刘海峰,刘浩天,李罗胤,等.基于改进BP模型的沙漠砂混凝土高温后抗压强度预测[J].河南理工大学学报(自然科学版)，doi:10.16186/j.cnki.1673-9787.2024030006.

LIU H F,LIU H T,LI L Y,et al．Prediction of compressive strength of desert sand concrete (DSC) after high temperature based on improved BP model [J].Journal of Henan Polytechnic University( Natural Science) ，doi:10.16186/j.cnki.1673-9787.2024030006.

基于改进BP模型的沙漠砂混凝土高温后抗压强度预测(网络首发)

刘海峰，刘浩天，李罗胤，陈小龙，车佳玲，杨维武

（宁夏大学 土木与水利工程学院,宁夏 银川 750021）

摘要: 目的 为探究高温历程对沙漠砂混凝土（DSC）抗压强度的影响，方法 考虑4个因素（沙漠砂替代率、温度、升温速率和静置时间），进行高温后沙漠砂混凝土抗压强度试验，并借助 X 射线衍射（XRD）和扫描电子显微镜（SEM）分析经历高温后沙漠砂混凝土微观形貌和物相组成变化规律。以反向传播算法（Back-Propagation, BP）为基准，融合粒子群算法（Particle Swarm Optimization，PSO）和遗传算法（Genetic Algorithm，GA）训练人工神经网络（Artificial Neural Network，ANN），建立高温后沙漠砂混凝土抗压强度预测模型，并采用十折交叉验证的方法对该模型进行验证。结果 结果表明：随着温度升高，DSC抗压强度呈下降趋势，材料内部水化产物大量分解，微观裂缝逐渐扩展并连接贯通；静置时间越长，抗压强度越高；升温速率越快，DSC破坏速率随之增大；沙漠砂替代率为20%时，DSC抗压强度达到最大值。三种预测模型预测值与实测值的平均绝对百分比误差均控制在8%以内。模型优化程度越高，误差范围越小。采用粒子群优化遗传混合算法（PSO-GA）神经网络模型预测结果更为精准，该模型预测值均方差（RMSE）为1.1272，平均绝对百分比误差（MAPE）为3.98%，28d抗压强度预测决定系数(R²)为0.9878，结论 显著提高了沙漠砂混凝土高温后力学性能预测的准确性。

关键词:沙漠砂混凝土；抗压强度；高温；神经网络模型

doi:10.16186/j.cnki.1673-9787.2024030006

基金项目:国家自然科学基金资助项目（52168034）；宁夏自然科学基金资助项目(2023AAC03039),2023年自治区级大学生创新项目，宁夏高等学校一流学科建设（水利工程学科）资助项目（NXYLXK2021A03）；2024年宁夏大学研究生创新项目（CXXM202454）

收稿日期：2024-03-03

修回日期：2024-05-21

网络首发日期：2024-09-09

Prediction of compressive strength of desert sand concrete (DSC) after high temperature based on improved BP model

LIU Haifeng,LIU Haotian, LI Luoying, CHEN Xiaolong, CHE Jialing, YANG Weiwu

(School of Civil and Hydraulic Engineering,Ningxia University,Yinchuan, 750021,Ningxia,China)

Abstract: Objectives In order to investigate the effect of high temperature history on the compressive strength of desert sand concrete (DSC),taking into account of factors such as desert sand replacement rate (DSRR),temperature,heating rate and settling time, Methods the compressive strength test of DSC after high temperature was carried out.At the same time,X-ray diffraction (XRD) and scanning electron microscope (SEM) was used to analyze the change rule of microscopic morphology and phase composition of DSC after experiencing high temperature environment.On the basis of Back-Propagation (BP) algorithm,the particle swarm optimization (PSO) and genetic algorithm (GA) were used to train the artificial neural network (ANN) and the compressive strength prediction model of DSC after high temperature was established,which was validated by ten-fold cross-validation. Results Research results indicated that with increasing temperature,the compressive strength showed a decreasing trend, the hydration products in the specimen was decomposed greatly and the micro-cracks gradually expanded and connected with each other.The longer the settling time, the higher the compressive strength was.The faster the heating rate,the greater the failure rate of DSC was.DSC compressive strength reached the maximum value with the DSRR of 20%.The average absolute percentage error between the predicted results from the three prediction models and measured values was controlled within 8%.The higher the degree of optimisation of the model,the smaller the error range was.The predicated results from the particle swarm optimization-genetic hybrid algorithm (PSO-GA) neural network model was more accurate,with a mean squared error (RMSE) of 1.1272,a mean absolute percentage error (MAPE) of 3.98% and coefficient of determination (R²) for 28d compressive strength predication of 0.9878, Conclusions which significantly improved the accuracy of predictions for the mechanical properties of DSC after high temperature.

Key words: desert sand concrete (DSC); compressive strength; high temperature; neural network model