PMSM Parameter Learning via Voltage Error Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine parameters in permanent magnet synchronous motor (PMSM) drives, such as back-EMF constant and motor circuit resistance, vary widely throughout the operating region and over the life of the system, making accurate estimation critical for optimal torque and current control but challenging with existing feedforward approaches.
Innovation Solution
The method involves online learning of PMSM parameters during feedback current control by using final and estimated voltage commands as inputs to learning algorithms to estimate machine parameters, particularly the back-EMF constant and motor circuit resistance, through a parameter learning system that determines regions for estimation based on motor velocity and current values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feedforward approaches are used to estimate machine parameters, then the control system is simple to implement, but the estimation accuracy deteriorates due to wide parameter variations throughout the operating region and over system life
Solution Approach 1:
The patent implements an online parameter learning system that continuously estimates machine parameters (back-EMF constant and motor circuit resistance) during feedback current control operation. The system uses measured currents in the synchronously rotating reference frame and compares them with commanded currents to compute parameter estimates, thereby improving estimation accuracy through closed-loop feedback rather than open-loop feedforward approaches
Solution Approach 2:
The parameter learning system utilizes existing measured currents and control signals already available in the feedback control architecture to estimate machine parameters. The system serves itself by using the operational data from normal motor control to continuously update parameter estimates without requiring separate identification tests or external measurement equipment
2Measurement precision
If machine parameters are estimated using online learning during feedback control, then parameter estimation accuracy improves, but the computational complexity and processing requirements increase
Solution Approach 1:
The parameter learning system operates continuously during normal feedback control operation, utilizing the ongoing measurement and control processes to continuously update parameter estimates. This continuous operation allows the system to track parameter variations throughout the operating region and over system life without interruption or additional computational overhead periods
Solution Approach 2:
The same feedback control architecture and current measurements used for torque control are simultaneously utilized for parameter estimation. The system performs multiple functions (torque control and parameter learning) using the same hardware and measurement infrastructure, thereby avoiding additional complexity while improving estimation accuracy
3Measurement precision
If part-specific calibration is performed for each motor, then initial parameter accuracy improves, but the time required for calibration and system setup increases
Solution Approach 1:
The system performs preliminary parameter estimation using standard motor data, then continuously refines these estimates during normal operation through online learning. This preliminary action provides sufficient initial accuracy for most applications while eliminating the need for time-consuming part-specific calibration procedures
Solution Approach 2:
The system adapts to parameter changes automatically through continuous online estimation during operation. Rather than requiring fixed calibration values, the system dynamically adjusts parameter estimates based on actual motor behavior, thereby eliminating calibration time while maintaining accuracy through adaptive parameter updates
Data Source
AI summary
Technical solutions are described for estimating machine parameters of a permanent magnet synchronous motor (PMSM) drive. An example method includes determining a region for estimating a machine parameter based on a motor velocity value and a motor current value. The method further includes, in response to the motor velocity value and the motor current value being in the region, estimating an error in estimated voltage command, and estimating the machine parameter using the error in estimated voltage command.