Research on the Prediction Algorithm of Daily Feeding Amount for Pregnant Sows Based on PSO-RF
The daily feeding amount has a great impact on the reproductive performance of pregnant sows and is of great significance for ensuring the health of sows and the growth and development of fetuses.In view of the unbalanced search ability of Particle Swarm Optimization(PSO)algorithm,a nonlinear decreasing inertial weight strategy was introduced to improve the PSO algorithm,and the improved particle swarm optimization random forest regression algorithm(PSO-RF)was used to accurately predict the daily feeding amount of pregnant sows,and ac-curately control the feed delivery of intelligent feeding device.The algorithm combines the high accuracy of the Random Forest algorithm with the strong parameter finding ability of the Particle Swarm Optimization(PSO)algo-rithm to enhance the prediction performance by optimizing the number and maximum depth of decision trees.The results demonstrate that the PSO-RF algorithm attains a coefficient of determination R2 value of 0.981 4,representing an enhancement of 1.19%,2.30%,and 3.25%in comparison to the RF algorithm,SVM(Support Vector Machine),and BP(Neural Network),respectively.The PSO-RF algorithm demonstrates superior accuracy in predicting the daily feeding amount of pregnant sows,which can facilitate enhancements in the intelligence of pig farm management,reduce production costs and improve the farm's breeding efficiency.Consequently,it possesses definite practical application value.
万方数据
维普
