Diagnostics and Prognostics in Power Plants: A systematic review

Abstract

Failures in power plants can lead to significant power interruptions and economic losses. Prognostics and Health Management (PHM) serves as a predictive maintenance technique by detecting and diagnosing faults while forecasting potential failures. This systematic review analyzes trends in diagnosis and prognosis in power plants using scientometric analysis, summarizes the datasets and components targeted by researchers, outlines the advantages and drawbacks of popular methods, and reports detailed methodologies from selected literature. The complex nature of power plants presents significant challenges for implementing PHM effectively. Data-driven techniques, particularly machine learning and deep learning, have emerged as popular solutions to address these challenges. While diagnostic methods have seen substantial advancements, prognostics in power plants remain underdeveloped and require further investigation. This paper discusses the challenges associated with fault diagnosis and prognosis, emphasizing that addressing these issues could significantly enhance the effectiveness of PHM. By reviewing recent methodological advancements, summarizing the pros and cons of various methods, and identifying key challenges, this paper contributes to a deeper understanding of the field and highlights opportunities for future research.