神经模糊组合预报冷连轧机轧制力

Prediction of roll force with neural and fuzzy combination for cold tandem mill

采用Elman动态递归网络方法,以生产实测数据为基础,建立了冷连轧机轧制力预报模型。在此基础上,提出了将基于误差反馈和专家经验的闭环模糊控制引入轧制力预报中,用于修正预报输出、提高预报精度和鲁棒性的设想。仿真结果表明,该方法是有效的,预报精度优于传统方法。预报结果的相对误差限制在±4%以内,实现了冷连轧机轧制力的高精度预报。

Based on the actual measured datum, the prediction model of the roll force for the cold tandem mill is established by using the Elman dynamic recursion network method. Further more, a good assumption is put forward, which brings the closed loop fuzzy control base on the error feedback and the expertise into the prediction of roll force to modify the predicted outputs and improve the predicted precision and the robustness. The simulated results indicate the method is effective, the predicted precision of which is better than the tradition method. The relative errors are less than ±4%, so the highaccuracy prediction of roll force for the cold tandem mill is realized.