Research On Data Analysis Technique Of Ultraweak Luminescence Wheat Kernels With Hidden Insects

The ultra-weak luminescence is the intrinsic function of the organism, and it has closerelationship with the vital processes of life. By measuring and profiling the ultra-weakluminescence signal, we can grasp to the internal change details of biological system andknow the life course better. The detection&analysis technology of ultra-weak luminescencesignal has developed rapidly in recent years. And its application has become a hotspot inbiology, medical, food, etc. Due to late start in our country, the data analysis methods in theultra-weak luminescence are simple and single. To find the sensitive characteristics which candescribe the relationship between ultra-weak luminescence signal and biological nature ofwheat, the ultra-weak luminescence signal of wheat varietals and wheat with hidden insect isanalyzed by various methods, which can be the foundation for the further study of hiddeninsects’ detection. The main work is as follows:Firstly, because of ultra-weakness, the ultra-weak luminescence signal may be easilyinterfered by the noise during the measurement. It is necessary to choose a datapre-processing method reducing the influence of noise to extract the useful signal. Aftersubtracting the background signal, the wavelet is adopted to de-noise in the preprocessing.Besides that, preliminary discussion for the application of the adaptive canceller in theultra-weak luminescence signal is served. It shows that these two de-noising methods areworkable to reduce the background signal.Secondly, the ultra-weak luminescence signal of four varieties is studied. Analysis intime domain and in frequency domain is used to describe the features of ultra-weakluminescence signal to find sensitive variable which can distinguish different varieties.Ultra-weak luminescence signal, belonged to four classes wheat, is analyzed in time domainand frequency domain. It shows that the luminous intensity of ultra-weak luminescencesignal for four classes’ wheat is relatively weak, and has dramatic changes over time. Meanand mean square value exist obvious difference in four varieties, and the range is7.3837and74.8926.Variance has relatively small differences. The range is1.1764. Ultra-weakluminescence signal is analyzed with the method of Classic spectrum estimation. It deduced that ultra-weak luminescence signal is a low frequency signal, and its power spectrum ismostly distributed in the frequency less than0.1Hz..And three parameters, which are spectraledge frequency，spectral gravity frequency and power spectral entropy, are adopted to explainthe features of the kernels’ ultra-weak luminescence signal. It showed that the parameters ofthe ultra-weak luminescence signal for different class wheat are similar, which lay thefoundation for the variety choice in the following experiments of analysis on wheat withhidden insect. The result helps the researchers describe the biological radiation characteristicsof wheat more fully,and it lays theoretical basis for the following research of signal modelingmechanism.Meanwhile,the result lay the basis for the following ultra-weak luminescencesignal of wheat with hidden insect research of experiment for the chioce of variety.Thirdly, the ultra-weak luminescence signal of wheat with hidden insect is studied，whichis used as experimental group. For hidden insect, Sitophilus Zeamais Motshulsky is the studyobject. According to the four development stages of insect, the statistical methods and AutoRegressive (AR) modeling are employed to illustrate the change of the ultra-weakluminescence signals of wheat with insect, while the normal wheat is used as control group. Itis shown that the data’s statistic features between normal wheat and unhealthy have visibledifferences in the measurement period, the results show that the parameters change with thedays，which are distinctly different between the experimental group and control group.In the end, to examine the validity of the selected features, K-means algorithm is adoptedto design the classifier. The highest recognition is80%, which shows that it is workable touse these characteristics.This study provided the analysis method for further studying the relationship betweenultra-weak luminescence signal and biological nature. And the results also laid the datafoundation for the bio-photon emission detection model of hidden insects.