Presentation Details
| Methodologies for Imputing Missing Data for Photovoltaic Performance Loss Estimation Nikola Hrelja1, Loic Guillemot2, Lluvia Ochoa3. 1TotalEnergies OneTech, Palaiseau, France.2TotalEnergies OneTech, Palaiseau, France.3TotalEnergies OneTech, Palaiseau, France |
Abstract
To assess the profitability of investing in photovoltaics, it is essential to accurately measure performance losses and predict the lifespan of the system. One significant challenge in photovoltaic monitoring is data loss, which can result from communication problems, maintenance interruptions, or inverter failures. This study presents a comparative analysis of multiple imputation methodologies applied to synthetically generated degraded PV performance data, with the objective of assessing their effects on degradation rate estimation under varying degrees of induced missing data. One of the challenges of data imputation is how to accurately capture the degradation trend in the missing data. Although numerous methods are available to evaluate performance loss rate, their sensitivity to different imputation methodologies on synthetic photovoltaic degraded data has not been considered. In this paper, we evaluate the impact of missing data on two different degradation rate estimation approaches: namely, the year-on-year and linear regression methods on synthetic data with known degradation rates. Missing data were introduced at different levels, and three imputation methods were applied before estimating the annual degradation rates. Results show that the estimated degradation rates are strongly affected by the added missing data and applied imputation techniques where different methods perform differently. Regardless of the imputation strategy, the year-on-year method demonstrates the most robust results and should be preferred over linear regression. The linear regression method is more sensitive to missing data and results can be improved using imputation methods.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
Content Locked. Log into a registered attendee account to access this presentation.