基于改进免疫算法优化支持向量机的钢材消费预测

Forecasting of Steel Demands by Using Support Vector Machine and Immune Algorithm

预测钢材需求量对我国钢铁工业及相关产业发展有重要意义。为了提高钢材需求量预测准确度，提出基于改进免疫算法优化支持向量机（IA—SVM）的钢材需求预测方法。IA—SVM采用免疫算法优化SVM参数，获得较优的SVM预测模型。为了提高IA收敛速度和寻优效果，提出基于有限随机思想的群体更新策略。针对我国1990—2009年的钢材需求数据进行实证分析，实验结果表明，改进的免疫算法能够找到支持向量机的最优参数组合，采用IA—SVM算法可以对钢材需求量进行有效预测。

It is very important for the Iron and Steel industry to forecast the steel demands. In order to im- prove the forecasting accuracy, a method by using support vector machine combined with immune algorithm （IA-SVM） is proposed. By this method, to obtain an effective SVM model, the parameters in the SVM are optimized by using immune algorithm. In the parameter optimization, a new policy called group updating is introduced to improve its convergence speed and performance. With a data set of steel demands in China from 1990 to 2009, empirical analysis is carried out. Results show that the proposed IA-SVM is effective.