A Method of False Alarm Recognition in Built-in Test Considering Its Time Series Characteristics

Built-in test (BIT) technology provides an effective way of state monitoring, and is widely used in many fields. However, the false alarm prevents its wider usage and reduces its reliability. The existing BIT methods typically assume that the probability of generating a false alarm is a stationary process, which is typically a far too simplifying assumption. In this article, we propose a new method of false alarm recognition, considering the time series characteristics of BIT. The presented method divides the evolution process of false alarm into three phases, and the false alarm recognition in the intermediate phase is discussed and studied. Multiclassification support-vector machine is introduced to identify the intermediate phase, and a new feature, the irrational entropy, is proposed as its input. Based on the phase identification result, we construct a hidden Markov model for every state-normal, false alarm, and faulty state. The state with the highest probability is taken as the true state. Finally, a dc-dc converter is taken as an example, and the proposed method is applied and compared with the repeated test method, artificial neural network and recurrent neural network. The results show that the presented method outperforms the other algorithms.