Encoding-Decoding-Based Quantized Learning Control Using Spherical Polar Coordinates.

Area of Science:

• Control Systems Engineering
• Signal Processing
• Robotics

Background:

• Iterative learning control (ILC) is crucial for systems requiring repetitive task precision.
• Quantization in control networks can degrade performance due to information loss.
• Existing ILC schemes often require complex parameter tuning when dealing with quantization.

Purpose of the Study:

• To develop and analyze a quantized iterative learning control strategy for discrete-time systems.
• To address the performance limitations imposed by signal quantization in control networks.
• To propose a scheme that avoids additional scaling parameters in the encoder and decoder.

Main Methods:

• Integration of an encoding-decoding mechanism with a spherical polar coordinate-based quantizer.
• Exploration of two scenarios: output-only encoding/decoding and output/input encoding/decoding.
• Design of a variable support sphere radius over iterations.
• Development of an ILC scheme based on the decoder's output.

Main Results:

• The proposed quantized ILC scheme achieves error-free tracking performance.
• The method eliminates the need for additional scaling parameters in the encoder and decoder.
• Demonstrated effectiveness through experimental validation on a permanent magnet synchronous motor.

Conclusions:

• The novel quantized iterative learning control strategy is effective for discrete-time systems.
• The proposed approach offers a simpler and more efficient solution compared to existing methods.
• Successful application in a practical system (PMSM) validates the theoretical findings.