PMSM Dataset with gym eletric motor platform

This dataset supports the research presented in the paper \Experimental Digital Twin with Evolvable PINNs for Real-Time Simulation and Field Data Adaptation\. It provides a structured collection of high-fidelity simulation data generated from the Gym Electric Motor (GEM) platform to evaluate digital twin models for permanent magnet synchronous motors (PMSMs) under dynamic and data-scarce conditions.The dataset is organized into two main parts. Part 1 contains baseline simulations of current, torque, and speed control for 50 motor configurations using a modified Model Predictive Control (MPC) strategy. Part 2 introduces extended scenarios by perturbing the angular velocity equation through varying damping coefficients, simulating system evolution and field data drift. For each motor, 200 new samples are derived across four damping intervals, providing a comprehensive benchmark for adaptive modeling.This dataset was used to train and evaluate the proposed Evolvable PINN framework, particularly for testing real-time state estimation, iterative forecasting, and online adaptation via residual learning modules. It is suitable for researchers working on digital twins, physics-informed learning, and industrial machine dynamics.