粒子群优化的支持向量回归机计算配电网理论线损方法

针对配电网理论线损精确计算,提出一种基于粒子群优化算法的支持向量回归机(SVR-PSO)的理论线损计算方法。SVR-PSO方法将理论线损计算抽象成多元回归分析,理论线损的若干影响因素作为自变量,理论线损值作为因变量,SVR-PSO通过对已知理论线损线路的数据样本训练学习生成配电网理论线损计算模型,进而利用该模型完成未知线路的理论线损计算。在SVR-PSO训练过程中,利用粒子群算法动态地搜索支持向量回归机的最优训练参数,提高了SVR-PSO的计算精度。最后横向对比实验证实了基于SVR-PSO的配电网理论线损计算方法的有效性,与传统方法相比,SVR-PSO方法在计算精度和运算耗时方面拥有更好的性能。

Theoretical line loss calculation based on SVR and PSO for distribution system

To precisely calculate the theoretical line loss of power distribution system,a method based on SVR(Support Vector Regression) and PSO(Particle Swarm Optimization) is proposed,which converts the calculation of theoretical line loss into MRA(Multi Regression Analysis).In MRA,all influencing factors are taken as independent variables and the line loss as dependent variable.The calculation model of theoretical line loss is generated by SVR-PSO through training with the samples of known lines and then used to calculate those of unknown lines.During SVR training,PSO is applied to dynamically search the optimal training parameters to improve calculation precision.Experiment verifies the effectiveness of the proposed calculation method,which is better in calculation accuracy and speed than traditional methods.