Daily milk yield prediction of dairy cows based on the GA-LSTM algorithm

Currently, the prediction on milk yield can help pasture managers coordinate production and transportation planning for a farm on time. However, the input of each current daily milk yield prediction research is simple and the current researches are not considering the time sequence characteristic of daily milk yield of dairy cows. In this study, we analyzed the correlation analysis on the input variables. After analysis, we selected parity, early lactation, peak lactation, mid-lactation, late lactation, weight, and the total amount of feed as input. we propose a prediction algorithm, GA-LSTM. The algorithm introduces the genetic algorithm (GA) into the long short-term memory (LSTM) algorithm structure, which considers the time sequence and correlations between the above input variables. The experiment demonstrates that the GA-LSTM algorithm is more accurate and stable than the traditional LSTM algorithm in predicting daily milk yields.