人工神经网络在配煤过程状态建模中的应用研究

INVESTIGATION OF ARTIFICIAL NEURAL NETWORKS IN COAL BLENDING

本文详细介绍了人工神经网络应用于状态建模的方法．对神经网络应用中的一些难点提出了切实可行且有效的解决措施，并举例作了应用示范．同时还介绍了神经网络方法应用于优化动力配煤的情况，并就神经网络方法在优化动力配煤中的进一步应用作了展望．

A complex engineering system is usually characterized by a number of interacting factors in which the relationship between these factors is not precisely known. So establishing an empirical model to predict these systems is a very troublesome task. This paper demonstrates the use of back-propagation (BP) neural network to alleviate this problem. How to model a complex system with BP network is introduced in detail, including the solutions ofsome key and knotty questions such as initialization of BP network, determination of the architecture of BP network and choice of training precision and so on. Based on our experience, some new and useful advice has been put forward. As a demonstration, applications of BP neural network in predicting coal ash fusion temperature and in optimizing coal blending are presented. The results indicate that our advice is helpful for modeling a complex system satisfactorily. At last, the prospect of application of artificial neural networks in coal blending are given.