攀钢高速重轨神经网络性能预报模型研究

Study on a neural network model to predict the mechanical properties of high speed heavy rail steel at Pan Steel

提高重轨钢的性能控制能力对其产品质量保证有重要作用。采用神经网络方法建立了重轨生产性能预报模型,并通过模型结构优化提高了模型预报的可靠性。通过模型自检、历史数据检验和离线应用,表明高速重轨的抗拉强度与伸长率预报命中率较高,可基本满足生产要求。

The control of mechanical properties of high speed heavy rail steel is very important to guarantee its quality. A mathematic model to predict the mechanical properties of heavy rail steel has been developed by means of neural network according to the demand of production. Owing to continuing optimization of the model structure the prediction reliability of the model is improved. The results of self- checking, testing by history data, and application of the model have proven that the model has higher prediction reliability to the tensile strength and elongatiori value, which can meet the production demand.