Research On Microwave Heating Real-time Detection System Based On Infrared Spectrum Analysis

Microwave heating is convenient and fast, the evenness of microwave heating is an important program, and the temperature measurement in microwave field is also a vital issue to be considered. Besides, it’s significant to research how microwave affects dielectric materials and use which method to detect the compositional change of dielectric material after microwave heating.The purpose of this graduated thesis is to research the issues of microwave heating and near infrared spectroscopy detection based on the microwave heating real-time detection system, including design the sample box in the rectangular waveguide by HFSS software in order to guarantee the minimum reflection coefficient and uniform electric field distribution when dielectric materials is heated in microwave, then design the cut-off waveguide of real-time detection system to combine the single mode microwave heating system with infrared spectrometer. This thesis analyses the factors that affect the evenness of microwave heating, using fluorescence fiber temperature sensor for the uniformity experiment of microwave heating, At last, this thesis research the noise reduction process of near-infrared spectroscopy detected from the material after heated by the system. This thesis completes the simulation analysis of the sample box and the waveguide in the real-time detection system, researches the distribution uniformity of electric field by HFSS simulation software based on finite element method, finds the higher electric field to heat the material and designs the cut-off waveguide in real-time detection system. Then the fluorescence fiber temperature sensor based on fluorescence lifetime is used in the experiment to test the evenness of microwave heating and the experiments about evenness containing pure water, pure milk, dough and prawn slices.This thesis realizes the pretreatment of near infrared spectroscopy to eliminate the spectrum noise by two methods of wavelet transform and wavelet packet transform. Through different wavelet bases, order numbers, decomposition layer numbers and appropriate threshold value, reducing the near infrared spectroscopy’s noise and comparing the different noise reduction effect of the signal-to-noise ratio and root-mean-square error, this thesis obtains the conclusion that the best way to reduce noise is to use sym8 wavelet base for level 3 decomposition, the threshold estimation method is soft-threshold method. The NIR spectrum becomes smooth after the noise reduction, noise peaks are perfectly eliminated, absorption peaks become more prominent and the distortion is suppressed.