A Survey on Machine Learning Applications in Storage Quality Monitoring and Fault Detection
Fault detection and diagnosis (FDD) techniques, like hardware redundancy and threshold-based monitoring, are becoming increasingly ineffective in detecting early-stage faults and ensuring operational reliability in increasingly complex and interconnected mechanical, electrochemical, chemical, thermal, and electrical energy storage systems. As a result, ML and AI methods are becoming more popular because to their strong capacity to model nonlinear, high-dimensional, and noisy data settings. ML-based solutions, including supervised, unsupervised, and deep learning methods, have demonstrated superior fault detection accuracy and adaptability across various applications from wind turbine systems and gearboxes to magnetic sensor arrays in nuclear fusion reactors. Recent trends in distributed ML architectures enable real-time, in-sensor diagnostics, further enhancing responsiveness and system resilience. However, practical implementation faces challenges such as sensor data inconsistencies, scarcity of labelled fault data, real-world data noise, multi-source data integration, and the gap between lab research and field deployment. The paper delves into the various types of systems, diagnostic methods, and failure stages in energy storage and mechanical systems that use ML-driven FDD. It also addresses important implementation issues and potential future research paths.
Authors
- Parth Gautam (ORCID: https://orcid.org/0009-0004-8999-9371)
Institutions
- Mandsaur University (IN)
Publication Details
- Journal
- International journal of data science and machine learning.
- Published
- 2026-11-18
- DOI
- https://doi.org/10.55640/ijdsml-06-02-17
- Primary Topic
- Fault Detection and Control Systems
- Type
- article
- Field-Weighted Citation Impact
- 0.00