Pipeline leakage signal compressed sensing based on wavelet packet hierarchical tree model

Aiming at the problems that in current pipeline leakage signal compressed sensing processing the required number of measurements is high, the performance index of leakage signal reconstruction is low under high compression ratio, and the inherent structural characteristics of pipeline leakage sparse signal are not fully considered, in this paper, a new method of pipeline leak signal sparse representation is proposed based on the wavelet packet hierarchical tree model; based on the characteristics of the sparse structural model of these signals, a new compress sensing method, named model-fast Bayesian matching pursuit (M-FBMP), is developed. The proposed method fully makes use of the characteristics of the pipeline leakage signal to select the sparse basis reasonably, constructs the wavelet package hierarchical tree model of the pipeline leakage signal, optimizes the standard Gaussian measurement matrix; then the M-FBMP algorithm for the pipeline leakage signal is deduced, the compressed sampling and reconstruction of the signal is achieved. The proposed method and the traditional fast Bayesian matching pursuits (FBMP) method were compared in experiments, the performance indices of the two compressed sensing methods are given under different data compression ratios. The experiment results show that under the same MSE and reconstruction SNR condition, the proposed method has a higher visual quality, and can improve the data compression ratio by 3.4 times compared with the traditional method. This algorithm can also decrease the requirement of network communication bandwidth in industrial data monitoring transmission system; under the data compression ratio of 30:1, the proposed method improves the reconstruction SNR by almost 2 times, the ERP is better than 1 × 10-3. The accuracy of the reconstructed signal can meet the requirement of the detection and positioning in pipeline leakage detection, and the proposed algorithm provides a new method for pipeline leakage detection.