational complexity in the design for the parameters of LLCL filter,this thesis combines the LLCL filter with the many-objective evolutionary algorithm,and studies the relevant theories of the many-objective evolutionary algorithm.Firstly,The thesis introduces the theory of many-objective evolutionary algorithm.Then,an improved RVEA algorithm,named RVEA-H,is proposed based on RVEA algorithm and DE(Differential Evolution).The algorithm is compared with NSGAIII,MOEA/D,and the original RVEA algorithm and the result shows the superiority of the RVEA-H algorithm.Next,we use the RVEA-H algorithm to design the parameter of LLCL filter and propose a novel parameter optimal design method for the LLCL filter.The LLCL filter inductance constraints,capacitive constraints,resonant frequenc y constraints and filter performance are taken into account to get the objective functi on and the ranges of parametersFinally,The photovoltaic grid-tied inverter model is built based on the NI platfor m to verify the effectiveness of the proposed optimization method.The experimental result shows that the proposed method can be used to design low inductance and high performance filters without depending on the experience of experts.This method further extends the application field of many-objective evolutionary algorithm.