Conventional and AI models for operational guidance and control of sponge iron rotary kilns at TATA sponge

Abstract

Prediction models for temperature, pressure, and quality control in rotary sponge iron kilns are developed from operational data. The conventional and AI-based methods which are used to develop the models include extreme learning machine (ELM), artificial neural net (ANN), and multiple linear regression (MLR). The performance of the developed models is tested on shop floor in actual operation and compared. Extensive plant data is used to develop and validate the models on day-to-day basis of operation so as to take care of the dynamically changing situation inside the kiln, giving first preference to quality control and then to accretion control. Accretion control increases the life of lining and thus also the available time for production. Automatic pressure control greatly helps in chaos control inside the kiln. Dynamically changing Lyapunov exponent acts a guide line for automatic pressure control. � Springer Nature Singapore Pte Ltd. 2019.