Article type: Research Article
Authors: Senthil Kumar, V.a; * | Aruna, R.b | Varalatchoumy, M.c | Manikannan, P.d | Santhana Krishnan, T.e | Usha Rani, B.f | Kumar, Ashokg | Rajaram, A.h
Affiliations:
[a] Department of Computer Science and Engineering, Faculty of Engineering and Technology, SRM Institute of Science and Technology, Tiruchirappalli, Tamil Nadu, India
|
[b] Department of Electronics and Communication Engineering, AMC Engineering College, Bengaluru, Karnataka, India
|
[c] Department of Computer Science and Engineering, Cambridge Institute of Technology, Bengaluru, Karnataka, India
|
[d] Department of Electrical and Electronics Engineering, A.K.T Memorial College of Engineering and Technology, Kallakurichi, Tamil Nadu, India
|
[e] Department of Electrical and Electronics Engineering, Rajalakshmi Engineering College (Autonomous), Chennai, Tamil Nadu, India
|
[f] Department of Electronics and Communication, AMC Engineering College, Bangalore, Karnataka, India
|
[g] Department of Computer Science, Banasthali Vidyapith, Rajasthan, India
|
[h] Department of Electronics and Communication Engineering, E.G.S Pillay Engineering College, Nagapattinam, India
Correspondence:
[*]
Corresponding author. V. Senthil Kumar, Department of Computer Science and Engineering, Faculty of Engineering and Technology, SRM Institute of Science and Technology, Tiruchirappalli, Tamil Nadu 621105, India. E-mail: [email protected].
Abstract: As the world embraces the transition towards renewable energy, the optimization of solar power plants becomes paramount. In this research, we present a comprehensive framework that leverages advanced analytical methodologies to address critical operational challenges and elevate the efficiency of solar power generation. Our framework encompasses data preprocessing, time series analysis, anomaly detection, and equipment performance assessment, synergistically combining their strengths to offer a holistic solution. The heart of our proposed approach lies in the precision and efficacy of anomaly detection. We introduce two powerful techniques—LSTM Autoencoder and Isolation Forest—to identify anomalies and equipment underperformance. Through meticulous evaluation, we showcase their comparative performance, revealing the nuanced strengths of each. Visualizations depict the model’s proficiency in pinpointing anomalies, with LSTM Autoencoder emerging as a standout performer, adept at capturing even subtle deviations from expected patterns. Our research extends beyond detection to equip stakeholders with real-time insights. The visualization of daily yield trends uncovers potential data anomalies, enabling timely intervention and rectification. Additionally, we address equipment failures by harnessing random forest modeling to establish a robust relationship between irradiance, temperature, and DC power. This approach provides a powerful tool for real-time condition monitoring and fault detection, enabling proactive maintenance and enhancing operational resilience.
DOI: 10.3233/JIFS-235578
Journal: Journal of Intelligent & Fuzzy Systems, vol. 46, no. 2, pp. 4995-5011, 2024
Published: 14 February 2024
