摘要： 提出具有9个输入节点,1个输出节点的RBF神经网络模型来模拟抗压强度及其影响因素之间复杂非线性关系。作为对比,作者同时比较了3种不同输入模型的RBF网络的预测效果并与传统的BP网络模型进行比较,结果表明,文章提出的RBF网络模型具有很高的预测精度和较强的泛化能力,可作为商品混凝土性能分析的一种新型有效的方法。

关键词: RBF神经网络, 抗压强度, 性能预测

Abstract: On the basis of comprehensive analysis,a prediction model of RBF network with 9 input vectors and 1 output vectors is set up to express the complex nonlinear relationship of performance index and their influencing factors.In contrast,the prediction results from three different input models of RBF network are compared with each other,and also compared with that of BP network.As a result,the RBF network model mentioned in this paper has high precision and good generalization capacity and can become a new effective prediction method.

中图分类号: 

• TU528.52