One-shot backpropagation for multi-step prediction in physics-based system identification -- EXTENDED VERSION
This work addresses system identification for physics-based models like space debris, though it appears incremental as it extends backpropagation with physical constraints rather than introducing a fundamentally new approach.
The paper tackles the problem of identifying unknown parameters in dynamical systems by developing a physics-based framework that computes gradients for multi-step prediction in closed form while enforcing physical constraints. The method successfully identified the inertia matrix of space debris, demonstrating reliable physical adherence of estimated parameters.
The aim of this paper is to present a novel physics-based framework for the identification of dynamical systems, in which the physical and structural insights are reflected directly into a backpropagation-based learning algorithm. The main result is a method to compute in closed form the gradient of a multi-step loss function, while enforcing physical properties and constraints. The derived algorithm has been exploited to identify the unknown inertia matrix of a space debris, and the results show the reliability of the method in capturing the physical adherence of the estimated parameters.