An extended physics informed neural network for preliminary analysis of parametric optimal control problems

Nicola Demo,Maria Strazzullo,Gianluigi Rozza

DOI: https://doi.org/10.1016/j.camwa.2023.05.004

2023-06-14

Abstract:In this work we propose an extension of physics informed supervised learning strategies to parametric partial differential equations. Indeed, even if the latter are indisputably useful in many applications, they can be computationally expensive most of all in a real-time and many-query setting. Thus, our main goal is to provide a physics informed learning paradigm to simulate parametrized phenomena in a small amount of time. The physics information will be exploited in many ways, in the loss function (standard physics informed neural networks), as an augmented input (extra feature employment) and as a guideline to build an effective structure for the neural network (physics informed architecture). These three aspects, combined together, will lead to a faster training phase and to a more accurate parametric prediction. The methodology has been tested for several equations and also in an optimal control framework.

Machine Learning,Numerical Analysis

What problem does this paper attempt to address?

The paper aims to address the following issues:

1. **Fast solution of parameterized partial differential equations (PDE(μ))**: The authors propose an extended Physics-Informed Neural Network (PINN) method to handle parameterized partial differential equations. This method can simulate parameterized phenomena in a short time, especially in real-time and multi-query settings.

2. **Optimization control problems (OCP(μ))**: The paper also focuses on parameterized optimization control problems. These problems have wide applications in fields such as fluid mechanics, biomedical science, and environmental science. The authors attempt to use the PINN method to handle these complex systems, particularly in situations where parameters change physical properties or model attributes.

Specifically, the main contributions of the paper include:
- Proposing a method to use physical information for neural network structure design (Physics-Informed Architecture, PI-Arch) to improve prediction accuracy.
- Utilizing extra features to accelerate the training process while maintaining prediction accuracy.
- Applying the PINN method to multiple equation systems, especially in parameterized settings, by combining physical information structures with extra features as enhanced inputs to accelerate the training process.

In summary, the goal of the paper is to reliably predict physical phenomena using the PINN method when dealing with complex systems and to achieve fast and accurate solutions in optimization control problems.