Dongkuk Steel

AI Transformation
July - Dec, 2020
#Energy #Digital Transformation #AI Modeling

Dongkuk Steel Mill Co., Ltd.

Dongkuk Steel Co., Ltd. is a steel manufacturing company that operates the primary steel manufacturing industry
in South Korea.

Project Details and Achievements

1. Dangjin factory’s heating furnace charging timing automation.

Dongkuk Steel Co., Ltd.'s Dangjin plant predicts the rolling time of each blade in the furnace and tries to build a control automation system by predicting the time the slab is put into the furnace.

SmartMind have developed a model with superior performance compared to the existing model by classifying the type of electric charge rolling (general rolling, double rolling start plate, double rolling 2 ~ thick plate) and building data pre-processing and model.

As a result, SmartMind were able to develop a model using an average RMSE value of 6.65 seconds for 80 consecutive slabs with an RMSE of 25.12 seconds in the total data. The analysis process and results were presented with insight through the power bi dashboard.

2. Forecast of shipment of machined reinforcing bars

Dongkuk Steel Co., Ltd. wanted to develop a model that could predict monthly processed rebar shipments during the construction period according to the characteristics of the construction site and to conduct trend analysis.

There was a problem in establishing a learning model on a site-by-site basis as contract weights, shipments, and trends were all different for each site and processing plant. SmartMind improved the model by calculating the relative delivery rate for each site and training on each cluster.

• As a result of model prediction, a model with an average accuracy of at least 90% was developed for both reinforcing bars and sections. We also provided an expandable model and a power-by dashboard, allowing Dongkuk Steel to directly customize the parameters to be included in the model.

3. Developing a demand prediction model for long-distance steel

Dongkuk Steel Co., Ltd. wanted to develop a monthly demand forecasting model for Long Long products through statistical indicators provided by public institutions and data within Dongkuk Steel Plant.

In this project, since it is a task to predict monthly shipments by learning monthly data of existing reinforcing bars and sections, there was a data shortage problem. Smart Mind augmented data through the window slide technique and developed a model through it

Samchully is now able to monitor phone call theme distribution and trends. In addition, phone call themes and contents can be appropriately categorized based on date, extension number, incoming/outgoing, etc. Moreover, phone call logs can be filtered by specific themes or keywords for review. This allows easy statistics extraction and call detail monitoring. Automatic scheduling was set up to update the monitoring result daily.