Prediction Model of Hot Rolled Strip Quality Based on K-Means Clustering and Neural Network

Effective quality prediction for hot rolled strip can effectively improve product quality. The traditional prediction model does not consider the characteristics of hot strip production data, this paper uses system clustering to determine the number of clusters, and then uses K-means algorithm to divide production data into K clusters. Each cluster of data is used to establish its own BP neural network prediction model, and particle swarm optimization algorithm is used to optimize the network parameters to avoid neural network falling into local optimum. The method is used to predict the finishing rolling temperature of hot strip. The simulation results are compared with the prediction model established by non-clustering analysis. It shows that the average error of the neural network prediction model based on clustering analysis is smaller and the prediction accuracy is higher.