Fault Diagnosis of Blast Furnace Throat Temperature Monitoring Device Based on Residual Analysis

Reasonable control of the distribution and development of gas flow is one of the most important operations for blast furnace (BF) ironmaking process. The BF throat temperature directly reflects internal coal gas flow distribution. Therefore, BF throat temperature monitoring is the key prerequisite for BF abnormal monitoring. In this paper, a novel fault diagnosis method for BF cross temperature measurement device is proposed, based on a residual analysis method combining principal component analysis (PCA) and furnace throat temperature estimation via multi-burden layer reconstruction. Firstly, the material surface profile of each BF burden layer is calculated according to the burden material distributing movement process, the material layer distribution model is established in combination with particle descending process, and the multi-burden layer reconstruction is implemented. Then, using the multilayer ore coke ratio (OCR) distribution concluded from the reconstructed multi-burden layers and the main BF state parameters, the long short-term memory (LSTM) network is adopted to estimate the furnace throat temperature. The residual difference between the estimated result and the actual temperature value is calculated. Finally, residual difference and BF main parameters are used as inputs of PCA to obtain fault diagnosis results, and the majority voting (MV) method is introduced to improve the stability and accuracy of the results. A case study is conducted using data from a BF located in the south of China to verify the effectiveness of the method. The fault diagnosis accuracy is above 90% when the offset is greater than 9 degrees C or the drift rate is greater than 0.4%