Machine learning approach for automated preform-to-product design mapping
An ML-based method that learns a mapping from preform designs to final products to support automated packaging design. It uses feature representations that capture key shapes and symmetries, combining physics-informed modeling with data-driven learning. The approach includes both forward and inverse maps to search for optimal outputs and infer input designs from target outputs, aiming to improve predictive performance and streamline automated design workflows.
Overview
The proposed solution is an automated packaging design approach that uses machine learning (ML) to map preform designs to final products. The method optimizes the design process by incorporating feature representations that capture essential shapes and symmetries, aiming to balance physics-based and data-driven models. By using ML, the solution targets improved predictive performance and streamlined automated design workflows.
Technical specifications
Key features:
- Feature Representations: Uses bespoke feature representations that exploit essential low-dimensional characteristics such as smoothness and symmetries.
- Dual Approach: Employs a forward map to search input space for optimal outputs and an inverse map to determine input designs from specified outputs.
- Physics-Informed ML Models: Integrates physics elements to enhance model accuracy and efficiency.
- Optimized Use of Historical Data: Efficiently uses a preform/product historical dataset to train models, reducing the need for extensive compute resources.
Applications:
- Automated packaging design
- Optimization of manufacturing processes
- Custom product design
Technology readiness level
This technology has reached Technology Readiness Level 6, indicating that it has been demonstrated in a relevant environment and is ready for further development and deployment in industrial settings.
About DAMSL Ltd
DAMSL Ltd is a research-led data science company founded in 2017 that provides bespoke solutions in data analytics, statistical modelling, and machine learning. Operated by a team of internationally renowned academic researchers and professors, the firm specializes in cutting-edge mathematics, computer science, and statistics. Their service offerings are categorized into three main areas: providing bespoke algorithms, code, and software products; delivering expert consultancy on data modelling and decision-making; and conducting tailored training courses for businesses. Their technical expertise includes Bayesian learning, uncertainty quantification, image analysis, and the development of digital twins, with capabilities that span R and Python-based data science environments.
By offering commercial access to advanced research methods, DAMSL helps partners across diverse sectors, such as retail, healthcare, energy, and pharmaceuticals, solve complex analytical problems and extract actionable value from their data. The company leverages deep academic knowledge to address challenges ranging from process parameter optimisation to clinical drug discovery and human behavioural data analysis. Their engagement model focuses on providing rigorous scientific insights, ensuring that clients benefit from evidence-based, robust statistical techniques and high-performance computing capabilities to drive business performance and innovation.
Inquire about this partnering listing
Industry applications
delivered to your inbox.