低压开关槽式变换器多目标优化

Multi-objective optimization of switched-tank converter for low-voltage applications

为满足低压应用场合下DC-DC变换器高效率、小尺寸和低成本等多方面需求,提出一种同时优化DC-DC变换器的功率损耗、物理尺寸和成本的方法。以低压开关槽式变换器(switched tank converter,STC)为例,在分析其工作原理基础上,首先建立元器件的功率损耗、面积和成本模型,以设计指标为约束条件,再建立一个以功率损耗、面积和成本为优化目标的变换器多目标优化模型,且优化模型中的参数可从器件数据手册查到。采用基于非支配排序引力搜索算法(non-dominated sorting gravitational search algorithm,NSGSA)改进得到的大范围改进的非支配排序引力搜索算法(large-scale improved NSGSA,LSINSGSA)求解变换器多目标优化模型。将所得优化结果与NSGSA和NSGA-Ⅱ(non-dominated sorting genetic algorithmⅡ)算法的优化结果进行比较。结果表明,提出的STC变换器多目标优化方法可以得到综合性能最优的器件组合方案,实现STC的效率、面积和成本达到折衷最优的目的,改进的LSINSGSA算法的收敛性与Pareto前沿中最优解均匀分布性均优于NSGSA和NSGA-Ⅱ。

In order to meet the requirements of high efficiency,small size and low cost of DC-DC converters in low-voltage application scenarios,taking the low-voltage switched tank converter(STC)as an example,a method for comprehensively optimizing the power loss,physical size and cost of DC-DC converters was extablished in this paper.First,the power loss,area and cost models of the components were separately established based on STC’s operation principle,and a multi-objective optimization model of STC with power loss,area,and cost as the optimization goals was derived by using the design requirements as constraints while the parameters in these models could be obtained from device data sheet.Then,the large-scale improved non-dominated sorting gravitational search algorithm(LSINSGSA),an improved algorithm based on the non-dominated sorting gravitational search algorithm(NSGSA),was established in this paper to solve the derived multi-objective optimization model.Finally,the optimization results were compared with those of NSGSA and NSGA-Ⅱ.The results indicate that the multi-objective optimization method of the STC converter proposed in this paper can facilitate a device combination scheme with the best comprehensive performance,realizing the best compromise between the efficiency,area and cost of the STC,and the convergence and uniform distribution of the optimal solutions in the optimal Pareto frontier of the LSINGSSA are better than that of NSGSA and NSGA-Ⅱ.