Research On Pipeline Leakage Detection Based On MEEMD-KF And SVM Optimization Model

Because pipeline transportation is widely used in various industries,so it is particularly important to study the safety of pipeline signals.In this paper,three kinds of pipeline signals collected in the pipeline leak detection experimental platform are studied and analyzed,and a pipeline signal denoising model using modified ensemble empirical mode decomposition algorithm combined with Kalman filter algorithm is proposed.In order to further verify the effectiveness of the algorithm,this paper uses dispersion entropy combined with fuzzy entropy feature extraction method to construct the pipeline signal feature vector.Finally,POA is used to optimize the penalty factor and kernel function parameters in SVM,and the feature vector is input into the model to identify the pipeline working conditions.For pipeline denoising pre-processing,this paper studies the principles of various adaptive decomposition algorithms and their improved algorithms,mainly studies the modified ensemble empirical mode decomposition algorithm and Kalman Filter,and establishes a MEEMD-KF denoising model for pipeline signals.In this model,each component is judged and screened by the arrangement entropy of components,and the dominant noise component is processed by scalar Kalman Filter and reconstructed with the normal component,so as to realize the signal denoising processing.Simulation experiments show that the algorithm proposed in this paper is effective,and comparison with other algorithms shows the superiority of the denoising algorithm.Using the pipeline signals collected on the pipeline leak detection experimental platform for verification,the results show that the algorithm is effective for pipeline signal denoising and its performance is more stable.For feature extraction of pipeline signals,the feature extraction method based on dispersion entropy and fuzzy entropy is put forward because the feature extraction of pipeline micro-leakage signal is not easy to extract.Through the research of this algorithm,it is found that the feature entropy can effectively characterize the feature of pipeline signal.After using MEEMD-KF algorithm to denoise three kinds of pipeline signals,the dispersion entropy and fuzzy entropy are calculated and the feature vectors are constructed.It is found that the discrimination of entropy values of three kinds of pipeline signals is obvious,which shows that the feature extraction method based on dispersion entropy and fuzzy entropy is effective in processing pipeline signals.In order to validate the effectiveness of the algorithm proposed in this paper,for the purpose of classification and identification of pipeline working conditions,the pelican optimization algorithm is used to optimize the penalty factor and kernel function parameters in SVM by studying the principle of support vector machine,and the extracted features are sent into the model and a new POA-SVM model is obtained by training,which solves the problems of long optimization time and low accuracy in traditional SVM.Through experimental comparison,it is proved that using POA to optimize SVM model for pipeline working conditions identification can further improve the efficiency and detection accuracy of the identification model.