Research On Gas Leakage Localization Based On Compressive Sensing

Gas leakage has been occurring occurred frequently in recent years,which brings great harm to people's lives and property.The present detection and localization methods to gas leakage show disadvantages on long detection period and poor localization accuracy.Hand-held ultrasound detection equipment which greatly depends on manual operation can hardly deal with the work which demands real-time detection and localization.Therefore,a method based on microphone array processing is studied in this thesis.And according to the distance between the gas leakage and microphone array,the gas leakage is divided into far-field condition and near-field condition.The work of this thesis is arranged as follows:?1?The DOA estimation research on gas leakage based on weighted l₁-norm.First of all,the l₁-SVD algorithm is applied to deal with the DOA estimation of the gas leakage.However,the l₁-SVD algorithm not only has a disadvantage on dealing with the DOA estimation of the coherent sources,but also has the problem that pseudo peaks appear in the spatial spectrum under the condition of few snapshots and low signal noise ratio.An improved algorithm called l₁-SVD-TMusic algorithm,of which the weighted function is the improved EST-MUSIC spatial spectrum function based on the MUSIC algorithm,is proposed to overcome the problem.Lastly,the traditional DOA estimation algorithm,l₁-SVD algorithm and the improved l₁-SVD-TMusic algorithm are simulated via matlab and comparison among them is made.And it is proved that the improved algorithm can solve the DOA estimation problem of coherent sources and effectively restrain the appearance of the pseudo peaks in spatial spectrum.?2?The DOA estimation research on gas leakage based on COBE algorithm.At present,the DOA estimation of gas leakage based on compressed sensing has the problems of large computation and high sampling rate.In order to solve the problem,the COBE algorithm based on time-domain compression is applied to deal with the DOA estimation of gas leakage.However,COBE algorithm has the limit that it is only capable of the one-dimensional DOA estimation.In order to improve the spatial applicability,an improved two-dimensional DOA estimation named 2D-COBE algorithm is proposed.Finally,the effectiveness and accuracy of the proposed algorithm is proved by simulation experiments.?3?The near-field narrow-band gas leakage DOA estimation research based on compressed sensing.At present,the near-field acoustic source localization problem is mostly based on one-dimensional linear array.In order to obtain more spatial information,the near-field sound source localization model is built with planar microphone array.Firstly,the near-field CBF algorithm is simulated based on different planar array models,which proves that random planar array is the best array model.Then,the near-field acoustic source localization model based on compressed sensing is established via performing the sound source plane division and the processing of spatial source sparse representation.After that,the near-field l₁-min algorithm and l₁-SVD algorithm are respectively used to solve DOA estimation problem.However,l₁-min algorithm has the problem of high computational complexity.Although the l₁-SVD algorithm reduces computation to a certain extent,the overall computational complexity of the system is still high.To deal with the problem,an improved SVD-SP reconstruction algorithm is used in the localization of near-field sound source.Three algorithms simulation experiments are carried out under different SNR conditions.The simulation results prove the effectiveness and accuracy of the proposed algorithm and the advantage of computation time.?4?The near-field broadband gas leakage localization research based on compressed sensing.A joint sparse model needs to be established to deal with the wideband sound sources localization.Firstly,the wideband l₁-SVD algorithm is applied to establish acoustic source localization model.To solve the drawbacks of large computational complexity of wideband l₁-SVD algorithm,an improved distributed singular value decomposition subspace tracking algorithm called DCS-SVD-SP algorithm is proved.The simulation experiments of broadband l₁-SVD algorithm,broadband l₁-min algorithm and the improved algorithm are carried out.The simulation proved the effectiveness and accuracy of the improved algorithm and the advantage of computational complexity.