基于GA-BP算法的高炉矿渣–粉煤灰混凝土抗压强度预测

Compressive Strength Prediction of Blast Furnace Slag-Fly Ash Concrete Based on GA-BP Algorithm

为更准确地预测高炉矿渣–粉煤灰混凝土抗压强度,在MATLAB平台上通过遗传算法对BP神经网络的初始权值和阈值进行改进,建立了抗压强度预测的GA-BP模型。将人工神经网络(BP)、随机森林(RF)、支持向量机(SVM)、极限学习机(ELM)和多元非线性回归(MnLR)的预测结果进行对比分析,GA-BP模型的预测精度和模型稳定性等方面优势明显。从而为高炉矿渣–粉煤灰混凝土质量评估提供指导,具有重要的实用价值。

In order to predict the compressive strength of blast furnace slag-fly ash concrete more accurately, a GA-BP model for compressive strength prediction was developed by improving the initial weights and thresholds of BP neural network through genetic algorithm on MATLAB platform. The prediction results of artificial neural network (BP), random forest (RF), support vector machine (SVM), extreme learning machine (ELM) and multiple nonlinear regression (MnLR) were compared and analyzed, and the GA-BP model has obvious advantages in terms of prediction accuracy and model stability. Thus, it provides guidance for the quality assessment of blast furnace slag-fly ash concrete, which has important practical value.