基于Matlab的BP神经网络在泥石流危险性评价中的应用

Application of BP neural network to the hazard assessment of debris flow based on Matlab

由于泥石流孕育环境、成灾条件及其诱发因素等的随机性、不确定性和模糊性,决定了泥石流是一种非常复杂的非线性系统。人工神经网络因其具有较强的自组织性、自适应性和自学习能力等优势,更适合于解决非线性问题。本文基于Matlab程序建立了区域泥石流危险性评价的BP神经网络模型,并将该模型应用于凉山州德昌县22个乡镇的区域泥石流危险性评价中,取得了良好的应用效果,评价预测的准确率高达95%。该方法不仅解决了泥石流危险度评价因子和评价等级之间的复杂非线性关系,而且过程简单,结果不受人为因素的影响,是一种具有应用价值、有效的泥石流危险性评价方法。

Because of the developing environment, the occurring conditions and the randomness, uncertainty and fuzziness of the inducing factors for debris flow, debris flow is a very complex nonlinear system. Artificial neural network is of the stronger self-organization, self-adaptability and self-learning capability, which is specially suited to solve nonlinear problems. BP neural network based on Matlab is set up and applied in regional debris flow hazard assessment in Dechang county of Anninghe watershed. The model gains a better application in practical projects and the accuracy of the assessment is above 95%. The said method solves the complex relationship between the indices and the grades for the hazard assessment of debris flow. It is simple and its assessment results are not easily affected by the human factors. It is effective and is worth to popularization and application in the hazard assessment of debris flow.