神经网络在酸轧自适应系统中的应用研究

冷连轧中的参数预计算是一个非常复杂的模型系统,根据轧制的带钢数据,为轧制过程提供各种参数设定值。通过研究现场原有系统的计算模型和算法,分离出了用于各种模型参数的系数计算的神经网络结构和功能,从理论上研究了改进神经网络训练样本数据库和结构的方案。离线计算实验结果表明,酸轧参数自适应计算中的神经网络设计和应用,有效地改进了传统轧制模型的计算精度,对于提高带材的板形质量和性能具有重要作用。

Application of Neural Network in Self-adjusting System in Cold Rolling

The pre-calculation of rolling parameters in cold rolling is a very complex mathematic model system.According to the steel band data,the system can work out all rolling parameters.By researching the calculation model and algorithm of the original model system,the structure and the function of all the coefficients of the model parameter computed by neural network are abstracted.The scheme to improve the training sample database and structure of the neural netwrk are studied in theoretic way.The experiment results of the practical production show that the the computing precision with neural network in self-adjusting algorithm are improved on the classical rolling model,which plays an important role to improve the quality and performance of the steel band.