Designing and manufacturing a new part or product, such as a car engine or wind turbine, can be time-consuming and costly. To combat process limitations, the US Department of Energy’s Argonne National Laboratory (Lemont, IL) is using cutting-edge machine learning techniques to help organizations reduce design time from months to days and slash development costs.
Machine learning is a type of artificial intelligence (AI) that trains computers to discover hidden patterns in data to make novel predictions without being explicitly programmed. This technology can be applied to manufacturing to quickly find the best design for a product or the most efficient production process.
“Machine learning … is similar to how biologists study fruit flies instead of humans. The flies share significant characteristics with humans, but they can generate and evolve much faster,” said Janardhan Kodavasal, mechanical engineer in Argonne’s Energy Systems division.
Traditional optimization of a new product design involves much experimental testing and evaluation of many prototypes. As the volume and complexity of data derived from these tests increase, industry relies more on high-fidelity computer models that virtually represent real-world devices and processes.
While they are an improvement over costly physical development and testing, high-fidelity models are computationally intensive and take a long time to run. Argonne’s solution is to augment high-fidelity modeling with machine learning. A job that might take hours to run using high-fidelity modeling takes milliseconds with machine learning.
“Machine learning converts the complex physical processes represented by the virtual model into a compact computational process that can be run in much less time,” said Kodavasal, who heads the initiative.
To start the process, the scientists run several thousand simulations of a high-fidelity model on Argonne’s supercomputer Mira, at the same time and with different inputs. This step generates virtual data that train the machine learning model to find the best input combination.
The scientists then use an evolutionary approach to find an optimal design by prioritizing the product’s desired outputs. With the ideal outputs specified, the model runs a set of designs and chooses the best ones from that generation. Those designs exchange some of their input features, like children taking genes from their parents, and the model is run again. The process repeats until the design can’t be enhanced any further. Once the machine learning model spits out the optimal inputs, the scientists run the original high-fidelity model with those inputs to verify it is the ideal set. The technique can help develop parts and products for virtually any industry, including materials, transportation and utilities.
Machine learning models can even maximize the efficiencies of processes, from ventilation systems in a building to the production of a car dashboard to 3D printing. “We can help companies at any step of the process, whether that is developing the initial high-fidelity model of the system or just implementing the machine learning aspect,” said Kodavasal.
As demand for machine learning grows, Argonne scientists will expand its competencies. Techniques such as active learning will allow machine learning models to interact with high-fidelity models to improve accuracy and efficiency as the models provide data and real-time optimization that help guide manufacturing processes as they happen.
Connect With Us