基于改进BP神经网络的路基材料性能预测

Forecast of properties of roadbed material based on improvable BP nerver network

结合人工神经网络原理,分析了路基材料性能的影响因素,采用GSL 变换改进后的BP神经网络来预测路基材料的有关性能,通过直接对实验数据的神经网络学习,发现各力学参数之间的关系,从而建造了路基材料性能的人工神经网络模型,对路基材料的无侧限抗压强度和回弹模量进行了预测,结果表明,该模型收敛速度及预测精度均得到改善,可以很好地拟合输入参数与输出参数之间的非线性关系,有较好的实际应用价值。

Based on artificial nerve network principle, influence factors of properties about roadbed material are analyzed. In the paper, it builds artificial nerve network model of forecasting the roadbed material properties with GSL improved BP never network. The result shows that the model`s mapping precision and convergence rate are high, it can mock unlinearity relationship of parameters about input and output.The proposed method is feasible and practical.