Digital Transformation of Coke Oven Batteries: A Path to Efficiency

Digitalization improves the production, quality, health & safety. This paper is about the digital models developed and integrated into Coke Oven Scheduling & Heating Monitoring System. Explored first principal methods, conditional based algorithms, various machine learning and artificial intelligence techniques, namely XGBoost, Random Forest, Decision tree & GLM to build predictive, prescriptive & descriptive models. Descriptive analysis & exploratory data analysis have been done to get insights from the data. Machine learning models were tuned for their optimum hyperparameters using grid search algorithms iteratively. Models with high accuracy have been deployed. Good operating zone/ranges have been defined for controllable parameters using Partial dependency plots.