论文部分内容阅读
泥石流的流速预测是泥石流灾害防治的核心问题之一。由于泥石流流速的影响因素众多,需要寻求能够综合反映泥石流流速影响因素的预测模型。移动最小二乘法(MLS)预测模型具有自学习和自组织及捕捉到影响因素数值微小变化的能力,可以解决泥石流流速预测存在的一些问题。采用云南蒋家沟泥石流流速实测数据作为训练样本和预测样本,以泥深、比降、密度、颗粒的平均粒径作为输入因子。讨论了用MLS方法进行泥石流平均流速预测的可行性与有效性,并将预测结果与经验公式、BP神经网络以及支持向量机进行了对比。结果表明,MLS方法的最大预测误差为4.6%,平均误差为2.7%,预测精度优于经验公式、BP神经网络及支持向量机方法。MLS方法可以为泥石流防治提供更准确的科学依据。
Debris flow velocity prediction is one of the core issues in the prevention and control of debris flow hazards. Because of the many influencing factors of debris flow velocity, we need to find a prediction model that can comprehensively reflect the influencing factors of debris flow velocity. The MLS prediction model possesses the ability of self-learning, self-organizing and capturing small changes of influencing factors, which can solve some problems in the prediction of debris flow velocity. The measured data of mudflow velocity of Jiangjiagou in Yunnan Province were used as training samples and prediction samples. The muddy depth, specific gravity, density and the average particle size of particles were taken as input factors. The feasibility and effectiveness of using MLS method to predict the average flow velocity of debris flow are discussed. The prediction results are compared with the empirical formula, BP neural network and support vector machine. The results show that MLS method has the maximum prediction error of 4.6% and the average error of 2.7%. The prediction accuracy is better than that of empirical formula, BP neural network and support vector machine. The MLS method can provide a more accurate scientific basis for debris flow prevention and control.