Abstract—In the present study, estimation of actual output parameters is carried out for a sponge iron production process by designing a Multilayer Perceptron model that uses a momentum learning algorithm. For this purpose data of temperature profile of rotary kiln are collected from typical sponge iron plant, which correlate four air inlets and twelve temperatures measured at different lengths of the kiln. Four different topologies are proposed for this data set to optimize the regression coefficients (R2). Firstly, these topologies are used to identify optimum value of output parameters based on value of R2. These values of output parameters meet the process requirements. Further, a better option is found to compute actual input parameters correspond to desired output. The analysis shows that to get desired output the input parameters are varied maximum by 38.9% in comparison to input parameters used in the industry.
Index Terms—Sponge iron process, rotary kiln, temperature profile, artificial neural network.
A. K. Poonia is with Dept. of Chemical Mechanical Engineering, National Institute of Technology, Raipur-492010, Chhattisgarh, India (email: ak_poonia@rediffmail.com)
S. Khanam is with Dept. of Chemical Engineering, Indian Institute of Technology, Roorkee-247667, Uttarakhand, India (e-mail: shabinahai@ gmail.com)
[PDF]
Cite:Anil K. Poonia and Shabina Khanam, "Simulation of Rotary Kiln Used in Sponge Iron Process Using ANN," International Journal of Engineering and Technology vol. 6, no. 2, pp. 95-98, 2014.
