Abstract:

This paper deals with the strip shape control in hot continuous rolling and proposes a shape prediction and optimization algorithm based on a chaotic optimized support vector machine model. In the investigation, first, an improved multi-scale chaotic optimization algorithm combined with the real-coded genetic algorithm is proposed to optimize the parameters of the least-square support vector machine model. Then, the model is trained using some on-line data and is used to predict the flatness value of strip steel. Moreover, the control parameters among the in- put parameters of the model are optimized to improve the shape control. Simulated results show that, as compared with the BP neural network, the proposed algorithm is more effective in predicting the flatness and improving the accuracy of shape control in hot continuous rolling, with a flatness value optimized by 40%.

Key words: hot continuous rolling, support vector machine, flatness, chaotic optimization