基于改进遗传算法的SVR短期电力负载预测

Short- term Electrical Load Forecasting Based on Modified Genetic Algorithm and SVR

为了有效且精确地预测电力负载,提出一种基于支持向量回归(Support Vector Regression,SVR)的预测方法对负载消耗进行建模,同时提出一种基于遗传算法(Genetic algorithm,GA)的两级改进遗传算法(Modified Genetic Algorithm,MGA)以调整SVR中的参数。在满足SVR约束条件的情况下选用平均绝对百分比误差(Mean Absolute Percentage Error,MAPE)作为MGA的适应度函数。最后使用一组实际数据对基于MGA的SVR预测方法的可行性和有效性进行了验证。

In order to effectively and accurately predict the power load,we proposed a Support Vector Regression（ SVR） based forecast method to modeling the load consumption,at the same time proposed a two stage modified Genetic Algorithm（ MGA）based on Genetic Algorithm to adjust the parameters of SVR. Under the SVR constraint conditions,we chose the Mean Absolute Percentage Error（ MAPE） as fitness function of MGA. Finally using a set of practical data,we verified the feasibility and effectiveness of MGA based SVR forecasting method.