Storage Pests Sound Pattern Recognition

The storedproducted insects severely endanger the stored materials. Their infringement target involves the human activities of different goods and materials. The storehouse insect causes enormous loss to national economy lift. Only in one item of grain, loss who whole world cause by worm endangers the storehouse every year reach 5% to 10% of total output.It is realistic significance to strengthen study on thing pest of storage.It can improve the preventive remedial, protect the materical, reduce the loss. The development which accelerates national economy is significant. For many years,it is an important subject of research field to expertize and classify the storehouse worm.It offers important scientific references to know the living beings characteristics of different types of pests. find out secure, economic and effective methods to kill off and measure the pest, and predict when the pest will take place and their quantities. At present, thing qualification of pests for storage have mainly two kind methods .One is based the extraneous form; the other utilize modem biochemistry and biophysics experiment methcd to classify the pest according to storedproducted insects inside physiology characteristic of pest. But those who adapt above-mentioned method to go on and classify kinds of the insect must possess certain professional technique knowledge. And they cant operate on-the-spot and must work long time .The general staff member of storehouse management are very difficult to utilize above-mentioned method to distinguish the kind of pest.So it is one of the important issues to automotive and classify the storage insect which is hoped to solve storedproducted insects on recognition technology computer in the research field. It has not been reported all over the world that documents about the problem of computer automatically identifying and classifying storedproducted insects by voice. A new method of insect's computer automatic classification based on voice-pattern recognition technology is first presented in this paper. First this paper briefly introduces the significance and status of storedproducted insects recognition and summarily discusses the general voice pattern recognition method in the introduction of chapter one. Second. in chapter two it does some processing to insect抯 voice signals such as noise reduction by Madline neural network, making an uproar, number value normalization and utilizing FFT(fast fourier transforn) as frequency spectrum to algorithm analyze. The feature of insects voice of the storedproducted insect finally is vectorial; Third, in chapter three these feature of insect抯 voice is sent into BP neural network to discern. With the energetically support of the Prevent and cure pest research group in the institute of technology of Zhengzhou , I has obtained domestic three kind common storedproducted insects . Each storedproducted insects uses 6 as BP neural network study specimen. It uses above-mentioned method to discern 30 storedproducted insects and the discerning rate is up to 81%. To the result of data analysis of preliminary experiment, I feel relatively satisfied. This research of subject have wide application prospect, and turn on way of innovation for storedproducted insects qualification and sort research.