Temperature prediction model of sheep barn in winter based on Neural Network

With the rapid development of large-scale sheep farming in recent years, the issue of environmental control in sheep housing has become increasingly prominent.Due to the complex mechanism of internal thermodynamics in sheep sheds, the temperature of sheep sheds is affected by a variety of factors, especially the influence of the external ambient temperature, and shows obvious non-linear characteristics that make modelling difficult. In this paper, the collected temperature data is viewed as a time series and the neural network is used to predict the future temperature trend, which is used to control the heating equipment. The BP feedforward network and Elman dynamic network were used to predict the temperature trend in the next 6 hours using the data of the first 48 hours respectively. Compared with the BP network, the recurrent neural network Elman adds a context layer, which serves to allow the network to have a memory function and to better adapt to dynamic changes in the data input.The prediction results show that the Elman prediction model has higher accuracy and better fitting effect, and can meet the demand of sheep barn temperature control.