Optimization And Application Of Endpoint Effects In Local Mean Decomposition

Local mean decomposition is a new method to adaptively deal with nonstationary signals.LMD as a new method of data decomposition,there are still some shortcomings in the algorithm,such as endpoint effect.The endpoint effect causes the components of data decomposition deformational at both ends.The effect of the endpoint effect is amplified when the LMD is used to extract feature and analysis deformation prediction of deformation data.In this paper,we will use the new method to suppress the endpoint effect of LMD and analyze it with examples.The theory of LMD method is studied and analyzed,and the existing problems and the direction of LMD method are summarized.The causes of the endpoint effect of LMD are analyzed and the reason of the end effect is illustrated by the simulation signal.The paper use a new continuation method that the SVM extension method based on the extreme point and distance,the actual engineering signal data is used as an example to verify that the continuation method can suppress the endpoint effects of LMD to a certain degree.Another method of suppressing endpoint effect of LMD is the adaptive iteration method.And the modified LMD method is used to decompose and analyze the deformation data of the dam and compared with other commonly used methods to suppress the LMD endpoint effects.It is proved that the optimized LMD method can effectively extract feature of deformation data.In the same way,the optimized LMD method is used to extract the trend items for one hundred years of sea level change data and compared with other commonly used methods to suppress the LMD endpoint effect.It is proved that the optimized LMD method extract the trend items of sea-level changes more effective than other methods.The LMD-SVM-GM(1,1)model is optimized by using the optimized LMD method(adaptive iteration method for suppressing endpoint effect of LMD)and SVM model and GM(1,1)model.The author use two different examples that storage power plant dam monitoring data and Poyang Lake into the lake total water data to verify themethod.The monitoring data is decomposed,the margin of decomposition and the other components are predicted by GM(1,1)model and SVM model respectively.The respective prediction results are re-fused to get the final forecast results.The experimental results show that LMD-SVM-GM(1,1)is suitable for multi-scale deformation monitoring data and is worth popularizing.