Vehicle Acoustic Target Detection Identification Technology And Engineering Realization

Acoustic target identification is an important part for the target detection system, with the coming of information battlefield, some related technologies have been paid attention by every scientific research institution and colleges or universities, and the stage achievement has been obtained. Especially, the development of artificial intelligence theory, bionics theory, signal processing theory and integration process of thick film provide the effective technological support for realizing the acoustic target identification system. However researches are insufficient for these key problems such as pretreatment, feature extraction, classifier and hardware implementation of the non-stationary signal. For these problems above, the key technology of acoustic target signal identification and the implementation of hardware algorithm have been studied.Combined with modern digital signal processing theory, target acoustic signal of actual collected has been taken theory investigation and algorithm verification, which includes producing mechanism analysis of the vehicle noise source, the target acoustic signal's pretreatment, feature extraction, classifier design and hardware engineering implementation of embedded digital signal processor (embedded DSP). In the pretreatment process of target acoustic signal, specific theoretic analysis and computer testing verification are taken through adopting FIR filter and LMS filter, combining the power spectral density theory; AR parameter model, wavelet energy model and the model of imitating auditory sense are used to pick up the feature of target acoustic signal in the process of feature extraction; based on the theory analysis, the parameter in the algorithm can be adjusted to pass through the computer testing verification; the system clustering algorithm based on from theory is adopted in the process of designing classifier, it can classify and identify four kinds of target acoustic signals effectively, which belong to jeep, tank, speech signals and background environment.In the implementation of hardware system, the hardware algorithm platform is adopted in this design the core of which is embedded DSP (TMS320F2812), it includes acquisition conditioning amplified circuit of the electret acoustic sensor, clock frequency setting module, signal acquisition module, serial communication module and so on. That makes the acoustic signal's acquisition, conditioning, amplification, pretreatment, feature extraction, classification and identification come true, meanwhile the TMS320F2812 hardware transplantation of every module algorithm in the identification system is achieved, part of algorithms are optimized to improve the accuracy of identification in the embedded DSP.