A Robust and Explainable Data-Driven Anomaly Detection Approach For Power Electronics

Beattie, Alexander; Mulink, Pavol; Sahoo, Subham; Christou, Ioannis T.; Kalalas, Charalampos; Gutiérrez-Rojas, Daniel; Nardelli, Pedro H. J.

doi:10.1109/smartgridcomm52983.2022.9961002

Cited by 4 publications

(4 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The matrix profile algorithm and anomaly transformer are two statistical methods used for anomaly detection. Power electronic signals and other time series data can be detected using these methods [12]. The Mahalanobis Distance (MD) technique is employed for the detection of anomalous behavior in electronic commodities through the comparison of MD values against baseline values [56].…”

Section: Methodsmentioning

confidence: 99%

“… Point anomalies: Occurs when data samples exhibit substantial deviation from the norm or expected behavior within a dataset [23]. For example, in the field of power electronics, a sudden drop or increase in the level of voltage in an electrical system is considered a point anomaly [12].  Contextual anomalies: These kinds of anomalies might be classified as normal or anomalous depending on the surroundings and situations around them [24].…”

Section: Anomaly Taxonomymentioning

confidence: 99%

“…Anomaly detection is essential to maintain complex systems, such as power electronic systems, by ensuring full coverage and minimal response time across all platforms, operating systems, and data centers [11]. This essentiality is because anomalies can result in system downtime, decreased efficiency, and higher maintenance expenses [12]. Furthermore, there is a growing need for real-time anomaly detection and classification systems in power electronic systems due to the integration of various technologies like renewable energy sources, smart grids, and cyber-physical systems [10].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Power electronics anomaly detection and diagnosis with machine learning and deep learning methods: A survey

Hossein Rahimighazvini,

Zeyad Khashroum,

Maryam Bahrami

et al. 2024

Int. J. Sci. Res. Arch.

View full text Add to dashboard Cite

Power electronics pertains to the conception, regulation, and utilization of electronic power circuits to proficiently administer and transform electrical energy. Power electronics play a crucial role in maintaining the reliability, efficiency, and security of complex production systems. Also, increasingly important in various applications such as renewable energy systems, electric vehicles, and industrial automation. However, modern power electronics systems are vulnerable to both cyber and physical anomalies due to the integration of information and communication technologies. So far, different methods have been used to detect abnormalities. This survey provides an overview of the state-of-the-art in anomaly detection in power electronics using machine learning and deep learning methods. It highlights the potential of these techniques in addressing the growing complexity and vulnerability of power electronics systems.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Anomaly Taxonomymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Power electronics anomaly detection and diagnosis with machine learning and deep learning methods: A survey

Hossein Rahimighazvini,

Zeyad Khashroum,

Maryam Bahrami

et al. 2024

Int. J. Sci. Res. Arch.

View full text Add to dashboard Cite

show abstract

“…Yet, a high quality dataset can rarely be achieved from data collected from real-world systems. That is because the available real data is usually not enough and most importantly, it can not accommodate different non-linear events, as such events tend to be rare [6], [7].…”

Section: Introductionmentioning

confidence: 99%

Data-Driven Controllability of Power Electronics Under Boundary Conditions - A Physics-Informed Neural Network Based Approach

Sahoo

Blaabjerg

2023

2023 IEEE Applied Power Electronics Conference and Exposition (APEC)

Self Cite

View full text Add to dashboard Cite

This paper introduces physics-informed neural network (PINN) for control of grid connected converter by fusing its underlying equations into the training process, thereby reducing the requirement of qualitative training data. In comparison to the traditional data-driven methods, which either incur a significant computational burden, or use overly conservative surrogate models, it is explored that PINN can be easily optimized as per the performance requirements and is significantly superior in terms of computation time, data requirements (trained using only 3000 datapoints), and prediction accuracy (an average of 98.76%). As a result, PINN unravels new modeling orientation for power electronics, and is well-suited for commercial applications. Finally, its robustness under various grid disturbances has been validated under experimental conditions.

show abstract

Toward eXplainabile Data-Driven Control (XDDC): The Property-Preserving Framework

Riva,

Formentin

2024

IEEE Control Syst. Lett.

View full text Add to dashboard Cite

A Robust and Explainable Data-Driven Anomaly Detection Approach For Power Electronics

Cited by 4 publications

References 21 publications

Power electronics anomaly detection and diagnosis with machine learning and deep learning methods: A survey

Power electronics anomaly detection and diagnosis with machine learning and deep learning methods: A survey

Data-Driven Controllability of Power Electronics Under Boundary Conditions - A Physics-Informed Neural Network Based Approach

Toward eXplainabile Data-Driven Control (XDDC): The Property-Preserving Framework

Contact Info

Product

Resources

About