Intellegens and Ansys push boundaries of additive manufacturing with ML
Machine learning (ML) specialist, Intellegens, and engineering simulation company, Ansys, have announced a collaboration to integrate machine learning methods into additive manufacturing (AM) workflows, to accelerate the development of reliable and repeatable AM processes. The partnership, combining both companies’ technologies will make it quick and easy for AM project teams to analyse data from experiment, simulation, or production generating models that capture vital insights, say the partners. Models will be used to optimise process parameters and powders, improving the quality of AM parts while cutting time to market.
The agreement will embed Intellegens’ ML technology, Alchemite, within the Ansys materials data management platform, Granta MI.
Intellegens provides an ML toolset that can train deep neural networks from fragmented, corrupt, or noisy datasets. The technique, originally developed at the University of Cambridge, is encapsulated in Alchemite. By unravelling data problems that are not accessible to traditional ML approaches, Alchemite offers accurate models that can predict missing values, find errors and optimise target properties. Suitable for any numeric dataset, Alchemite has demonstrated particular value in the design, development, and application of materials, chemicals, formulations, and drugs. It has enabled organisations to break through data analysis bottlenecks, reduce the amount of time and money spent on research, and support better, faster decision-making, says Intellegens.
Alchemite deep learning algorithms rapidly find relationships within complex datasets, even when that data is ‘sparse’ (i.e., has many empty values). This makes Alchemite particularly well suited to AM teams seeking to exploit data brought together from multiple sources. It extracts all possible knowledge from the data to identify the critical combinations of factors that ultimately control the performance of AM parts. Alchemite needs no prior knowledge of which parameters are likely to be important, Ansys says this is a significant advantage in the emerging field of AM.
Alchemite optimises process parameters for AM processes and provides computational design for AM materials. It also provides failure analysis and quality control, with data validation and gap-filling and assisted design of dxperiments (DoE) for AM.
Granta MI is the de facto standard for materials data management in engineering enterprises and is applied in AM applications to capture, in a single place, all of a company’s AM data. This includes data on the properties of powders and raw materials, machine build parameters, post-build processing data, test results for AM parts, and simulation data from the Ansys AM simulation suite. Integrating Alchemite into this holistic system will make it straightforward to analyse the full range of this data in the search for key process/property relationships and to continuously improve models as the data is updated, says Ansys.
“Intellegens’ machine learning technology offers a ready-made solution to key data analysis challenges faced by our AM customers,” commented Rob Davis, director of product management, at Ansys. “Integration with Ansys Granta MI creates a unified workflow for capturing and applying results from AM testing, simulation, and production,” he added.
“Merging the data management capabilities of Ansys’ Granta MI with the machine learning prowess of Alchemite is a perfect fit, promising to deliver deep insights to AM workflows,” explained Intellegens’ CTO, Dr Gareth Conduit.
