A Novel Data-Driven Control Method for a Class Magnetic Levitation Systems with Time Delay and Noise Interference

A Novel Data-Driven Control Method for a Class Magnetic Levitation Systems with Time Delay and Noise Interference

연구 분야: Software Development

학회: 2024 IEEE 13th Data Driven Control and Learning Systems Conference (DDCLS)

In response to the challenges posed by nonlinear large time delays and external noise interference in the control process of magnetic levitation systems, this paper proposes a model-free adaptive control algorithm based on time delay compensation and filtering (DCF-MFAC) to attain magnetic levitation systems control. Firstly, an adaptive control algorithm is designed by introducing a dynamic linearization technique with time-varying parameters based on pseudo-gradient (PG), which linearizes the unknown nonlinear system and facilitates controller design. Subsequently, the algorithm utilizes Smith predictor and second-order integrator-based estimation to estimate the output differentials of the controlled object, compensating for system time delays. Additionally, the algorithm introduces a filter to attenuate or suppress noise in the system, ensuring smoother control signals. The effectiveness of the algorithm in magnetic levitation system control is verified through joint simulation using HUMUSOFT real-time toolbox and Matlab.