An Ammonia Concentration Prediction System For Pig Houses Based On The Android Platform

With the improvement of the quality of life,people's demands for pork and its products have also gradually increased.The pig breeding of large-scale and intensivity has rapidly developed.In large-scale pig feeding production,pig house environment plays a vital role in the growth and development of pigs.The good living environment can effectively improve the health level,reproductive capacity and pork output of live pig.Therefore,great importance has been attached to the pig breeding environment in many aspects.Ammonia is the main component of air pollution in large-scale closed pig house breeding,which has the greatest influence on pigs.It not only reduces the health and production performance of livestock and poultry,but also easily causes various diseases,and has become the main air pollution source in livestock farms.There are some common indexes that were used to measure the air quality of livestock and poultry farms in the world.Due to the nonlinear and time-varying characteristics existing in the environment of pig house,it is difficult to predict and control the environment of pig house,especially to realize the intelligent and accurate prediction and control of these environmental factors.The main method of predicting ammonia concentration in closed pig farms is to establish a model for the prediction of ammonia concentration,which not only is one of the important indexes of air quality evaluation in closed pig farms,but also the basis of accurate prediction and management of pig growth environment.Since the concentration of ammonia in pig house has a great influence on the growth and development of pigs,it is necessary to establish an accurate model for predicting the concentration of ammonia.Although there are some researches on the prediction of ammonia concentration in pig house,the concentration of ammonia is affected by many environmental factors within the pig house.In addition,the relationship between ammonia concentration and environment of pig house is nonlinear,and there is a lack of accurate prediction model.Therefore,this study selects 2 880 groups of data for 120 days for 4 months basing on the measured environmental data of pig house(including the ammonia concentration,temperature,humidity,pig's activity,ventilation,fan opening,ventilation rate),and utilizes BP neural network,Linear neural network and Elman neural network basing on the L-M algorithm optimization to predict the ammonia concentration in pig house.The results show that the9-19-19-1 four-layer structure prediction model established by Elman neural network has achieved the target error after 7,047 steps,and the maximum absolute error between predicted value and true value is 0.9466.The 9-19-19-1 four-layer structure prediction model established by BP neural network based on L-M algorithm optimization has reached the target error through262 steps and the maximum absolute error between predicted value and true value is 1.1647.Although the training speed of BP neural network prediction model based on L-M algorithm optimization is fast,the prediction model of BP neural network based on L-M algorithm optimization is unstable along with the increase number of training data,and the prediction error for a certain time is large.However,the Elman neural network has better associative memory function and stable prediction due to it has the internal local memory unit.Compared with the Linear neural network and BP neural network based on L-M algorithm optimization,it can improve the accuracy and timeliness of prediction to the ammonia concentration and provide support for the environmental pre-warning of pig house.It also provides a reference for the establishment of prediction models in the other industries.Through the ammonia concentration prediction model established,the prediction system of ammonia concentration in pig house based on the Android platform was established.The ammonia concentration prediction and pre-warning system in the large-scale closed pig house breeding environment can effectively predict the ammonia concentration in the course of breeding live pigs in real time,and automatically notify the relevant personnel when monitoring or predicting high concentration of ammonia,which can effectively reduce the risks of live pig breeding.