Blast Furnace Fault Diagnosis Method Based On Improved ICA

As an important basic raw material,steel is extremely important in our daily life,which can be found everywhere.Reports show that China has now become the world’s largest steel producer,as well as the biggest steel consumer.This industry has developed rapidly,and has now become the important pillar industry in our country.However,it must be noted that the iron and steel industry in China still has a ton of problem to be reckoned with,such as high energy consumption;scattered industry;low production quality and bad supply chain management.So,we must strive to improve the quality of the product and reduce energy consumption.The blast furnace is an important equipment in the steel industry with a large energy consumption and has great influence on the subsequent smelting process.Thus,in order to reduce the energy consumption and improve the quality of the products,we decided to study the fault detection and diagnosis method of blast furnace in order to ensure smooth process.The blast furnace itself is a very huge and complicatced system with complex mechanism model.To ensure safe and smooth melting process,we have to detect and diagnose the fault happened in the blast furnace accurately so that we are able to take accurate actions.Taken the large amount of non-Gaussian data collected during the actual process of blast furnace smelting,we decided to use the independent component analysis method to detect and diagnose the blast furnace fault.However,ICA algorithm itself has some disadvantages.So,first of all,we have to improve the ICA algorithm by considering the number of independent component chosen and the importance of each component.Thus,we are able to detect the fault much more accurately.Secondly.we decided to combine the ICA algorithm with MEWMA method based on the slow changing characteristics of the blast furnace fault.By taking the influence of historical data into account,we are able to detect the fault much more accurately.Finally,since the fault samples of the blast furnace smelting process is limited due to safe producing,we decided to use the SVM algorithm to diagnosis the fault of the blast furnace once the improved ICA has already detected fault.