Inspiration
An AI-powered smart manufacturing platform that optimizes alloy composition using machine learning and real-time spectrometer integration, reducing material waste, energy consumption, production costs, and carbon emissions while accelerating industrial innovation. Designing and testing new alloys is a slow, expensive, and resource-intensive process. Engineers often have to manufacture trial samples and perform multiple laboratory tests before finding the right composition. We wanted to reduce this trial-and-error process by using Artificial Intelligence to predict alloy properties and recommend optimized compositions before physical production. Our goal was to make alloy development faster, smarter, and more cost-effective.
What it does
AI-Based Alloy Optimization predicts important material properties such as yield strength, tensile strength, hardness, elongation, and melting temperature from an alloy's chemical composition. The system then recommends optimized alloy compositions based on user-defined targets. It also integrates with a simulated spectrometer using OPC UA, allowing real-time composition data to be automatically imported into the application without manual entry.
How we built it
Collected and prepared alloy composition and property datasets.
Trained Machine Learning models using Python and Scikit-learn for property prediction.
Built an optimization engine to suggest improved alloy compositions.
Developed the backend using Django with REST APIs.
Built an interactive frontend using React.js.
Simulated a spectrometer using OPC UA so that alloy composition values could be streamed into the application in real time.
Connected all components into a complete workflow from data acquisition to AI prediction and optimization.
Challenges we ran into
Finding high-quality alloy datasets with consistent material properties.
Cleaning and preprocessing metallurgical data for machine learning.
Balancing prediction accuracy with model performance.
Integrating OPC UA communication between systems.
Designing an optimization algorithm that recommends practical and manufacturable alloy compositions.
Building a smooth end-to-end workflow between the frontend, backend, AI models, and OPC UA simulation.
Accomplishments that we're proud of
Built a working AI-powered alloy optimization platform.
Successfully predicted multiple alloy properties using machine learning.
Implemented automatic alloy composition import through OPC UA simulation.
Created a user-friendly dashboard for engineers to analyze and optimize alloys.
Reduced the need for repeated trial-and-error experiments through AI-based recommendations.
Demonstrated how Industry 4.0 technologies and AI can work together in materials engineering.
What we learned
Applying Machine Learning to real-world materials engineering problems.
Data preprocessing and feature engineering for metallurgical datasets.
Integrating AI models into a full-stack web application.
Working with industrial communication protocols like OPC UA.
The importance of combining domain knowledge with AI to create practical engineering solutions.
Building scalable software that connects data acquisition, prediction, and optimization into one platform.
What's next for AI-Based Alloy Optimization
Integrate with real Optical Emission Spectrometer (OES) hardware instead of simulation.
Expand support for additional alloy families such as aluminum, titanium, and nickel-based alloys.
Improve prediction accuracy using advanced deep learning and ensemble models.
Add explainable AI features to show why specific alloy compositions are recommended.
Incorporate manufacturing constraints, cost, and sustainability into the optimization process.
Deploy the platform on the cloud for collaborative industrial use.
Develop a digital twin for alloy development and manufacturing.
Enable continuous learning from new production and laboratory data to improve recommendations over time.
Log in or sign up for Devpost to join the conversation.