A Robust Prognostic Indicator for Renewable Energy Fuel Cells: A Hybrid Data-Driven Prediction Approach

Daming Zhou (), Zhuang Tian and Jinping Liang
Additional contact information
Daming Zhou: Northwestern Polytechnical University
Zhuang Tian: Northwestern Polytechnical University
Jinping Liang: Northwestern Polytechnical University

A chapter in Sustainability, 2023, pp 167-197 from Springer

Abstract: Abstract As a power generation device, fuel cells are now widely studied as a new type of energy device due to their high energy density and non-pollution advantages. However, the large-scale industrialization of fuel cells is still difficult to realize. One of the important reasons is that its aging failure problem can lead to the degradation of performance and even the decay of useful lifetime. Prognostic and health management (PHM) is an effective technology to make reasonable predictions of fuel cell lifetime and state of health (SOH) to prevent economic loss and safety hazards due to aging failure. Prognostic is an important component of PHM, and it can predict the subsequent data trends of fuel cells using known measured data such as voltage, power, etc., and thus predict the SOH of fuel cells in the future period. This chapter first develops a hybrid prediction method with a state space model and a data-driven method. Then a prediction method with sliding prediction length is proposed. Finally, the accuracy and reliability of the hybrid method are verified. The main contributions of this chapter are as follows: 1. The proposed hybrid prediction method combines the advantages of the respective prognostic approaches, thus being able to fully make best the advantages of the state space model and data-driven prediction method. In this case, the hybrid prediction method can accurately predict linear degradation trends and the local fluctuations and nonlinear characteristics. Thus, the method compensates for the drawbacks of the single prediction method and has higher prediction accuracy. 2. The sliding prediction length method can update the aging data in the multi-step prediction process in time to ensure the data source of the training set. In addition, the method facilitates the assignment of weight factors for the fusion of different prediction methods to obtain better prediction accuracy. 3. A comprehensive comparison experiment is designed to verify the advancement of the proposed hybrid prediction method aiming at the whole dataset range and sliding prediction length range. It provides a feasible solution for the aging prediction method of fuel cells, especially for the multi-step prediction under actual operating conditions, thus avoiding the risks caused by the sudden degradation of fuel cells in actual operation.

Keywords: Sustainability; Prognostic and health management (PHM); State space model; Machine learning; Aging prediction; Maintenance management (search for similar items in EconPapers)
Date: 2023
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:isochp:978-3-031-16620-4_10

Ordering information: This item can be ordered from
http://www.springer.com/9783031166204

DOI: 10.1007/978-3-031-16620-4_10

Access Statistics for this chapter

More chapters in International Series in Operations Research & Management Science from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().