A locally adaptive ensemble approach for data-driven prognostics of heterogeneous fleets

Abstract

Abstract. In this work, we consider the problem of predicting the Remaining Useful Life (RUL) of a piece of equipment, based on data collected from an heterogeneous fleet working under different operating conditions. When the equipment experiences variable operating conditions, individual datadriven prognostics models are not able to accurately predict the RUL during the entire equipment life. The objective of the present work is to develop an ensemble approach of different prognostics models for aggregating their RUL predictions in an adaptive way, for good performance throughout the degradation progression. Two data-driven prognostics models are considered, an Homogeneous Discrete-Time Finite-State Semi-Markov Model (HDTFSSMM) and a Fuzzy Similarity-Based (FSB) model. The ensemble approach is based on a locally weighted strategy that aggregates the outcomes of the two prognostic models of the ensemble by assigning to each model a weight and a bias related to its local performance, i.e., the accuracy in predicting the RUL of patterns of a validation set similar to the one under study. The proposed approach is applied to a case study regarding an heterogeneous fleet of aluminum electrolytic capacitors used in electric vehicles powertrains. The results have shown that the proposed ensemble approach is able to provide more accurate RUL predictions throughout the entire life of the equipment compared to an alternative ensemble approach, and to each individual HDTFSSMM and FSB models.

An ensemble of models for integrating dependent sources of information for the prognosis of the remaining useful life of Proton Exchange Membrane Fuel Cells

“…To evaluate the quality of prognostic outcomes, a synthesis of the prognostic metrics is used [21,45,46,47,24]. Figure 20 illustrates the RUL predictions with uncertainties at different prediction time steps t λ .…”

“…In this work, we develop a hybrid approach based on an ensemble of models, which uses prediction outcomes provided by different degradation models fed by different deterioration measurements and properly combines them to provide the prognostic results. Ensemble of models have shown promising results for the prognostics of industrial systems [20,21,22,23,24,25]. For example in [23], an ensemble approach based on a semi-Markov model and a fuzzy similarity model has been developed for the predictions of the RUL of a heterogeneous fleet of aluminum electrolytic capacitors used in electric vehicle power trains.…”

Bootstrapped Ensemble of Artificial Neural Networks Technique for Quantifying Uncertainty in Prediction of Wind Energy Production

“…Ensemble approaches have been applied in various application fields to enhance the accuracy of the predictions and quantify their uncertainty [34,35]. The basic idea is that the individual models of the ensemble can complement each other by leveraging their strengths and overcoming their drawbacks: thus, the aggregation of their outcomes can boost the performance of the models [35,[65][66][67].…”

Ensemble Approach of Optimized Artificial Neural Networks for Solar Photovoltaic Power Prediction

“…With respect to 2), once the base prediction models of the ensemble are developed, an effective strategy for aggregating their prediction outcomes is required. Aggregation strategies can be generally categorized into statistics-based and model performance-based [39], [45], [46]. The former strategies assume that the base models contribute equally to the final prediction outcome of the ensemble by simply computing statistical values, e.g., average or median, of the base models' predictions.…”