The Research Of Online Evaluation System For Sintering Quality Grade Based On Tail Section Image And Bellows Temperature

As one of the most important raw materials of economic construction,Iron and steel prompte the rapid development of metallurgical industry.Iron ore is one of the most important raw materials in blast furnace ironmaking,but natural and rich iron ore has can’t meet the production requirements of raw materials in blast furnace. Abundant in the nature of the lean ore begin to get a large number of mining applications,however,it can’t be directly used to blast furnace ironmaking,it needs sintering to form the sinter in order to improve the quality.Therefore,at the present stage sintering process is essential technology of metallurgical plants.Due to the sintering process is extremely complex,so people need real-time observe and monitor the quality of sintering,in order to adjust the process parameters timely.The chemical detection method is more accurate, but the sample testing in the lab time for several hours,it can only reflect the sintering quality of quite a long time ago,So it is poor real-time performance. The method of artificial observation from tail section has richer real-time performance,it judge the quality of sintering by the fire-watching workers according to production experience for many years by observing the tail section,however,this method also has the unavoidable defects, it could be mainly influenced by subjective factors,such as visual error, experience is not unified and so on,thus to fluctuate more easily on the judgment; In addition, this method is difficult to achieve long-term real-time monitoring,it is lack of continuity on the Judgement of slight burning and excessive burning.But as a result of this method is highly intuitive,it is still widely used in the industry.At present,The mainstream approach to judge the sintering is slight burning or excessive burning is based on the temperature rise of the bellows,however,it can only forecast the BTP(Burning Through Point),and it does not directly reflect the quality of sintering at finally,and it is not intuitive.Therefore,the organic combination of the above two methods can overcome their own shortcomings,it can not only ensure the correctness,but also be Highly intuitive and convenient.This method forms sintering indexes by the morphological analysis of tail section and the monitoring data of bellows temperature,and forms a corresponding functional relationship with sintering quality grade.Finally,it can get the output of sintering quality according to the input of real-time sintering indexes.Sintering is a highly nonlinear and high coupling process,using simple mathematical relationships can not express this function.In recent years,neural networks get more extensive application in the field of function approximation,and the traditional neural network is difficult to find the optimal solution in the global,it may effect the precision of judgement. For this reason,when using neural network to determine sintering quality,Requires the optimization of neural network algorithm is necessary,using the SA(simulated annealing) algorithm can obtain the optimal solution in the global,it improves the precision and robust of neural network.In combination with the morphological analysis of the tail section and the tracking of the bellows temperature,and relying on the improved neural network as a judgement algorithm,to realize online evaluation of the sintering quality levels. After a long time of the actual running result at the sintering production line of the division of stainless steel in baosteel.This method is not only real-time, highly effective, intuitive and convenient,but also significantly improves the accuracy of the evaluation,it has the high application value.