A Nonlinear Monitoring Strategy for Blast Furnace Characteristics

EasyChair Preprint 6980

Abstract

The article delves about the development of a Process Monitoring Strategy of the Blast Furnace of an Integrated Steel Plant. Blast furnaces are employed for production of molten pig iron from sintered iron ore for subsequent production of plain or alloy carbon steel. Proper monitoring strategy of Blast Furnace assumes great importance owing to the fact that the quality of the molten pig iron thus produced has a bearing on the ultimate quality of the steel being produced by the concerned steel making plant. The Process Monitoring Strategy devised considered the nonlinear relationship existing amongst the process and feedstock characteristics associated with the Blast Furnace. The methodology being employed for development of the strategy is an amalgamation of Artificial Neural Network (ANN) and Principal Component Analysis (PCA) collectively termed as Neural Network Fitting-Principal Component Analysis (NNF-PCA) model. The ANN model was used for addressing the issue of nonlinearity by transforming the nonlinear process and feedstock characteristics observations of the Blast Furnace into its fully or partially linear counterpart and PCA was used for development of the nominal model for monitoring of the characteristics associated with the Blast Furnace. The monitoring of the Blast Furnace included detection of the fault by employment of PCA score based control chart and their subsequent diagnosis attained by the usage of appropriate fault diagnostic statistic.

Keyphrases: Artificial Neural Network, Fault Diagnostic Statistic, Principal Component Analysis, Process Monitoring Strategy, blast furnace, control chart

@booklet{EasyChair:6980,
  author    = {Ashish Kumar and Anupam Das and Swarnambuj Suman},
  title     = {A Nonlinear Monitoring Strategy for Blast Furnace Characteristics},
  howpublished = {EasyChair Preprint 6980},
  year      = {EasyChair, 2021}}