Abstract: In order to improve the egg production rate prediction effect of laying rate, the improved neural network algorithm is proposed. Firstly, the weight of neural network was optimized according to the error and dynamic factor of neural network, and the self-adaption learning rate was operated neural network larger in the early stage, speeding up the learning, and reduced rapidly in the late, accelerating the convergence, so that error data were reduced and the accuracy of network training was improved. Secondly, the activation function of neural network was selected the positive linear activation function ReLu, the calculation was accelerated. Thirdly, the prediction model was established. Finally, the prediction model and the prediction and evaluation indexes of egg production rate were given. The simulation results show that the correlation coefficient of the improved neural network algorithm improve 0.58%, 0.48%, 0.40%, 0.28%, 0.23% comparing with SVM, ELM, NN, PSONN, ACNN, and the mean square error reduce 5.89%, 4.92%, 3.93%, 2.67%,1.81% comparing with SVM, ELM, NN, PSONN, ACNN, so it is better than other algorithms, and provides a new idea and method for the prediction of egg production rate.