Monitoring And Fault Diagnosis In Continuous Annealing Process Base On Mechanism And Data Analytics

Continuous annealing is an important procedure in the production process of strip steels.In the continuous annealing process of a domestic large-scale iron and steel enterprise,the strip tension in the soaking furnace cannot be measured in an on-line way and the fluctuations of strip tension are quite frequent.So the strip running deviation often occurs in pratical production.If the strip running deviations are not timely warned and effectively handled,some major production accidents such as broken down may be caused,which will result in great loss of economic benifits for enterprises.Taking the actual strip operation inside the soaking furnace in a large iron and steel enterprise as the background,this paper proposed a hybrid modeling method of mechanism and data analytics to realize the real-time measurement of strip tension,as well as the monitoring and fault diagnosis of strip running deviation,in the soaking furnace,which was very important to guarantee the normal and continuous production of the continuous annealing.The detailed research contents included the following parts.(1)Due to the fact that the strip tensions cannot be measured in an on-line way in the soaking furnace,a mechanism model of the strip tension at each roller was established based on the mechanism of tension generation.The test results based on the data of the actual production process showed that the model could realize the real-time measurement of the strip tension in the furnace.However,the accuracy of the model remained to be improved.(2)To further improve the accuracy of the strip tension model,a hybrid modeling method of mechanism and data analytics was proposed.The hybrid method adopted the least square support vector machine to correct the errors of the mechanism model,so that the accuracy can be improved.The test results based on the data of the actual production process showed that the hybrid model was significantly better than the mechanism model and could obtain a higher accuracy that met the actual production requirements.(3)To handle the strip running deviation and the unacquirable fault data,a monitoring and fault diagnosis method based on the combination of mechanism and principle component analysis was put forward so as to reduce the false alarm rate and improve the accuracy of fault diagnosis.The test results based on the actual production process data showed that the hybrid method was quite effective.(4)Based on the above methods,the monitoring and fault diagnosis system for strip running deviation in the soaking furnace of continuous annealing was developed.The practical application of a large domestic iron and steel enterprises showed that this system could satisfy the needs of the real-time measurement of strip tension,and the monitoring and fault diagnosis of strip running deviation.