It is the first time that a converting furnace endpoint prediction model based on an improved BP neural network and error compensation of linear regression. By simulating test, it is proved that the model possesses higher precision and practicability.

By the full use of the ergodic property of chaos movement, a Chaos Genetic Algorithm(CGA) is proposed in this paper. The basic principle of CGA is that the small disturbance is added to child generation group by using the chaos variable and the disturbance extent is adjusted little by little as the search is going on. The computation results indicate that the CGA has good performance and significantly improves the computational efficiency in optimization.

本文通过理论计算,建立了铜锍吹炼造铜期出炉烟气温度的数学模型,提出一种对造铜期终点进行预报的简便新方法,并将之应用到工业生产上,烟气温度信号预报值与生产经验相比,偏差约为2～3分钟,显然要比经验法更科学,而且具有成本低、操作简单的优点,值得其它有条件的铜冶炼厂推广应用.