基于布谷鸟算法的电子变压器输出端谐振参数优化

OPTIMIZATION FOR RESONANT PARAMETERS OF OUTPUT PORT OF POWER ELECTRONIC TRANSFORMER BASED ON CUCKOO SEARCH ALGORITHM

常用来降低电子变压器主开关损耗的方法。目前是基于LLC谐振网络的软开关技术。其中谐振参数的选取是否合适直接影响到电子变压器输出性能的好坏。为此,该文引入布谷鸟搜索算法对谐振参数进行智能优化设计。该搜索算法能够对存在不可导点的目标函数进行优化,并且其特有的巢寄生行为和莱维飞行特性能够使得谐振参数避免陷入局部最优。其优化过程首先利用基波分析法对变换器的参数进行建模,然后分析和研究谐振电感、励磁电感与谐振网络直流增益、输入阻抗角以及原、副边电流、励磁电流的直接关系,从而得到参数优化的约束条件。在此基础上,利用布谷鸟搜索算法优化得到使电路损耗最小的谐振网络参数。最后,以半桥LLC谐振变换电路为例,通过实验验证了在宽输入电压下,全负载范围内,该优化设计方法比"FHA+仿真"法的效率提高了1%~2%,样机最高效率达97%,证明了论文所提优化方法的有效性。

The soft-switching technique based on the LLC resonant topology is widely used in the medium-high frequency DCDC converter as the output port of the power electronic transformer(PET)to reduce switching losses. The most important part is the design of the resonant parameters which directly affects the performance of the PET output. In this paper,Cuckoo Search(CS)algorithm is introduced to realize the resonant parameter optimization. CS algorithm can optimize the objective function with non derivable points,and its unique nest parasitic behavior and Lévy-flight characteristics can avoid the resonance parameters falling into local optimum. During the intimation,the fundamental harmonic approximation(FHA)is applied to analyze the resonant process and the mathematical model of this converter is established. The direct relationships between resonant parameters and the characteristics of the resonant network including the DC gain,input impedance angle and keypoint currents are analyzed in detail so as to obtain the constrain functions. Then the Cuckoo Search algorithm is used to select a set of resonant parameters that minimize switching losses. Finally,taking an LLC half-bridge resonant converter as an example,the experiment results show that under the wide-input-voltage and full-load-range condition,the efficiency of the optimized design in this paper is 1%-2% higher than that of the"FHA + simulation"method. The maximal efficiency of the prototype is97%,which proves the effectiveness of the optimization method proposed in this paper.