A simple state-based prognostic model for filter clogging

In today's maintenance planning, fuel filters are replaced or cleaned on a regular basis. Monitoring and implementation of prognostics on filtration system have the potential to avoid costs and increase safety. Prognostics is a fundamental technology within Integrated Vehicle Health Management (IVHM). Prognostic models can be categorised into three major categories: 1) Physics-based models 2) Data-driven models 3) Experience-based models. One of the challenges in the progression of the clogging filter failure is the inability to observe the natural clogging filter failure due to time constraint. This paper presents a simple solution to collect data for a clogging filter failure. Also, it represents a simple state-based prognostic with duration information (SSPD) method that aims to detect and forecast clogging of filter in a laboratory based fuel rig system. The progression of the clogging filter failure is created unnaturally. The degradation level is divided into several groups. Each group is defined as a state in the failure progression of clogging filter. Then, the data is collected to create the clogging filter progression states unnaturally. The SSPD method consists of three steps: clustering, clustering evaluation, and remaining useful life (RUL) estimation. Prognosis results show that the SSPD method is able to predicate the RUL of the clogging filter accurately.