Environment And Health Of Poultry In Neural Network Model Analysis

Chicken industry in the disease prevention and feed is success, but the problem of sub-clinical cause by the environment quality is more and more obvious in hazards and the importance in now days. By controlling the environment, we can reduce the mortality, pathogen pollution and improve the potency of vaccine. Environmental control had not been taken seriously, the situation of small-scale farms and farmer type are more prominent. There are many ways in the evaluation of environmental quality or degree of pollution. Both the pathogenic effects and disinfection of environment in poultry evaluation, there are two quantity of E. coli as a standard. Good quality of the environment is a prerequisite for healthy chicken, an important measure to improve the performance of production, an assurance improve product quality.In this study, by detecting environment of small-scale farms and farmer type about the total number of bacteria, E. coli counts and part of micro-climate factors, the application of SPSS software analysis the impact of environmental factors and then set up neural network model evaluation, mortality and sub-clinical state of chicken in the same period. First of all, use SPSS software analysis correlation effects of environmental factors in the incidence and mortality, selection the impact of significant factors, as the study parameters of neural network, then set up neural network model. EMAP (Relative error) and EMA (Absolute error) are the two points as a mixed model evaluation of the target to assess the field data in non-linear BP (Back-propagation) neural network model. After repeated testing, choose the conjugate gradient algorithm of Fletcher-Reeves algorithm for constructing the model ultimately. The total of 20 groups of environmental samples, 16 groups for model testing and self-study training, another group of 4 for testing the generalization ability of model. The model is can reveal the environment and the relevance of health status in small farms.Sub-health (sub-clinical) status is the fundamental reasons for cannot further improve the efficiency, but it has often been neglected. It is the first time application of neural network technology in the field of veterinary epidemiology. By Matlab programming language achieve the neural network computing and the learning process. Make up disadvantage of dynamic model in infectious diseases.