Research On Explosives Detection System

Over the years, terrorists’hazardous activities are very active in parts of the world, and explosive destruction is also in an increasing trend. With the increasing of terrorist activities, the difficulty of preventing terrorist attacks is increasing, and the urgent demand for explosives detection and explosion-proof measures has emerged. Explosive detection as a very important antiterrorist means focuses on the development of new explosive detection equipment and technique which vigorously improve the validity and reliability of explosive detection in order to curb terrorist activities in the attempted state.The main subject study TNT, RDX, HMX, ammonium nitrate explosives trace detection, signal processing, and some computer processing algorithms. Existing explosives detection technologies rely mainly on the physical, biological, chemical, and other technical means for detecting and identifying explosives. Explosives detection system based on molecularly imprinted piezoelectric biosensor is designed in this paper, mainly including the sensor oscillator circuit waveform conversion mixer-difference frequency circuit, the MSP430F249microcontroller micro control system and the host machine LS-SVR software algorithms. The least squares support vector machine regression algorithm is exploited which can effectively detect explosives with the limited sample data.A lot of experiments have been studied,and molecular imprinting piezoelectric wensors which are signal acquisition components are used in the experiment. By the frequency corresponding to the sensors output, a qualitative analysis of the measured object is deployed. In order to achieve a quantitative analysis of the measured object, the host computer software algorithm LS-SVR trains the data samples to get modeling parameters, and further a regression model based on MCU measurement of the difference frequency is established. The experimental results show that the MCU measured frequency error is less than0.05%, the mean square error of analytical model is0.000268183and the correlation coefficient99.7641%. These conclusions above shows that the system desingned in this paper is able to detect trace TNT.