Data Processing In Animal Barn Based On Linear Regression And GRNN

Intelligentized measure and control technology in animal breeding field is a very important direction for developmental stockbreeding in 21st Century. It will promote the change from extensive and low effective stockbreeding to knowledge, technology_ based and aggiornamento, highly effective in pasture husbandry. Compared with the developed countries, animal industry is undeveloped in our country. To increase the efficiency and output of animal production, in recent years, animal industry in our country is growing as scale and factory. Therefor the research for the letting,detecting and controlling of nocuous gases in the animal bar is already arose widely attention in the national agricultural and environmental section and in the academe.This paper aims at addressing the current pollution of environment in animal barn,proposes the methods to recovery estimate the missing data of the nocuous gases concentration. For the missing data duing to the sensor faults, considering the temporal,spacial and environmental influences to the measure value of one nocuous gas,using multi-sensors to monitor the concentration of the nocuous gases, environmental temperature, humidity and etc, puts forward the methods of multiple linear stepwise regression and the data fusion based on GRNN to recover the missing data. In the choice of variables, using PLOT to analyse a good many genes influencing the concentration of nocuous gases, and finding out the remarkable influenced genes. On this basis, study the data estimation methods of multiple linear stepwise regression and the data fusion based on GRNN. And compiles the corresponding software, achieving the recovery estimation of the missing data. The methods researched not only solves the practical problem, but also analyzes the scientific and feasibility of the methods in the theory. In this paper, we take the data processing of the concentration of NH3 for example, gathering the annals of the concentration of NH3 three days in the animal bar in America to analysis and dispose. As to actual research object, the maximal comparative error of estimate to step regression of mathematical statistics is 9.37%,and to the data fusion based on GRNN is 9.80%.These two methods can easily actualize and make the data recover adequate precision. This research provides a reliable basis for measuring the gross of nocuous gases which let continuously in the bar accurately, then control the nocuous gases beyond normal.