基于改进FOA-SVR的电网工程概算预测研究

Research on Power Grid Project Budget Prediction Based on Improved FOA-SVR Method

为减少电网工程概算对于工程数据的依赖性并提高预测精度,构建了用于电网工程概算预测的改进FOA-SVR模型。通过灰色关联法进行工程样本数据筛选,将筛选后的数据导入SVR训练,并在标准FOA和自适应FOA的基础上,设置局部最优解跳出机制和飞行步长收敛模式,提出一种全局寻优能力和收敛效率均较高的改进FOA算法,寻找合适参数优化SVR模型。以某电力设计院2020年的30个电网工程架空线路为样本进行模型训练与概算预测,多次测试发现,模型能够稳定输出高精度的预测结果,且改进FOA算法参数寻优的效率较高,为电网工程概算的自动化计算和FOA算法的参数寻优应用提供了一定参考。

To reduce the dependence of power grid project budget on engineering data and improve the prediction accuracy,an improved FOA-SVR model for power grid project budget prediction is proposed.The engineering sample data are processed by GRA method,then the data are used for SVR training.On the basis of standard FOA and adaptive FOA algorithm,an improved FOA algorithm with local optimal solution jump out mechanism and flight step convergence method,which has high global optimization ability and convergence efficiency,are proposed for parameter optimization in the SVR model.Taking 30 overhead line samples of power grid projects from a power design institute in 2020 as an example,the data are brought into the model for training and predicting budget estimation.Plentiful tests prove that the model can stably output high-precision prediction results,and the efficiency of parameter optimization by improved FOA algorithm is high,which provides a reference for automatic calculation of power grid project budget estimation and parameter optimization by FOA algorithm.