Research On Object Recognition Based On Sound Field Simulation Platform Of Ultrasonic Array

To address the limitations of the impact of radiation, electromagnetic interference difficult to work in the earthquake, nuclear radiation, tsunami, and other high-risk environment, the object identification technical solution based on ultrasonic sensor array is designed.To address the limitations of the detection process cumbersome, too time-comsuming, high cost in ultrasonic testing, the experiment platform based on ultrasonic phased array sound field simulation is developed and the object identification algorithm is researched on the platform.Open GL graphics and CAD design software are combined to build the objects, and Open GL and 3DSMax are combined to build the three-dimensional scene in VS2010. The object built by the CAD graphic design software is saved as STL file of ASCII format and is imported into the three-dimensional scene by reading the data of triangular facets. Two simple cylinders built by Open GL and round sides indicate the transmitting and the receiving array elements. The transmitted and reflected sound lines are determined by traversing the gravity of the triangular patches of the objects according to the ultrasonic propagation model in the air, ray tracing method, law of reflection.According to the acoustic radiation directivity of transducer in the air, the sound field of ultrasonic transmitting array elements is modeled. The ultrasonic echo signal is collected by the oscilloscope when the transmitting array element is single and the receiving array element is single, and the echo signal waveform is fitted by Matlab tools, and the original signal of the ultrasonic transmitting array element by using calibration method.Considering the attenuation of ultrasonic in the air and the absorption attenuation of the objects, the signal of transmitting and receiving ray is obtained according to the original signal and the acoustic radiation directivity. The echo signal of receiving array in the platform is obtained averaging the acoustic pressure signals of the receiving rays.The echo signal obtained in the simulation platform and that captured in the real scene are analysised, and the relative error is compared 7 spaced amplitude points. the relative error is less than 5% indicating that the simulation platform is valid.The echo signal obtained in the simulation platform and that captured in the real scene are processed by Fourier transform to extract the normalized characteristics of amplitude and phase to obtain the distance of spectral characteristics. Genetic algorithm is used to search the minimum distance of spectral characteristics when the object is translated and rotated to realize the object recognition. Many experiments and analyses show that the simulation platform can recognize the objects such as cylinder, square pyramid, cone and the recognition rate is over 90%.