Business Problem: One of the world’s largest steel producers wanted to improve the steel plant’s efficiency using data science. For them. the steel being produced must meet a set of composition tolerances (% of constituents). The metallurgical techniques currently used are very accurate but are time-consuming. The client wanted to predict the composition of the output in near real-time using operating parameters as an input to the model.
If we can learn composition as a function of operational parameters with sufficient accuracy, the client’s long term goal is to optimize/adjust the operating parameters inside the blast furnace to control the composition of the steel.
Our Approach: We built a predictive model to correlate the composition of the steel (Carbon, Silicon content) and a large number of operating parameters, chemistry and environment. The attributes are a mix of time series (operational) and causal (raw chemistry).
Our model building process included understanding the chemistry of steel production, nature of relationships and selecting the model structure to suit the underlying dependencies and training the models.
We evaluated gradient boosting machines, neural nets and hybrid models.
Results: For 85% of the cases, the prediction was within the RMSE desired by the client.
INSOFE provides a variety of consulting solutions to clients. Ranging from custom education from internal teams to complete delivery ownership, we can design a solution, delivery methodology that works for you. Click here to learn how our 75+ ML consultants with 75 patents and 100 consulting projects help you solve your pressing business problems.