基于灰色关联分析的极限学习机在混凝土抗压强度预测分析中的应用

Application of extreme learning machine based on gray relational analysis in prediction of concrete compressive strength

利用灰色关联分析法筛选出表征混凝土抗压强度的重要因素指标,并以选取的因素指标为输入变量、以抗压强度为输出变量,创建极限学习机(ELM)模型,克服了冗余因素对模型精度的影响。实例分析表明经指标优化选择的ELM模型具有较高的精度,对抗压强度的预测效果明显优于未经指标筛选的ELM模型,也远好于支持向量机的预测效果,为混凝土抗压强度预测提供了一种新思路。

The gray correlation analysis method is used to select the important factors to characterize the compressive strength of concrete. Extreme learning machine（ELM） model is established by using the selected factor as the input variable and the compressive strength as the output variable, which overcomes the influence of redundant factors on the accuracy of the model.An example analysis shows that the ELM model based on index optimization has higher accuracy,The prediction effect of compressive strength is superior to ELM model without index selection, and it is much better than predictive effect of support vector machine（SVM）, which provides a new idea for concrete compressive strength prediction.