Magnetic Position Estimation in Ferromagnetic Systems Involving Significant Hysteresis

The proximity of a magnetic sensor to a magnet has been utilized for position measurement in many applications involving small ranges of motion. However, such position measurement becomes challenging when the application involves a ferromagnetic environment and larger ranges of motion. The movement of a magnet magnetizes and demagnetizes the ferromagnetic surroundings and creates significant hysteresis in the magnetic field versus position relationship. This relationship can no longer be represented by an algebraic mathematical function and needs to take the history of previous motions into consideration for calculation of position from magnetic field measurement. This paper presents a position estimation system that utilizes various models to represent the hysteresis behavior, including an interacting multiple model approach, a Preisach model, and a modified Preisach model. The developed position estimation system is evaluated on a large hydraulic actuator used in mobile construction vehicle applications. Experimental results show that an unscented Kalman filter using the modified Preisach hysteresis model can provide a position estimate with an accuracy better than 2% of stroke, in spite of significant hysteresis.