BP网络预测再生混凝土性能研究

利用BP神经网络模型,对再生混凝土强度及工作性能的预测方法进行了探讨。根据再生混凝土的特殊性,找出影响其强度和坍落度、保水性的主要因素,对试验中通过主观观察得到的数据进行量化,在此基础上建立预测其强度和工作性能的BP神经网络模型,针对所建模型,输入一定量的实测数据样本,对网络进行训练。为了验证训练好的网络的推广性能,用预留的一组试验数据进行仿真训练的效率和误差及仿真计算的结果表明,采用优化的BP网络模型及合适的样本参数训练出的预测系统对再生混凝土的强度及工作性能进行预测是可行的。

PREDICTION ON PERFORMANCE OF RECYCLED CONCRETE BY BP NEURAL NETWORK

Prediction on strength and workability of recycled concrete by means of BP neural network is introduced in this paper. The key factors affecting the strength, slump and the water retentiveness are found out basing on the specialty of recycled concrete. The data are quantified, and the BP neural network model was created on the basis of strength and workability. Some experimental data are input to testify the network model in order to prove the feasibility of the network model. Some other data are used to simulate the application of the model. The results indicate that to predict the strength and the workability of the recycled concrete by using the optimized neural network is suitable.