Assessing and improving learning outcomes for power management experiments using cognitive graph

Abstract

The series of Power Management Lab Kits (PMLK), released by Texas Instruments (TI), have been globally adopted in classroom settings. We propose a cognitive graph-based method to assist better adoption of TI-PMLK in Chinese power electronics education and specifically assessment of the experiment-based learning experience. First, construct a power management cognitive graph. Then, identify knowledge weaknesses using a Deterministic Inputs Noisy And Gate (DINA) model based cognitive diagnosis method. An Automatic Items Generation System (AIG) is then developed to generate personalized experiment items for any given student. Finally, learning outcomes evaluation is generated from the AIG experimental items using Bayesian psychometric models.