Field Programmable Gate Array Based Torque Predictive Control for Permanent Magnet Servo Motors

Area of Science:

• Robotics
• Control Systems Engineering
• Electrical Engineering

Background:

• Increasing demand for legged robots necessitates improved joint drive performance.
• The inner torque/current control loop is critical for overall joint system capabilities.
• Existing control methods like Field Oriented Control (FOC) have limitations.

Purpose of the Study:

• To propose a novel direct torque control algorithm for robotic joint drives.
• To enhance the dynamic performance and efficiency of motor control.
• To reduce the switching losses in the inverter.

Main Methods:

• Developed a direct torque control strategy incorporating a prediction model.
• Estimated motor torque by accounting for calculation and measurement delays.
• Implemented error estimation and torque tracking error compensation.
• Utilized a Field Programmable Gate Array (FPGA) for concurrent computation.

Main Results:

• The proposed controller significantly improved motor dynamic performance.
• Experimental verification confirmed the effectiveness of the predictive torque control.
• Reduced average switching times of the inverter compared to FOC.
• Demonstrated superior torque tracking accuracy.

Conclusions:

• The direct torque control with prediction offers enhanced dynamic performance for robotic joints.
• FPGA implementation enables efficient real-time control for complex algorithms.
• This approach is a viable solution for high-performance legged robot applications.