Fault diagnosis for e-seal unreadability using learning Bayesian networks

Due to the physics of radio frequency, the read rate of RFID e-seals has trouble achieving optimal levels in real-world applications. Hence, continuous monitoring of fault causes and immediate treatment of fault occurrence is critical for the applications of e-seal technology. To provide a flexible and easy-to-implement fault diagnosis tool for the problem of e-seal unreadability, this study adopts a noisy OR-gate and learning Bayesian networks approach to model the problem and make probabilistic inference. Marginalization mechanism is also applied to update the posterior probability distribution of parameters. With the help of this proposed tool, port administrators can keep monitoring the marginal probabilities of causes and effect to improve the reliability of e-seals. When unreadability occurs, the recommendation list for inspection can be produced immediately through the computation of prediction probabilities. ICIC International