经验模态分解和遗传小波神经网络法用于边坡变形预测

发布时间：2019-03-27 17:38

【摘要】：从挖掘边坡变形特性出发,提出一种基于经验模态分解(EMD)和遗传小波神经网络(GA-WNN)法的新型边坡变形预测模型。该模型首先对边坡变形序列进行EMD分解,有效分离出不同尺度特征的子序列;其次基于相空间重构挖掘各子序列的特性,以避免预测模型输入维数选取的随意性;然后采用遗传算法优化小波神经网络的权值和阈值,进而对各子序列建立预测模型;最后叠加各子序列预测值得到边坡预测结果。经过了算例计算,并与SVM和GA-WNN对比分析。结果表明:该模型具有较强的非线性拟合和自适应能力;在一定程度上保证较优的局部预测值和较好的全局预测精度,均方根误差为0.68 mm;在边坡变形预测中具有一定的实用意义。
[Abstract]:A new slope deformation prediction model based on empirical mode decomposition (EMD) and genetic wavelet neural network (GA-WNN) is proposed in this paper. Firstly, the slope deformation sequence is decomposed by EMD to separate the sub-sequences of different scale features effectively, secondly, the characteristics of each sub-sequence are mined based on the reconstruction of phase space to avoid the random selection of input dimension of the prediction model. Then genetic algorithm is used to optimize the weights and thresholds of wavelet neural network, and then the prediction model of each sub-sequence is established. Finally, the prediction results of slope are obtained by adding the predicted values of each sub-sequence. A numerical example is given and compared with SVM and GA-WNN. The results show that the model has strong nonlinear fitting and adaptive ability, and to a certain extent, the better local predictive value and better global prediction precision are guaranteed, and the root mean square error is 0.68 mm;. It has certain practical significance in slope deformation prediction.
【作者单位】：桂林理工大学测绘地理信息学院;广西空间信息与测绘重点实验室;
【基金】：国家自然科学基金项目(41461089) 广西自然科学基金项目(2014GXNSFAA118288) 广西“八桂学者”岗位专项经费资助项目广西空间信息与测绘重点实验室课题(桂科能130511402;130511407)
【分类号】：TU196.1

【参考文献】