基于BP神经网络的汽车车身拉延油的研制

Development of the car body's drawing oil based on the BP neural network

由于拉延油内涉及的添加剂品种多,其性能指标与添加剂间的相互关系又很复杂,所以普通数学方法很难做到在不考虑这个复杂关系的前提下寻找出最优配方组成。BP神经网络是一个优秀的并行数据处理方法,具有暗箱操作性、以及强大的学习能力和推广能力。本文使用BP神经网络计算出最佳添加剂加量,并研制出具有很好极压性、清洗性、防锈性和安全性的拉延油。

The drawing oil contains many kinds of additives, and oil's performance depends on the complex interaction among additives. The optimum formulation of the drawing oil cannot be worked out by using the common mathematics methods that must take the interaction into consideration. The BP neural network was proved to be an excellent parallel data-handling method with dark-box operating performance, powerful studying and generalizing ability. Therefore, the BP method was adopted in this paper. On the basic of the data, the drawing oil with good load carrying, clean-out, anti-rust ability, and secure performance is finally developed.