Neural Network Tuning of FSMPC for Drives
AI system automatically tunes complex predictive controllers for five-phase induction motors using experimental data.
A new research paper from Juana M. Martínez-Heredia and José L. Mora introduces an AI-powered method for tuning complex motor controllers. The system uses a neural network to automatically adjust the parameters of a Finite State Model Predictive Control (FSMPC) system for an induction motor. This controller manages two critical loops: one for motor speed and another for stator current. The researchers validated their approach using a five-phase induction machine in an experimental setup, training the neural network with data collected from step tests performed on the physical hardware.
This work represents a significant application of machine learning in industrial control systems. Manually tuning FSMPC parameters is a time-consuming and expert-dependent task. By automating this process with a data-driven neural network, the researchers aim to reduce setup time, improve consistency, and potentially unlock better performance from electric drives. The use of experimental data for training grounds the AI model in real-world physics, making it a practical tool for engineers. This fusion of AI and control theory, published on arXiv (ID: 2603.08816), points toward more autonomous and efficient industrial automation.
- Automates tuning of Finite State Model Predictive Control (FSMPC) for induction motors using a neural network.
- Manages parameters for both the speed control loop and the stator current control loop simultaneously.
- Trained and validated on a real five-phase induction machine using data from experimental step tests.
Why It Matters
Automates a complex engineering task, potentially cutting motor drive setup time and improving industrial efficiency.