LPV Phase Current Observer for PMSM Sensor Failure Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional current sensor failure detection algorithms in PMSM motor control systems are inadequate for comprehensive failure detection and estimation of phase current signals, leading to operational instability.
Innovation Solution
A linear parameter-varying (LPV) observer that estimates phase current values using reference control loop voltages, angular position, and angular velocity signals, with a failure detector to replace sensed current values with estimated values in case of sensor failure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional current sensor failure detection algorithms are used, then the system can detect specific sensor faults, but the detection is not comprehensive and fails to detect all types of sensor failures
Solution Approach 1:
The LPV observer is designed to perform multiple functions: it estimates phase current signals for normal operation and simultaneously detects all types of current sensor failures including stuck-at faults, gain errors, and offset errors. This multi-functional approach allows a single system to handle diverse failure modes comprehensively.
Solution Approach 2:
The system continuously compares the estimated current signals from the LPV observer with the actual sensor measurements and uses this feedback to detect discrepancies indicating sensor failure. The observer gain is adjusted based on feedback from the residual signal between estimated and measured currents, enabling adaptive failure detection.
2Measurement precision
If conventional observers such as Luenberger observer or linear observer are used, then the system can estimate current signals, but the estimation performance is inadequate for reliable failure detection
Solution Approach 1:
The observer transitions from a static gain structure to a dynamic parameter-varying structure where the observer gain adapts based on operating conditions. The LPV observer uses scheduling parameters (such as motor speed and load) to dynamically adjust observer parameters, improving estimation accuracy across varying operational states and enhancing failure detection reliability.
Solution Approach 2:
The system changes observer parameters dynamically based on operating conditions. The LPV observer modifies its gain matrix and other parameters according to scheduling variables that reflect the current motor state, allowing the estimation algorithm to adapt to different operating regimes and maintain high precision throughout the operating range.
3Productivity
If the system operates with closed-loop current control, then the control performance is improved, but sensor failures become undetectable in the output feedback system
Solution Approach 1:
The LPV observer acts as an intermediary that independently estimates the current signals without relying solely on sensor feedback. By creating an alternative estimation path that uses motor model and measured voltages/positions, the system can detect sensor failures even while operating in closed-loop control mode, as the observer provides a reference against which sensor accuracy can be validated.
Data Source
AI summary
A linear parameter-varying (LPV) observer for an electric motor controller. The controller includes a control loop with a current sensor that senses phase current signals driving an electric motor and that provides sensed current values used in the control loop. The observer includes an input interface and a processor. The input interface provides a sensed angular speed value, a sensed angular position value, physical parameters of the electric motor, and a pair of control loop control voltages used in the control loop. The processor calculates estimated current values indicative of the sensed current values using the angular speed value, the angular position value, the pair of control voltages, and the electric motor physical parameters. A failure detector may compare the sensed current values with the estimated current values to detect a failure, and substitute the estimated current signals for the sensed current signals when a failure is detected.