Improved performance of Kalman filter for junction temperature estimation with Autocovariance Least-Squares method

Junction temperature estimation is a crucial topic for assessing the state-of-health of semiconductors. In this work a measurement board provides an initial estimation of the temperature. Afterwards, a Kalman filter improves this estimation. The ALS method is used for obtaining an optimal performance of the Kalman filter. Finally, the proposed methodology is deployed in an experimental setup and the temperature estimation is validated with the help of a real measurement through fiber optics sensors.