基于人工神经网络预报汽车齿轮钢的端淬值

Prediction of Top Quench Value for Automobile Gear Steel Based on Artifical Neural Network

采用人工神经网络建立了关于国内某特殊钢厂生产的汽车齿轮钢的淬透性与化学成分之间关系的非线性网络模型 ,并验证了该模型的准确性。该方法不同于以线性回归为基础推导出的经验公式 ,它具有容错性好、通用性强等优点 ,可用于预报要求多点控制淬透性的汽车齿轮钢。

A non linear neural network model used to study the relationship between hardenability and chemical composition of automobile gear steel produced by a domestic special steel factory has been established. The model was proved by examination. This method is different from the traditional empirical formula based on linear regression. It has the advantages of good fault tolerance and generality. The neural network provides a new and reliable way for hardenability forecasting of automobile gear steel.