Research And Analysis Of Electronic Nose Data Based On Data Mining Method

Electronic nose technology based on sensor technology,signal processing technology and patternrecognition technology is one of the fastest growing gas phase analysis and gas detection technologies in the past two decades.It has gradually developed in biomedicine,environmental monitoring,agricultural production,food.Detection and other fields have been applied.The electronic nose is a sensor array having cross sensitivity to the gas to be measured,and the information of the mixed odor component in the gas to be tested is converted into a measurable physical signal group related to time,composition,concentration or content.The signal acquisition system is used to output the digital signal containing the characteristic information of the gas to be tested,and the digital signal is analyzed by the pattern recognition system to obtain the integrated odor information and hidden features of the gas to be tested,so as to realize the rapid,systematic and accurate identification and analysis of the gas to be tested.Some classical algorithms are not ideal for classifiers when using the electronic nose(E-nose)for classes.We need to develop a new algorithm to solve this problem.Use less data and get a higher classification accuracy for the electronic nose.In this paper,K-means clustering is used to extract the features of the data,then the adaboost-SVM model is established,and the electronic nose data of the pig farm is used for pattern recognition,and different time periods are classified to preventthe harmful effects of pig farms in different periods.Gas concentration is enriched.At the sametime,we consider more common classification models to compare with them,so that adaboot-SVM hasobvious advantages in the classification of pig farm electronic nose data.