| One of the important aspects in array signal processing is signal source localizing which is accomplished by effectively detecting and accurately estimating to the characteristics of the space signal. It can be classified as far field direction of arrival (DOA) and near field source localization according to the relationship of the distance between signal sources and receiving transducers array and the size of array aperture. The basic theory and algorithm in the field of direction of arrival is mature relatively, on the contrary the research on near field source localization is lag apparently. There are many difficult problems needed to be solved urgently. The first problem is how to improve the estimate accuracy of near field source localization in traditional algorithms and solve the contradiction between the parameter estimate accuracy and the number of computer operation which is commonly encountered in practice. Secondly, so far the research on near field source localization is mostly concentrated on the narrow band signal model, the research on the wideband model is underway oppositely, so there is still a long way to go. Thirdly, the research of utility acoustic source localization is always a difficult problem in the field. According to the problems quoted above, wideband near field source localization, narrow band near field source localization, estimation of time delay of arrival in acoustic source localization are investigated deeply in virtue of high order cumulants and high order cycle cumulants in the paper. Research on auto guided vehicle (AGV)based on ultrasonic localization still have number of difficulties in technologies, for example, localization technology, navigation technology and the power supply technology, etc. To explore algorithm of high accuracy, high real time and easy to realize with DSP is still the central problem in localization as the basic technology of AGV. An AGV system based on ultrasonic localization association with the research result of wireless localization in the paper.is designed.The innovation points of this paper focus on the three aspects as follows. Firstly, research on the localization of wideband near-field source is relatively weak with less ffective algorithms , explore research is made by the author in the field and an algorithm for wideband near-field sources is proposed first time using coherent signal subspace algorithm (CSM). The wideband of near-field source is divided into multiple narrowband frequency bins. Coherent focusing of the received data over multiple frequency bins is studied. A cumulant-based subspace algorithm is then presented to obtain estimated parameters with the Eigen decomposition of the cumulant matrix. Compared with the wideband range and bearing estimation algorithm using MUSIC, the advantages of the proposed algorithm are that multiple-dimension search computation is not required and Gaussian colored noise is effectively removedSecondly, a new algorithm is presented based on fourth order cyclic cumulant matrix and MUSIC algorithm. The algorithm can estimate bearing and range of near field non-Gaussian cyclostationary sources and remove cyclostationary interference with different cyclic frequency from the desired signals and suppress the cyclostationary Gaussian white noise. In addition a fourth-order cumulant based TLS-ESPRIT like 3-D algorithm which utilizes adequately the error between estimation value and observations to pair the estimation parameter is proposed, the algorithm does not require any spectral peak search. The algorithm can be effective only if one of the three parameters is distinct.Finally, the AGV system is designed by incorporated the proposed TDOA algorithm based on cycle correlation technology with the localization model of double transmitting and double receiving ultrasonic transducers is proposed, which, by using the felicitous time sequence of system, can obtain the position and pose information at one time with no use of angle transducer like coder.The main work can be summarized as follows:In introduction, the wireless localization algorithm and development situation, the AGV system and development prospects are analyzed and summarized. The relative mathematics models and algorithms are introduced, which include the relative mathematics model with this paper, the knowledge of matrix algebra, properties of higher order cyclic statistics and higher order cumulant. Then a new algorithm for wideband near-field sources is proposed using coherent signal subspace algorithm (CSM). The wideband of near-field source is divided into multiple narrowband frequency bins. Coherent focusing of the received data over multiple frequency bins is studied to obtain estimated parameters from the eigenvectors and Eigen values of the Eigen-decomposition of the cumulant matrix.The advantages of the proposed algorithm are that multiple-dimension search computation is not required and Gaussian colored noise is effectively removed.Next, to reduce algorithms'compute burden and restrain complex noises a new algorithm based on fourth order cyclic cumulant matrix and MUSIC algorithm is presented. The algorithm can estimate bearing and range of near field non-Gaussian cyclostationary sources through a uniform linear array (ULA) and remove cyclostationary interference with different cyclic frequency from the desired signals and suppress the cyclostationary Gaussian white noise.The problem of parameter estimation in three-dimensional space domain is discussed in chapter five. For the sake of limitations of the traditional pair algorithms and restrain complex noise, the three dimensional parameters estimate algorithm based on four order cumulant is proposed. The proposed algorithm utilizes adequately the error between estimation value and observations to pair the estimation parameter without requiring any spectral peak search, which can be effective only one of the three parameters is distinct, at the same restrain Gaussian noise.In chapter six, the multiple path time delay of arrival (MTDOA) algorithm based on cyclic statistics has been proposed, which can suppress multiple path interference and complex background noise. The estimation performance is prior to the correlation algorithm and it is prior to the high order cumulant algorithm in computer burden and adaptability in background noise.On the basis of introduction of AGV development history, research of AGV localization theory based on ultrasonic is stressed in chapter seven. The localization model of double transmitting and double receiving ultrasonic transducers is proposed, which, by using the felicitous time sequence of system, can obtain the position and pose information at one time without use of angle transducer like coder. Subsequently, the algorithm of dynamic dual-threshold time delay extraction is presented by analyzing the waveform characteristics of ultrasonic signal, and is realized in the practical system; next the two-step correlation time-delay extraction algorithm is presented to extract the time delay of continue points. Having already completing the localization of AGV the lead sight path following navigation algorithm is proposed on the analysis of AGV dynamical model. The system possesses the perfect performance in such aspects: localization precision, real time and agile all shown in the experiments. .At last, the main work of the paper is summarized and the future work is also expected.
|
PDF Full Text Request