Research On Acoustic Signal Processing Of Water Leakage In Underground Water Supply Pipeline Based On Compressed Sensing

In the current context of water scarcity,the waste of water resources is becoming more and more serious,and the leakage of underground water supply pipeline will not only cause water waste but also affect people’s normal production and life.In recent years,the development status of water leakage monitoring at home and abroad shows that most of the water leakage monitoring methods mainly rely on wireless sensor nodes with low power consumption,low cost and strong self-organization ability,but the nodes themselves have limited storage and insufficient energy.Therefore,the leakage signal collected by the node will lose some valid information during the transmission process.After research,it`s found that compressed sensing theory,as an emerging sampling theory,can make up for the shortcomings of the nodes and effectively solve the problems arising in the signal acquisition process,so it is proposed to use compressed sensing theory to solve the problems arising in the signal acquisition process.In this thesis,the main research contents to address the above problems are as follows:（1）Through previous research,it was found that the acoustic signal of underground water supply pipeline leakage has two important characteristics: time-varying and low-frequency,that is,the signal in the low-frequency band changes with time,so the signal needs to be pre-processed.It is verified by experiments that the method of adding windows by frame is more suitable for pre-processing the acoustic signal of underground water supply pipeline leakage.（2）After pre-processing the signal,the sparse basis,observation matrix and reconstruction algorithm suitable for processing the signal were selected through several sets of comparative simulation experiments,and a set of compression perception methods suitable for processing the sound signal of underground water supply pipe leakage was found theoretically,and an improvement idea is proposed for the reconstruction algorithm:when initialising the residuals,make the residuals equal to the best output after adaptive filtering,which can reduce the noise interference while retaining more low-frequency signals to facilitate the reconstructed results.This improved method effectively improves the reconstructed success rate and obtains the reconstructed signal with water leakage information.（3）In order to implement the compressed sensing theory in hardware,the actual signals collected by the nodes are further processed and studied,and the complexity of the operation needs to be reduced,so an observation matrix suitable for implementation in hardware is constructed: the time intervals at different times in the sampling interval satisfying the Gaussian distribution are stored in the observation matrix to be constructed,and these values are randomly distributed on the diagonal of the matrix so that the points on each row roughly satisfy the diagonal distribution.The performance of the constructed observation matrix was verified by reconstructing a 100 Hz sine wave,and the results showed that the reconstruction success rate using the constructed observation matrix could reach 98.06%,and a high reconstruction accuracy was obtained.（4）In the hardware implementation,the IOT-NODE433 node is connected with the LM386 sound sensor as the sending node;the IOT-NODE433 node is connected with the PC end as the receiving node.The sending node sends the obtained sampling value to the receiving node through sparse sampling,and uses the constructed observation matrix to compare with the universal random Gaussian observation matrix.After obtaining the observed values,the reconstruction is carried out using the improved reconstruction algorithm in this thesis,and the reconstruction results prove that the constructed observation matrix is more suitable for processing the signals collected by the nodes and is better for implementation in hardware.In summary,a set of compressed sensing method is found to be theoretically suitable for processing the sound signal of underground water supply pipeline leakage,and it is proved through simulation experiments that compressed sensing can effectively compensate for the lack of storage and energy of nodes and recover reconstructed signal of water leakage information;in hardware implementation,by using devices such as IOT-NODE433 nodes and sound sensors,the compressed sensing method for the acoustic signal of underground water supply pipeline leakage is verified,and the observation matrix constructed in this thesis is used for the acoustic signal of underground water supply pipeline water leakage collected by the node,and the reconstruction result with high accuracy is obtained,and the reconstruction success rate is 97.89%.This result proves that the observation matrix constructed and improved reconstruction algorithm in this paper can effectively reduce the computational complexity and is more suitable for hardware implementation.