Study on Growth Modeling for Pearl Gentian Grouper Based on RBF Neural Network.

In this study, under the environmental conditions of factory circulating aquaculture, by measuring various parameters affecting the growth and development of pearl gentian grouper, a Radial Basis Function (RBF) neural network model for its growth was established, and the fit of the model was analyzed which provides a reference for the breeding of pearl gentian grouper. Aiming at the problems of RBF neural network, the Principal Component Analysis (PCA) method was used to screen out the main growth factors as the input of RBF, and the iterative fitting effect of RBF neural network was improved. The model in this paper decreases rapidly after the start of training, about 30 iterations, and the error decline rate gradually slows down. After 281 iterations, the error starts to converge, the value tends to stabilize, and the stable value error is about 0.05. The relative error of the prediction result of the trained model on the test set gradually decreases as the true value becomes larger. When the true value is larger and stable, the relative error is smaller and the overall fitting effect is better. The growth model of pearl gentian grouper was established by using RBF neural network and traditional mathematical method stepwise regression analysis, and the established growth model was used to predict the weight of pearl gentian grouper. By comparing the results, we can find that the RBF neural network prediction algorithm is closer to the real result in the prediction result, with a minimum error of 0.24209.