Research On Direction-of-Arrival Tracking Algorithm Based On Acoustic Array

With the rapid development of unmanned aerial vehicle(UAV)technology in the military and civil fields,there is a growing demand for the development and application of anti UAV technology.Especially with the gradual opening of low altitude airspace in China,the emergence of “low,small and slow” aircraft makes the existing low altitude defense system face new challenges.For the traditional radar and photoelectric detection technology,its detection performance is susceptible to ground buildings,unobvious doppler effect,small radar cross-section and bad weather environment,making “low,small and slow” target detection become a worldwide problem.Passive detection technology based on acoustic array has become an effective way and research hotspot to solve the problem of “low,small and slow” target detection because of its high concealment,convenience,non electromagnetic interference and all-weather working characteristics.According to the practical application requirements of “low,small and slow”acoustic target detection,based on different application scenarios of acoustic array,this dissertation has carried out exploratory research on the test platform of air acoustic array detection system and the theory and methods of direction of arrival(DOA)tracking using linear acoustic array,L-shaped acoustic array and coprime acoustic array.In order to verify the practicability of acoustic array detection algorithms,a test platform of acoustic array detection system is designed.The performance of the existing array DOA tracking algorithms are studied through outdoor acoustic experiments,the performance of each algorithm for DOA tracking is compared and analyzed,and the application conditions and processing methods of each algorithm are concluded.The experimental study of MUSIC-like algorithm is carried out.Aiming at the limitation that traditional beamforming and subspace methods need the number of sources to be known,an improved MUSIC-like algorithm based on diagonal loading technique is proposed.The experimental results based on linear acoustic array show that the proposed algorithm can effectively suppress the appearance of false peaks in the single or dual source scenarios,and can significantly improve the robustness of the algorithm in practical application.The above experimental results also verify the effectiveness and practicability of the test platform.In order to realize fast DOA tracking of weak acoustic target in two-dimensional space,DOA tracking using uniform linear array has been studied in this dissertation.According to the shortcoming of the subspace tracking method or the traditional PF algorithm,that is the tracking precision descend and bad real-time performance due to low SNR enviroment and the direction-of-arrival(DOA)of maneuvering targets changing rapidly.An improved PF tracking algorithm based on the spatial spectrum function of the PM algorithm is proposed.The spatial spectrum function of PM algorithm is used to construct the likelihood function of the PF algorithm,which does not need to perform eigendecomposition of the array convariance matrix,and can reduce the calculation time of the algorithm while retain the high-resolution characteristics of subspace algorithm,the improved likelihood function has the ability to update particles more efficiently.Theoretical analyses and simulation results are presented to demonstrate the effectiveness of the proposed PF-based algorithm,compared with subspace tracking algorithm and traditional PF algorithm,the proposed algorithm has higher calculation efficiency and lower tracking root mean square error which is more suitable under the conditions of fast change of DOA and low SNR.Finally,based on the prototype of the acoustic detection system,the array data acquisition experiment of outdoor moving acoustic source is carried out.Results from real acoustic data processing are provided to demonstrate the effectiveness of the PF-based algorithm.In order to realize the DOA tracking of coherent acoustic targets in three-dimensional space,2-D DOA tracking using L-shaped array has been studied in this dissertation.According to the problem that the traditional beamforming and subspace methods need to perform spectral peaks search in 2-D angle space,which leads to a large amount of calculation and can not deal with the DOA tracking problem of coherent sources,and the problem that the traditional sparse signal reconstruction methods need to construct an over complete atomic database in two-dimensional angle space,which leads to an exponential increase in the computation in the process of sparse signal reconstruction,and can not solve the matching problem between the azimuth and the elevation under scenario of multi-targets.Two DOA tracking alogorithms based on sparse bayesian learning(SBL)algorithm with two different L arrays are proposed.For the first alogorithm,the sparse signal reconstruction problem of the two-dimensional over complete atom library is reduced to two signal reconstruction problems of one-dimensional over complete dictionaries,which can effiently reduce the computational complexity.Then,SBL algorithm is applied to reconstruct the amplitude difference between signals and complete the parameters to be estimated Automatic pairing between numbers,the proposed algorithm is suitable for DOA estimation of coherent sources.For the second alogorithm,according to the conjugate symmetry of the array covariance matrix,the problem of 2D DOA sparse reconstruction of L-array is transformed into two sparse signal reconstruction problems using linear array extended aperture.Then,SBL algorithm is used to reconstruct the sprase signals respectively,and the matching problem of the parameters to be estimated can be solved according to the difference of the source signals amplitude.Theoretical analysis and simulation results show that the proposed algorithm has lower computational complexity,larger array aperture and lower DOA estimating error.The experimental results verify the effectiveness of the proposed algorithm.In order to achieve high efficiency and high precision DOA tracking of acoustic source target in special application scenarios,DOA tracking using coprime array has been studied in this dissertation.According to the problem that the ULA has limited angular resolution due to Nyquist sampling theorem,a linear coprime array is constructed by sparse linear array,and a PF DOA tracking algorithm with the coprime array is proposed.Facing the issue that the ability of the PF to update the particles will be reduced in the case of low SNR or small number of snapshots,we reconstruct the likelihood model based on the SSMUSIC spatial spectrum,however,the acquisition of SS-MUSIC spatial spectrum function need perform eigendecomposition of the array convariance matrix,resulting in larger calculation load.To get around this,the SS-PM spatial spectrum function of the coprime array is derived and applied to reshape the likelihood function.Through the theoretical analysis and numerical simulation,compared with the spatial smoothing beamforming algorithm,spatial smoothing subspace algorithm and sparse reconstruction algorithm,the proposed algorithm exhibits better performance in terms of the computational complexity and tracking accuracy.The experimental results verify the effectiveness of the proposed algorithm.