Research And Design Of Abnormal Condition Monitoring System For Blast Furnace Tuyere Injection Lance

At present,most iron and steel enterprises adopt the manual remote online observation method to realize the monitoring of the blast furnace tuyere coal injection gun coal injection status,this method requires the blast furnace operators have rich observation experience,long time observation will lead to the operator fatigue and efficiency decline and other problems,can not timely judge the blast furnace tuyere coal injection gun coal injection status.Increased the risk of accidents.In this thesis,digital image processing technology,deep learning algorithm and fuzzy control algorithm are used to study and design the abnormal state monitoring system of blast furnace tuyere coal injection gun,aiming at solving the problems of periodic identification,abnormal state alarm and abnormal state warning.The main research work of this thesis is as follows:(1)In view of the fuzzy and uncertain category information contained in the single frame image of the coal injection state of the blast furnace coal injection gun,the decision of image category label by using manual experience in the image classification task will lead to ambiguity.This thesis presents the method of fuzzy statistics to determine the category label of single frame image.It makes the determination of category label of single frame image more comprehensive,comprehensive and accurate.(2)The classification result of a single frame image as the output value of the coal injection gun will lead to the hopping phenomenon of individual frame image recognition results in the overall change trend of adjacent frame images,with one-sidedness and loss of the association information before and after adjacent frame images.In this thesis,a method based on voting strategy is proposed to identify the injection state period of blast furnace tuyere coal injection gun.The replacement of coal injection state has a relative stability,more in line with the actual site blast furnace tuyere coal injection gun coal injection state replacement.(3)Before the coal injection state of the coal injection gun is changed from normal state to abnormal state,the category probability recognition cycle statistics of normal state will decrease;Before the coal injection state of the coal injection gun is changed from abnormal state to normal state,the statistical value of category probability recognition cycle of normal state will rise.However,it is not possible to establish an accurate mathematical model to describe the amplitude of the rise and fall of the statistical value of the normal state category probability recognition cycle,and predict the change trend of the abnormal state of the coal injection gun in the subsequent recognition cycle according to the established mathematical model.This thesis presents a fuzzy inference machine designed by fuzzy control algorithm to predict the abnormal state of coal injection gun.The warning time of coal injection gun abnormal state is longer than the warning time,and the warning accuracy is high.In this thesis,LabVIEW,Python and C++ programming languages are used to design and implement the image acquisition module,coal injection state cycle identification module,coal injection abnormal state alarm module,coal injection abnormal state forecast alarm module and human-computer interaction module.After testing,the system can meet the actual production needs.