基于模糊聚类的BP神经网络模型预报中厚板轧制力

以实测数据为基础,在中厚板轧制设定中采用BP神经网络的方法取代传统的轧制力数学模型,并对神经网络输入项和训练样本进行分析,将传统轧制力模型的自学习过程引入神经元网络用于轧制力预报,改善预报精度.采用模糊聚类分析方法,科学选取学习样本,解决了由于样本多学习速度慢的问题.通过在线数据分析,可知这种方法对轧制力的预报精度有很大改善,而且神经元网络的结构也得到简化.此方法可以作为神经元网络应用的一个拓展.

Prediction of plate rolling load of BP neural network model based on fuzzy clustering

On the basis of the measured data of the plate mill, the neural networks have been used for prediction of rolling load instead of traditional models for presetting in plate rolling. The input of networks and training data have been analyzed and a self-adaption of rolling load model was integrated with BP neural networks for prediction of rolling load in order to improve the prediction precision of traditional rolling load model. The fuzzy cluster analysis was used as the preprocessing to select the sample set, which can solve the problem of study speed. It is shown that the prediction precision of BP neural networks can be improved greatly according to many on-line data and the structure of the BP neural networks is also simplified by this method. This process can be a necessary supplement to the application of BP neural network.