A hidden fault prediction model based on the belief rule base with power set and considering attribute reliability

Hidden faults are important characteristics of complex systems that cannot be observed directly. The hidden behavior of a system, such as the health state and safety level, has a direct correlation with these hidden faults. After the predicted hidden behavior reaches a fault boundary, certain measures must be taken to avoid fault occurrences. Thus, hidden faults can be predicted by the hidden behavior of a system. The belief rule base (BRB) has been used to predict hidden behaviors. However, two problems remain to be solved in engineering practice. First, when the observed information is absent, ignorance may exist in the output. If only global ignorance is considered, it may be unreasonable in certain cases, which can influence the prediction model. Second, the effects of disturbance factors such as noise and sensor quality may cause the reliability of the gathered information to decline, which indirectly leads to unreliability of the hidden behavior. Thus, to address the global ignorance and unreliable hidden behavior, a new hidden BRB model with a power set and considering attribute reliability (PHBRB-r) is proposed for hidden fault prediction. In the PHBRB-r model, the effects of disturbance factors on hidden behavior are considered using attribute reliability, and the discernment frame is a power set. A case study of hidden fault prediction is conducted to demonstrate the effectiveness of the PHBRB-r model.