Study On Methods Of Acoustic Vector Array Signal Processing Based On Quaternion

Acoustic vector sensor array signal processing is one of the new research topicsin the array signal processing field. Analyzing and processing the received acousticvector signals from the sensor array has become a key factor to reflect the superiorityof the acoustic vector sensor. The acoustic vector sensor can provide the vibrationvelocity of the sound field which is not provided by the traditional sound pressuresensor. Therefore, the performance of the acoustic vector sensor technology issuperior to that of the traditional sound pressure sensor array signal processingtechnology.The traditional acoustic vector sensor array signal processing methods do notemploy the orthogonality between each output component of the acoustic vectorsensors and the superiority of the acoustic vector sensor technology isn’t reflected.Hence, we need to find new ways to reflect the superiority of the acoustic vectorsensor technology. Employing the quaternion theory in the field of acoustic vectorsensor signal processing, the advantages of the acoustic vector sensor can be reflectedbetter.In this thesis, we study for quaternion theory based acoustic vector sensor arraysignal processing method on the basis of quaternion theory and acoustic vector sensorarray signal processing theory. We employ the quaternion to express the soundpressure and particle velocity output of the acoustic vector sensor and derive thequaternion model of the acoustic vector sensor. Moreover, we derive the quaternionmodel of the acoustic vector sensor of different array structure. On the basis of thesemodels, we study the direction of arrival (DOA) estimation method of array signals.For the L-type, the double parallel line and the plane acoustic vector sensor arrayquaternion models, we propose the quaternion MUSIC algorithm. We derive thespectrum estimation formula of the quaternion MUSIC algorithm and compare theestimation performances of the quaternion MUSIC algorithm and those of thetraditional vector MUSIC algorithm after the detailed theoretical analysis. For theuniform linear array and the L-type acoustic vector sensors with two dimensions, wepropose the quaternion ESPRIT algorithm. We derive the DOA estimation formula ofthe acoustic source and compare the estimation performances of the quaternion ESPRIT algorithm and those of the traditional vector ESPRIT algorithm after thedetailed theoretical analysis.The theoretical analysis and the simulation results indicate that the DOAestimation performances of the acoustic source signal by the proposed quaternionMUSIC algorithm and the quaternion ESPRIT algorithm are better than those of thetraditional MUSIC algorithm and the ESPRIT algorithm.