2019 Pittsburgh AISTech

Sessions Speakers Exhibitors Products Press Releases Show Specials Floor Plan Buzz

Development of a Predictive Model for Minimizing Ladle Desulfurization Cycle Time and the Associated Costs (Room 411)

Steel plant operators rely mostly on heuristics (i.e., past experience and conducting sample tests after the batch process) to tune the sulfur content to the desired value. However, the heuristics approach is often unable to reproduce (a) desired steel sulfur levels and (b) process cycle times due to individualistic approach by shop floor personnel. In this study, the support vector machine (SVM) algorithm, accompanied by support vector regression (SVR) and artificial neural networks (ANN), was utilized to develop a predictive model capable of suggesting optimum addition patterns, material input mix and process parameters, thereby minimizing process cycle time and the associated costs.