基于SPSS回归分析的锂渣混凝土抗压强度预测模型-《建筑科学与工程学报》

文章信息/Info

Title:: Prediction Model of Compressive Strength of Lithium Slag Concrete Based on SPSS Regression Analysis

Author(s):: XU Kai-cheng, BI Li-ping, CHEN Meng-cheng; School of Civil Engineering and Architecture, East China Jiaotong University

Keywords:: SPSS; stepwise regression analysis; nonlinear regression; lithium slag concrete; compressive strength; residual analysis

摘要:: 利用SPSS软件的逐步回归分析法、多元非线性回归法建立锂渣混凝土的强度预测模型，分析各模型的残差图、预测值与试验值的对比，并结合均方根误差、平均绝对误差、平均绝对百分比误差和模型可决系数值对各模型的精确度等进行综合评价，最终确定出较优的锂渣混凝土强度预测模型。结果表明：水胶比、锂渣掺量和减水剂掺量对锂渣混凝土强度的影响十分显著；经残差分析和95%预测值区间检验，5个建议模型都有较好的精确度；经综合评价建议最佳的锂渣混凝土强度预测模型是以水泥强度、胶水比、锂渣掺量和减水剂掺量为自变量的非线性回归方程，其相应的可决系数R²=0.920，均方根误差为3.684，平均绝对误差为3.15，平均绝对百分比误差为5.44。

Abstract:: The strength prediction model of lithium slag concrete was established by the stepwise regression method and multiple nonlinear regression analysis method in SPSS software. Then the residual plots and comparison between predicted and experimental values of the proposed models were analyzed. Combined with root mean square error, mean absolute error, mean absolute percentage error and model determination coefficient value, the accuracy of the proposed model was evaluated synthetically, and the better strength prediction model of lithium slag concrete was determined. The results show that the effects of waterbinder ratio, the amount of lithium slag and water reducing agent on the strength of lithium slag concrete are very significant. By the residual analysis and 95% predictive value interval test, the 5 proposed models have good accuracy. Based on the comprehensive evaluation, it is recommended that the best strength prediction model of lithium slag concrete is a nonlinear regression equation with the strength of cement, the binderwater ratio, the amount of lithium slag and water reducing agent as the independent variable, and its corresponding performance measurement values include that determination coefficient ?R² is 0.920, root mean square error is 3.684, mean absolute error is ?3.15, mean absolute percentage error is 5.44.