基于改进粒子群组合算法的电网基建年度投资预测

摘要

Optimization of power grid infrastructure invest-ment decision-making depends on reasonable prediction of grid infrastructure investment. Multiple linear regression,BP neural network and grey systems theory are often used in prediction since infrastructure investment is influenced by multiple and complex factors. By comparison,these methods have both pros and cons,which could cause errors in solving practical pro-blems. Firstly,this paper analyzes different influence factors for infrastructure investment by SPSS and predicts the investment separately in multiple linear regression,BP neural network and grey system method. Secondly,by using a combination algori-thm based on improved PSO the results obtained by these three methods are combined and optimized to get the final prediction value. Finally,the analysis of the calculation example shows the method proposes in this paper can improve the accuracy of the prediction efficiently.%利用SPSS软件对基建投资的主要影响因素进行了分析，并分别利用多元线性回归法、BP神经网络法、灰色系统理论法对基建投资进行了预测。为了获得更加准确的预测结果，采用了一种基于改进粒子群的组合算法，将3种方法的预测结果进行优化组合，得到最终的预测值。经过算例分析表明，该方法能够有效提高预测精度，得到较好的预测结果。