基于遗传支持向量机的多维灰色变形预测模型研究

Study on multi-dimensional grey prediction model of deformation based on genetic algorithm and support vector machines

多维灰色模型适合对多因素影响下的贫信息系统问题进行建模,但对多因素影响下的非线性变形系统建模和预测精不高,针对该问题进行分析研究.利用支持向量机算法建立多维灰色变形预测模型的残差与变形影响因素之间的非线性关系,对多维灰色变形预测模型的残差进行预测,并与多维灰色变形预测模型相加,对多维灰色变形预测模型进行修正,构建基于支持向量机的多维灰色变形预测模型.利用遗传算法优化支持向量机模型参数,提高支持向量机建模精度.该方法较好地解决了多维灰色变形预测模型精度不高的问题.把该模型应用于大坝变形预测,并与多种传统变形预测方法进行对比,结果证实该方法有效提高多维灰色变形预测模型的精度,且新模型精度远优于传统方法,是一种新的有效的变形预测模型.

Multi-dimensional grey model （GM（1, N）） can be used for poor data information system modeling with different influencing factors effectively. But when GM （1, N） is used for complicated deformation forecasting, the forecasting accuracy is poor. This paper studies the problem. The paper obtained a relation between residual error of GM（1, N） and influencing factors based on support vector machines （SVM）. Residual error was forecasted by the method. Predictions were added to the result of GM（1 ,N）of deformation. This method can modify GM（1, N） of deformation. Finally, GM（1, N） of deformation based on SVM was constructed. The parameter of SVM could be optimized by GA to improve the precision of SVM efficiently. The new method can effectively solve the poor accuracy problem of GM（1, N）of deformation. The new model was used for prediction of dam deformation and is compared with traditional deformation forecasting models. Case study showed that new method can effectively improve the accuracy of GM（1 ,N）of deformation. The accuracy of new model is much higher than traditional methods. GM（1,N） of deformation based on genetic algorithm and SVM is new effective deformation prediction model.