Narrow Results

Prognostics and health management transform predictive maintenance by identifying which components need maintenance and when they should be performed. While there are several metrics to quantitatively assess the accuracy of remaining useful life predictions, fewer studies have explicitly modeled the economic benefits of implementing prognostics and health management, such as return on investment, cost reduction in the life cycle, and maintenance metrics driven by data over a period. We extend data-driven metrics from renewal theory, such as average cost per unit time, utilization per unit time, safety, and availability, rendering them suitable for application in the context of PHM methods. Furthermore, a comparative framework provides systematic evaluation and selection of PHM methods for multi-objective decision analysis. We explicitly decouple degradation models and include an unscented Kalman filter and particle filtering algorithms that iteratively update estimates of a model’s parameters. This decoupling approach enables direct comparison of alternative combinations of degradation models and PHM algorithms and a method to select a time horizon that balances tradeoffs between competing metrics according to stakeholder preference. The approach is applied to lithium-ion batteries. Based on data-driven maintenance metrics, the framework can be applied to select a combination of model and algorithm that balances tradeoffs between competing objectives such as cost and utilization. In addition, the framework is general and accommodates both existing and future degradation models and algorithms.

Fault prognosis in industrial plants is a complex problem, and time is an important factor for the resolution of this problem. The main indicator for the task of fault prognosis is the estimate of remaining useful life (RUL), which essentially depends on the predicted time to failure. This paper introduces a temporal neuro-fuzzy system (TNFS) for performing the fault prognosis task and exactly estimating the RUL of preheater cyclones in a cement plant. The main component of the TNFS is a set of temporal fuzzy rules that have been chosen for their ability to explain the behavior of the entire system, the components’ degradation, and the RUL estimation. The benefit of introducing time in the structure of fuzzy rules is that a local memory of the TNFS is created to capture the dynamics of the prognostic task. More precisely, the paper emphasizes improving the performance of TNFSs for prediction. The RUL estimation process is broken down into four generic processes: building a predictive model, selecting the most critical parameters, training the TNFS, and predicting RUL through the generated temporal fuzzy rules. Finally, the performance of the proposed TNFS is evaluated using a real preheater cement cyclone dataset. The results show that our TNFS produces better results than classical neuro-fuzzy systems and neural networks.