利用经验模态分解和LSSVM预测隧道不均匀沉降

• 作 者: 汤俊 邹自力 张晓平
• 机 构: 东华理工大学资环系,东华理工大学地球科学与测绘工程学院,
• 出 处: 测绘科学
• 发表日期: 2011年03期
• 关键词: LSSVM 不均匀沉降 经验模态分解 二乘 原始序列 不均匀沉降预测 支持向量机 预测模型 核函数 预测值

隧道不均匀沉降是个复杂的系统。针对其非线性、非平稳的特点,本文将经验模态分解(EMD)和最小二乘支持向量机(LSSVM)引入该领域,建立了一种基于EMD和LSSVM的预测模型。首先,利用EMD方法将原始序列作平稳化处理,得到一系列本征模态分量(IMF);其次,根据各个IMF的变化规律,采用合适的参数构造不同的LSSVM进行预测;最后合成原始序列的预测值,并与单一的LSSVM模型和AR模型相比较,结果表明,该方法具有较高的精度和较强的泛化能力。

Uneven tunnel subsidence is a complex system. In view of its non-linear and non-stationary characteristics, this paper introduces empirical mode decomposition (EMD) and least square support vector machine (LSSVM) into this field and establishes a prediction model based on EMD and LSSVM. First, the EMD method is used to stabilize the original sequence to obtain a series of IMFs. Secondly, different LSSVMs are constructed using the appropriate parameters according to the variation law of each IMF. Finally, the original sequence Compared with the single LSSVM model and the AR model, the results show that this method has higher accuracy and stronger generalization ability.