Solutions

Process optimisation

Process control: eliminating deviations in oxygen steelmaking


Root Cause Analysis for Basic Oxygen Furnaces

AI Components Developed Specifically for BOF Analysis Combined With Metallurgical Expertise

We talk to you first and work with you to identify appropriate optimisation measures that match your plant, your product mix, and your equipment. We develop optimisation measures which result in permanent quality improvements and cost reductions.

Smart Steel Technologies provides a tried and tested system for analysis of BOF processes. Hot metal analysis is fused with tapping and steelmaking data including all time series and audio signals. AI models specially developed for oxygen steelmaking plants allow fast and reliable determination of complex fault causes.

Process control: eliminating deviations in electric steelmaking


Root Cause Analysis for Electric Arc Furnaces

AI Components Developed Specifically for EAF Analysis Combined With Metallurgical Expertise

We talk to you first and work with you to identify appropriate optimisation measures that match your plant, your product mix, and your equipment. We develop optimisation measures which result in permanent quality improvements and cost reductions.

Smart Steel Technologies provides a tried and tested system for analysis of EAF processes. Charge material analysis is fused with flat bath, tapping and steelmaking data including all time series and audio signals. AI models specially developed for electric steelmaking plants allow fast and reliable determination of complex fault causes.

Increase yield and quality through precise temperature control


Precise Temperature Guidance from Ladle Furnace to Caster

Predictive Models Based on Artificial Intelligence and Metallurgical Expertise

Precise temperature guidance from ladle furnace to caster avoids process deviations like ladle choking and casting sequence interruptions. It will increase yield, improve quality by controlled solidification during casting and thus save energy and reduce consumptions.

Our process experts analyse your production paths from BOF / EAF to the caster in close collaboration with your specialists. Our IT experts pull together all required data signals, tune our models to the specific situation at your plant and integrate our software components smoothly into your existing production IT systems.

Identifying and eliminating casting defect root causes


Root Cause Analysis for Continuous Casting for Flat Products

Specialised Artificial Intelligence Software Modules Combined With Metallurgical Expertise

Our metallurgy and AI experts talk to you first and work with you to identify the specific optimisation potentials of your plant. From many years of practical experience, we know the limitations of various plant types and the deficiencies in measurement and quality data. We develop optimisation measures that can be implemented without major construction projects and quickly lead to a substantial reduction in casting defects in your plant.

Smart Steel Technologies provides a tried and tested modular system of data-driven analysis components. All process and surface data are mapped to the exact strand position. Instead of the defect classes of your surface inspection system, we use deep learning based defect classifiers adapted to the specific objective of the defect cause analysis. AI models specially developed for continuous casting plants allow fast and reliable determination of complex fault causes.

Identifying and eliminating casting defect root causes


Root Cause Analysis for Continuous Casting for Long Products

Specialised Artificial Intelligence Software Modules Combined With Metallurgical Expertise

Our metallurgy and AI experts talk to you first and work with you to identify the specific optimisation potentials of your plant. Comprising many years of practical experience, we know the limitations of various plant types and the deficiencies in measurement and quality data. We develop optimisation measures that can be implemented without major construction projects and quickly lead to a substantial reduction in casting defects in your plant.

Smart Steel Technologies provides a tried and tested modular system of data-driven analysis components. All process and surface data are mapped to the exact strand position. Instead of the defect classes of your surface inspection system, we use deep learning based defect classifiers adapted to the specific objective of the defect cause analysis. AI models specially developed for continuous casting plants allow fast and reliable determination of complex fault causes.

Reducing hot mill downtimes


Prediction of Hot Rolling Interruptions

AI Components Based on Merged Cross-Process Data

Disturbances during rolling present a productivity threat to every hot mill operator. Together with you, our metallurgy and AI experts identify the specific event types relevant for your plant, for example, cobbles.

Smart Steel Technologies merges steel, casting, reheating furnace, descaling, and rolling data into AI models specifically designed for hot mill event analysis. This allows to provide online predictions of abnormalities in the rolling process with a high degree of statistical confidence.

Optimising cold mills for precision strip production


Prediction of Optimal Rolling Parameters

Adhere to even tighter tolerance limits through complementing classic automation with AI

Prior to cold rolling of each coil, an Artificial Intelligence model designed specifically for applications in cold rolling analyses quality data and process data from upstream manufacturing steps associated to the given coil and then optimises the rolling configuration computed by the mill's automation software.

Surface inspection

Making production defect images accessible using deep learning


Deep Learning Based Search Engine for Steel Defect Images

Quickly search through millions of defect images and product data using Deep Learning
  • Check where and when a particular defect type occured
  • Quickly extend ASIS training sets with defect images from production
  • Applicable to all surface inspection systems of hot rolled, pickled, cold rolled, galvanized strip
  • Boost performance of existing surface inspection systems through englarged training sets
  • Easy to use web interface. Software installed in local data center

Boosting defect classification accuracy


Deep Convolutional Neural Networks for Defect Classification

Defect classification accuracy increased by thirty percentage points

Our advanced defect classification modules can be integrated as post-classifiers with all surface inspection systems from hot rolling to galvanising. Their highly accurate classification results benefit all downstream systems such as coil grading and quality control.

  • Transfer learning via multiple pre-trained network models
  • Detailed Fine Tuning (retraining of network layers, parameter tuning).
  • Image Data Augmentation to compensate small training sets

Detecting complex defects and defect clusters


Context Based Surface Inspection

Accurate detection of complex surface defects that stretch across multiple images
  • Increased throughput (less manual inspections)
  • Reduced Risk of warranty claims
  • Self-learning system: automatic, immediate retraining after labeling saves time and increases accuracy
  • Applicable to surface inspection of hot rolled, pickled, cold rolled, galvanized strip
  • Easy to use web interface for tagging groups of defect images. Software installed on local data center

We look forward to presenting our solution to you

Smart Steel Technologies GmbH

Data protection information

Smart Steel Technologies is tackling the last efficiency frontier in steel manufacturing through Artificial Intelligence.

Copyright 2019 Smart Steel Technologies GmbH