Research and application of genetic algorithm-based optimized radial basis neural network model parameter design

In this paper, the structural features of the radial basis network as well as both the center value of the hidden node and the width parameter´s influence on the structure are analyzed; the strategy of optimizing the center value and the width parameter by genetic algorithm is researched. An above-algorithm based yarn quality forecast model is established, and the result shows that the predictive output of the model basically matches with the actually measured sample, and the network trained is capable of fast and accurately predict the quality indexes.