Integration of asset prognostics and maintenance optimization for solar power plants

Şakir KARAKAYA, Ministry of Industry and Technology

Abstract

This talk includes two interlinked studies that utilize sensor-based degradation analytics for asset prognostics and operations management in energy systems. The first study proposes a diagnostic model that can harness sensor data for detecting the state-of-health of PV inverters, which are significant contributors to reliability risks in PV systems. The model (i) transforms functional sensor data from PV inverters to time-frequency domain features in an effort to capture both summary statistics and signal dynamics, and (ii) uses the produced signal features to build a diagnostic model that predicts degradation severity in PV inverters based on multinomial logistic regression. The results using inverter data from an accelerated life testing experiment show that the proposed model offers 91% accuracy in predicting degradation severity that can be used for determining individual maintenance schedules for inverters. The second study shows how the output of the diagnostic model can be utilized for asset prognostics and how the maintenance and operations planning process can be improved in energy systems by integrating prognostics and optimization models. In this context, a maintenance and operations planning model for PV solar plants, which aims at making optimal maintenance crew routing and, preventive and corrective maintenance decisions, is developed to maximize total profit. The model harnesses sensor-driven data reflecting the degradation in each unit’s performance, including PV inverters, arrays, and transformers. It is characterized as a two-stage stochastic program in which energy generation and unit degradation are handled as uncertain. The experimental results show that the proposed model outperforms the traditional periodic maintenance policy in terms of optimum solution values obtained.

Şakir Karakaya received his Ph.D. (2018) and MSc (2008) degrees in the Industrial Engineering Department from the Middle East Technical University. He worked as a postdoctoral fellow at Wayne State University (Department of Systems and Industrial Engineering) in 2022-2023. He also completed an MSc program on Public Management and Governance at the London School of Economics and Political Science (LSE) in 2016. He has been working for the Ministry of Industry and Technology since 2004 and currently is the Head of Market Surveillance and Product Safety Department. His research interests are quality and product management, sensor-driven operations and maintenance management, and public management.

Venue

Friday, October 6, 2023, 4.00 pm - IE03(Halim Doğrusöz Auditorium)