Engineering Design Exploration through numerical analysis alone can be a lengthy, expensive task especially when number of design iterations increases. Employing power of predictive models in engineering design provides more flexibility which translates into more time for genuine creativity.
Altair's Design Exploration tool, HyperStudy, provides supervised learning models both in regression (Least Squares, Moving Least Squares and Radial Basis Function) and classification (SciPy-based Random Forest) categories. In addition to the out-of-the-box algorithms, HyperStudy allows the use of custom algorithms which was covered in a recent blog, DIY Machine Learning in Altair Simulation Suite is Shockingly Easy. However, this is not where all the "fun" stops. One of Altair's Data Analytics tool, Knowledge Studio, offers wide range of both supervised and unsupervised algorithms which can be integrated into HyperStudy as a model type.
Knowledge Studio is a machine learning and artificial intelligence platform which depends on data generated elsewhere and in engineering, data is generally collected by performing series of numerical analysis. HyperStudy, through appropriate Design of Experiments methods, can generate the necessary data. Once the data is generated, it can be exported to an Excel file (*.xlsx) via an option dedicated for Knowledge Studio in Report step.
After data (run matrix) is exported to an excel file, it is then imported into Knowledge Studio to build a machine learning model of your choice.
For each built model, there is a *.kdm file (i.e., Deep Learning.kdm) in the project directory and the kdm file of interest is used as a resource file for the Knowledge Studio connection.
