Research On Artificial Intelligence Method Of Sintering Quality Based On Sectional Image Of Sintering Machine Discharge End

Sinter is the main raw material of iron-making. In order to operate blast furnace and improve the qualified rate of production efficiently, we hope to produce sinter with the best properties. In great majority of sintering plants of our country, the control of sintering process is still dependent on the experience of the sintering experts. That's to say the sintering operation is mainly dependent on the intuitive observation of the fire-workers about the cross-sectional image of the sintering machine tail, and the sintering quality is judged with experience. However, the environment of the sintering machine is bad, and there is experience difference among the fire-workers, which have a great impact on judgment of the sinter quality.The cross-section of sintering machine tail was set as the research object in this paper, through a large number of survey of this field, the fire-workers' experience and knowledge were converted into inference rules associated with the image features of sintering machine tail. Intelligent methods were used to make more accurate judgment of the sinter quality. The method has reduced the sintering fire-workers' labor intensity in some extent. The sintering quality information and the images were displayed on the screen of the computer, providing the information of guidance operation for sintering process in real time.In order to make the intelligence detection method of sintering quality meet with the experience of the sintering fire-workers, a lot of work has been done as follows.First, through a large number of survey and study of the field, the fire-workers' experience and knowledge were converted into inference rules associated with the image features of sintering machine tail. Multiple cycle of sintering machine tail sectional images were analysised, as the lack of external trigger and the fixed interval method of cross-sectional images, a novel method which is by comparing brightness between frames of sintering machine tail video was proposed in this paper, and this method is more accurate to find the key frames of sintering machine tail.Secondly, based on experimental analysis, regarding the image features of the cross-sectional sintering machine tail, the images preprocessing methods were studied for accurately extracting images feature lay a good foundation. The features of the sectional sintering machine tail images were extracted, such as area, perimeter, and gravity of the booming area. Characteristics which are related with sintering FeO content were extracted too. That's the average luminance and the average area of the pore. In order to judge the quality of sintering, the feature parameters with the inference rules were combined.Finally, the current devices were analysised, and the improved system devices selection program were put forward. On the software side, the program used a modular design and Matlab programming language. Each module of the program conducted a detailed description and explanation. At the same time, the author used LabVIEW to make an image display interface, which can display the images' characteristics of sintering machine tail and give guidance functions of sintering quality information.