基于自组织神经网络的烧结终点自适应预报系统的开发

烧结终点的在线检测和提前预报对于稳定终点，进而提高烧结矿强度和产量、降低能耗有重要意义。文章介绍了烧结终点的软测量方法；提出了一个新的预报参数——风箱废气温度曲线拐点；将多层前向人工神经网络应用于烧结终点的预报，对BP算法做了较大改进，使学习算法可以自组织神经网络的结构。应用这些技术开发的烧结终点自适应预报系统能够快速、准确地判断和预报烧结终点的状态，可用于在线操作指导或作为自动控制的依据。

Adaptive Prediction Systemof Sintering Through Point Based on Self-organizing Artificial Neural Network

The on-line measurement and prediction of burning through point(BTP)in sintering process is significant to optimize operation which aims at improving the strength and yield,decreasing the energy-consumption. The present paper describes a soft-measurement method of BTP and proposes a new predictive parameter,the mathematics inflexion point of waste gas temperature curve in the middle of the strand. The artificial neural network is firstly used in predicting BTP,the author makes great modification on backpropagation algorithm in order to improve the convergence and self-organizing the hidden-layer neurons. The adaptive prediction system developed on these techniques shows its characters such as fast,accuracy,less dependence on production data and good robustness. The prediction of BTP can be used as operation guidance or control parameter.