基于LSTM模型的再生混凝土抗压强度预测分析

再生混凝土抗压强度预测已经有很多方法,但目前的研究大多集中在对混凝土28 d抗压强度预测上,对于超长龄期的预测较少。为更好地对再生混凝土超长龄期强度进行预测,该研究使用数理统计方法对56 d抗压强度进行统计学分析,并使用箱线图对所有变量进行异常值检测,使用LSTM模型对再生混凝土56 d抗压强度进行预测,为更加准确地对模型预测效果进行评价,采用决定系数R2、均方根误差(RMSE)、平均绝对误差(MAE)作为评价指标。最终模型结果表明,LSTM模型可以较为精确地实现再生混凝土超长龄期抗压强度的预测,为工程实践提供指导。

There have been many methods to predict the compressive strength of recycled concrete,but the current research is mostly focused on the prediction of 28-day compressive strength of concrete,but there are few predictions for ultra-long age.In order to better predict the super-long age strength of recycled concrete,this study uses mathematical statistics to analyze the 56-day compressive strength,and uses the box chart to detect the abnormal values of all variables.The LSTM model is used to predict the 56-day compressive strength of recycled concrete.In order to evaluate the prediction effect of the model more accurately,the determination coefficient R2,root mean square error(RMSE)and mean absolute error(MAE)are used as the evaluation index.The final model results show that the LSTM model can accurately predict the compressive strength of recycled concrete at ultra-long age and provide guidance for engineering practice.