An Integrated Approach Improved Fast S-transform and SVD Noise Reduction for Classification of Power Quality Disruptions in Noisy Environments. Issue 14 (14th September 2022)

• Record Type: Journal Article
• Authors: Goh, Hui Hwang; Liao, Ling; Zhang, Dongdong; Dai, Wei; Lim, Chee Shen; Kurniawan, Tonni Agustiono; Goh, Kai Chen
• Is Part Of: Electric power components and systems. Volume 50:Issue 14/15(2022)
• Page Start: 868
• Page End: 885
• Publication Date: 2022-09-14
• Subjects: improved fast S-transform -- noise reduction -- power quality disturbances -- singular value decomposition; Electric machinery -- Periodicals; 621.3104205
• DOI: 10.1080/15325008.2022.2141928
• Languages: English
• ISSNs: 1532-5008

Abstract

Abstract: Degraded power quality (PQ) significantly jeopardizes the safety, economics, and efficiency of electricity users. Effective power quality disturbance (PQD) classification is critical for power quality control. To address the issue of noise obscuring the time-frequency domain characteristics of PQD extraction, this research introduces a novel method based on singular value decomposition (SVD) and an improved fast S-transform (IFST). To begin, the disturbance signal is noise reduced using SVD to produce a denoised signal. This denoised signal is then processed using differential sum to generate the exact feature F1, which is used to distinguish stationary from nonstationary disturbances. Additionally, the denoised signal is subjected to IFST to extract features F2–F5. Finally, the five most effective features are fed into a simple ruled decision tree (DT) to automate the classification of disturbances, which includes seven single PQDs and six complex PQDs. The utility of the proposed technique was proven in this research by utilizing both simulated disturbances under varied noise conditions and real data. In comparison to established techniques, the unique method outperforms them in terms of anti-noise performance, allowing for more precise classification of varied types of disturbances, and the extracted specific features can intuitively reflect the dynamic changes in disturbances.