摘要
针对滑坡位移时间序列含大量噪声、具有明显的非线性等特征,本文提出了一种基于粒子群优化(PSO)的小波分析(WA)-支持向量机(SVM)滑坡位移预测模型(即WA-SVM模型)。该模型在混沌分析的基础上,首先用WA将滑坡位移序列分解成不同频率的分量,然后采用PSO算法优选SVM模型参数,并利用SVM模型预测各分量值,最后将各分量预测值组合得到最终预测值。结合滑坡位移序列实例,将基于粒子群优化的WA-SVM模型的预测结果与WA-BP模型、单独SVM模型进行对比,结果表明该滑坡位移序列存在混沌特性,基于粒子群优化的WA-SVM模型克服了噪声的干扰和参数优选的问题,具有较高的预测精度和预测效率,为滑坡位移的预测提供了一种新的方法。
To deal with the time series of landslide displacement containing a large amount of noise and showing obvious characteristics of nonlinear,this paper applies a wavelet analysis and support vector machine（WA-SVM）model based on particle swarm optimization to predicting the displacement of landslide.On the basis of using chaos theory,firstly,this model decomposes landslide displacement sequence into different frequency components through wavelet analysis;then,it selects the parameters of SVM by particle swarm optimization and predicts each component through optimal SVM model;finally,it integrates the prediction results of each component into ultimate forecast value.Combined with the examples of landslide displacement,the paper compares the results of the WA-SVM model based on particle swarm optimization with the results of WA-BP model and SVM model,which shows that the presence of landslide displacement has chaotic characteristics,and that the WA-SVM model based on particle swarm optimization solves the problem of noise and parameter optimization and has higher efficiency and accuracy so as to provide a new approach to forecast landslide displacement.
出处
《安全与环境工程》
CAS
北大核心
2014年第4期13-18,共6页
Safety and Environmental Engineering
关键词
滑坡
位移预测
混沌
粒子群优化
小波分析
支持向量机
landslide
displacement prediction
chaos
particle swarm optimization
wavelet analysis
support vector machine
