A Data-Driven Control Algorithm With Time Delay Compensation Based on Optimized Recursive Adaptive Identification for Magnetic Levitation Systems

This study addresses the challenge of precise control of the maglev system under the condition of rapid time-varying dynamics and large time delay. Maglev technology has been applied in many fields, such as maglev train and magnetic bearing, for its non-contact operation, low maintenance cost and high energy efficiency. However, the inherent nonlinear characteristics and delayed response characteristics of the magnetic levitation system make it very difficult to control. In practical scenarios, it is difficult to establish an accurate magnetic suspension system model, which limits the use of model-based control methods. PID control method is also difficult to guarantee the control performance when dealing with unmeasurable load disturbance and system delay scenarios. This study proposes a data-driven control solution, which integrates an optimized recursive identification algorithm and an adaptive delay compensation algorithm. Uniquely, the control solution achieves real-time iterative optimization with only a few input and output data, reducing the data storage resource requirements, avoiding offline training, and ensuring real-time performance and efficient resource utilization in practical applications. Practitioners can use this new method to enhance the performance and reliability of maglev applications, such as maglev trains, industrial processes and precision instruments, so as to achieve more robust and adaptive control in the face of rapid dynamic changes and large time delays. Note to Practitioners—This study addresses the challenge of precise control of the maglev system under the condition of rapid time-varying dynamics and large time delay. Maglev technology has been applied in many fields, such as maglev train and magnetic bearing, for its non-contact operation, low maintenance cost and high energy efficiency. However, the inherent nonlinear characteristics and delayed response characteristics of the meglev system make it very difficult to control. In practical scenarios, it is difficult to establish an accurate magnetic suspension system model, which limits the use of model-based control methods. PID control method is also difficult to guarantee the control performance when dealing with unbearable load disturbance and system delay scenarios. This study proposes a data-driven control solution, which integrates an optimized recursive identification algorithm and an adaptive delay compensation algorithm. Uniquely, the control solution achieves real-time iterative optimization with only a few input and output data, reducing the data storage resource requirements, avoiding offline training, and ensuring real-time performance and efficient resource utilization in practical applications. Practitioners can use this new method to enhance the performance and reliability of maglev applications, such as maglev trains, industrial processes and precision instruments, so as to achieve more robust and adaptive control in the Phase of rapid dynamic changes and large time delays.