基于遗传算法和人工神经网络的高性能混凝土劈裂抗拉强度预测

Splitting Tensile Strength Prediction of High Performance Concrete Based on Genetic Algorithm and Artificial Neural Network

为实现高性能混凝土劈裂抗拉强度的准确预测,提出结合人工神经网络(ANN)和遗传算法(GA)的GA-ANN预测模型。首先通过遗传算法得到合适的人工神经网络初始权值和阈值,然后在GA-ANN预测模型基础上对数据集进行训练和测试,并利用相关系数误差指标对模型性能进行比较和评估。结果表明,经过遗传算法优化的模型预测结果与实际抗拉强度相关系数均>0.97,误差分布更集中于0附近,预测精度较优化前得到较大程度提升。所提出的GA-ANN预测模型为高性能混凝土劈裂抗拉强度提供了一种高精度、方便快捷的预测方法。

To achieve accurate prediction of splitting tensile strength of high performance concrete,a GA-ANN prediction model combining artificial neural network(ANN)and genetic algorithm(GA)was proposed in this paper.The appropriate initial weights and thresholds of the artificial neural network were first obtained by the genetic algorithm,and then the data set was trained and tested based on the GA-ANN prediction model.Then,the model performance was compared and evaluated by using the correlation coefficient error index.The results show that the correlation coefficient between the prediction results and the actual tensile strength of the model optimized by the genetic algorithm exceeds 0.97,and the error distribution is more concentrated around 0.The prediction accuracy is improved to a greater extent than before the optimization.The GA-ANN prediction model proposed provides a highly accurate,convenient,and fast method for splitting tensile strength prediction of high-performance concrete.