Machine learning approximation#
The practical guide section approximation by machine learning methods will provide the skills to use ML models to solve physical problems, as well as the skills to use these models in the REPEAT VISION platform. The application of ML models on the REPEAT VISION platform allows:
Integrate machine learning models for interdisciplinary modeling of complex systems;
Use 2D and 3D model approximations in the project. Models of lager dimensionality can be created using MKE, МКО, PINN or other numerical methods;
Use real data for specific systems in the project. Data can be obtained from experiments or from diagnostic systems installed on facilities under operation. The data is approximated using machine learning models.
Note
If you do not have basic skills in using ML models on the REPEAT VISION platform, please read introductory tutorial before continuing with the material in this practical guide.
This practical guide presents the steps for implementing a simple machine learning model using calculated data for a heat transfer problem involving conduction in the wall of a product part.
The practical guide provides a detailed mathematical description of the models of physically informed neural networks (PINN), as well as ideas from current scientific articles on achieving better convergence of the solution of the two-dimensional non-stationary equation of thermal conductivity.
This practical guide presents the steps for implementing a simple machine learning model using calculated data for a heat transfer problem involving conduction in the wall of a product part.
В практическом руководстве приведено подробное описание создания модели машинного обучения для аппроксимации максимальных напряжений, действующих в отводе и возникающих при эрозионном износе в процессе эксплуатации газопровода-шлейфа.