Online Multi-Parameter Identification for PMSM Parameter Monitoring Based on a ZOH Model and Dual-Sampling Strategy

Area of Science:

• Electrical Engineering
• Control Systems

Background:

• Online parameter identification for PMSMs faces challenges from model discretization, matrix rank deficiency, and inverter nonlinearities.
• Existing methods struggle with accuracy, particularly under high-speed operation.

Purpose of the Study:

• To propose a high-precision virtual sensor for multi-parameter identification of PMSMs.
• To overcome limitations of current identification techniques, enhancing accuracy across diverse operating conditions.

Main Methods:

• Utilized Zero-Order Hold (ZOH) discretization for accurate PMSM modeling, accounting for rotor position variations.
• Implemented a dual-sampling strategy combined with d-axis small-signal injection to address matrix rank deficiency.
• Employed the Forgetting Factor Recursive Least Squares (FFRLS) algorithm for online multi-parameter identification.
• Developed an inverter nonlinear voltage compensation strategy considering dead-time effects, voltage drops, and turn-on delays.

Main Results:

• The ZOH method significantly reduced discretization errors compared to forward Euler, especially at high speeds.
• The combined strategy effectively resolved rank deficiency issues in the identification matrix.
• Experimental validation confirmed enhanced parameter identification accuracy across the entire speed range.
• High-speed identification errors for resistance, inductance, and flux linkage were within 5.47%, 4.05%, and 2.46%, respectively.

Conclusions:

• The proposed method offers a robust solution for accurate online PMSM parameter identification.
• The integration of ZOH discretization and inverter nonlinearity compensation leads to superior performance, particularly under demanding high-speed conditions.