Research On UAV Location Method Based On Microphone Array

With the reduction of the autonomous control threshold of the Low,Slow and Small Unmanned Aerial Vehicle(LSS-UAV),the "unregistered flight" incidents of UAVs are frequent,causing major security risks.In order to detect the LSS-UAV's intrusion as early as possible,methods such as radar detection,data link tracking,human eye lookout and image recognition have been developed.However,due to the factors such as high equipment cost and limited application scenarios,the existing methods still have limitations.Considering the noise generated by propeller rotation during the LSS-UAV flight,the sound source localization algorithm is used in this thesis to locate LSS-UAVs.Thanks to the advantages of simple assembly and high openness,the sound source localization system can realize the large-scale coverage of the core monitoring area,and be combined with the existing LSS-UAV localization technology.When the existing technology is restricted,the sound source localization system can be used as an auxiliary means to ensure the normal operation of the positioning system,and provide a supplementary plan for the effective localization of LSS-UAVs.The main work of this thesis is as follows:(1)Analyze the advantages of microphone array sound source location system in unmanned airport scenario,analyze the principles and compare the characteristics of Conventional Beamforming(CBF)algorithm,Multiple Signal Classification(MUSIC)algorithm and Time Difference of Arrival(TDOA)algorithm,and choose the TDOA algorithm with small amount of computation and high real-time as the positioning algorithm of the UAV.The frequency spectrum and background noise spectrum of the UAV during flight are analyzed,and 5KHz~10KHz is selected as the filtering frequency band.(2)Based on the array layout of plane quaternion uniform cross array and plane five-element uniform cross array,a nine-element uniform cross array type is proposed,and the geometric principles of near-field and far-field Direction of Arrive(DOA)estimation for three arrays are derived.The formulas for estimating the azimuth estimation error and pitch error of three arrays layout in far-field models are derived.At the same time,the average consuming time of the far-field DOA estimation for the three arrays layout is quantitatively analyzed.Finally,the nine-element uniform cross array which takes into account the positioning accuracy and real time is chosen as the microphone array of the far-field DOA estimation of UAVs.(3)In the environment of low Signal to Noise Ratio(SNR),the Phase Transform(PHAT)weight function has low accuracy in time delay estimation.To improve the PHAT weight function,this paper chooses to raise the SNR of sampled signals.Taking into account the UAV voice signals collected in three scenarios,simulation results show that the improved weight function can reduce the time delay estimation error.The improved weight function is applied to the nine-element uniform cross array acoustic source localization system,and the far-field DOA estimation of the UAV is carried out.The measured results show that the improved weight function reduces the azimuth estimation relative error by 3.18% compared with the PHAT weight function,and the pitch angle estimation relative error is reduced by 1.77%.(4)Analyze the performance of microphone array and explore the influence of array layout on array performance.For the spatial confusion problem of uniform arrays,genetic algorithm is used to optimize the array layout.The simulation results show that while the aperture of the array remains unchanged,the one-dimensional twenty-element uniform linear array can be optimized to a ten-element non-uniform linear array.After optimization,the maximum sidelobe level of the array is increased by 0.53 d B and the sidelobe suppression ability is improved.When the number of array elements remains unchanged,the nineelement uniform cross array can be optimized to a nine-element non-uniform cross array.After optimization,the maximum sidelobe level of the array is increased by 9.39 d B,the main lobe width is reduced by 0.04%,and array sidelobe suppression and array resolution are both improved.Compared with the nine-element uniform cross array,the optimized nineelement non-uniform cross array used in UAV measurement has larger array element spacing,which can effectively reduce the delay estimation error and improve the DOA estimation accuracy.