高炉料面的分类与案例匹配算法

本文基于改进的k-means算法及分级案例库匹配技术提出一种研究高炉料面和煤气流关系的方法.为了获取煤气流分布情况,首先提出了改进的基于新型有效性指标评价的k-means算法,并将该算法与其它多种算法进行比较,证明了该方法的高效性和准确性.继而在此基础上提出了案例库匹配技术,从而获得与当前料面最为匹配的历史料面.最后,将匹配算法与改进的灰色相似性匹配算法和欧式近邻匹配算法进行比较.结果表明,分级匹配算法具有更高的分辨率和效率.在多次试验中匹配准确率高达92.5%,比其他几种算法更加准确,更适合研究料面与煤气流关系,指导布料操作.

Classification and case matching algorithm on blast furnace burden surface

A method is supposed in this paper based on the improved k-means algorithm and graded case based matching method to study the relationship between the burden surface and the gas flow in blast furnace. To get the gas flow distribution from historical data, first of all, an improved k-means algorithm was proposed based on a new evaluation approach of effectiveness index. Comparison with other clustering algorithms proved that the proposed algorithm has a high accuracy and high efficiency. A matching technique was put forward on the basis of the above clustering results to obtain the most matched historical burden surface. At last, compared with the improved grey similarity matching algorithm and Euclidean nearest neighbor matching algorithm, the results showed that the proposed method has higher resolution and efficiency. Matching accuracy is as high as 92.5% in the experiments which is more accurate than the other methods. The approach is more suitable for the investigation of the relationship between burden surface and gas flow so as to assist monitors of blast furnace to control burden surface.