Joint identification of permanent magnet synchronous machine and inverter
arXiv:2607.25739
2026
Dynamics
1 ideas extracted · analyzed Aug 31, 2026
What the math gives to ML
The paper provides a constructive joint-identification mechanism for separating intrinsic system dynamics from command-dependent actuator distortion. Its key asset is the recognition that fitting a flexible dynamics model to commanded inputs alone creates omitted-variable bias when the realized input contains an unknown nonlinear disturbance; jointly estimating a low-dimensional disturbance model removes this bias. A transferable neural-network design is to augment a learned dynamics model with a parsimonious actuator-distortion branch, train both branches from observed state transitions, and test whether the learned state dynamics become invariant across actuator regimes.
Ideas from this paper
✓✓ Beats tuned baseline
2026
Add a low-dimensional actuator-distortion model alongside a neural state-transition model instead of assuming that commanded control is the realized control. For a transition $x_{t+1}=F_\theta(x_t,u_t^{\mathrm{cmd}}+d_\phi(x_t,u_t^{\mathrm{cmd}}))$, jointly fit the intrinsic dynamics parameters $\theta$ and disturbance parameters $\phi$, with a strong simplicity prior on $d_\phi$. This should prevent the dynamics network from absorbing systematic actuator errors and improve cross-regime…
Useful7/10
Difficulty5/10
Novelty7/10