Assessing placement efficiency of photovoltaic installations using Mask R-CNN

Bänziger, Tim Claude

doi:10.21256/zhaw-27103

Please use this identifier to cite or link to this item: https://doi.org/10.21256/zhaw-27103

Publication type:	Master thesis
Title:	Assessing placement efficiency of photovoltaic installations using Mask R-CNN
Authors:	Bänziger, Tim Claude
Advisors / Reviewers:	Gavagnin, Elena Gellrich, Mario
DOI:	10.21256/zhaw-27103
Extent:	62
Issue Date:	2022
Publisher / Ed. Institution:	ZHAW Zürcher Hochschule für Angewandte Wissenschaften
Publisher / Ed. Institution:	Winterthur
Language:	English
Subject (DDC):	005: Computer programming, programs and data 333.79: Energy
Abstract:	Photovoltaic (PV) energy production has experienced strong growth over the past years and is forecasted to greatly contribute to the successful transition to renewable energy production as demanded by Switzerland’s Energy Strategy 2050. Several studies attempted to estimate the national PV potential on building rooftops but arrived at strongly varying results ranging from 15 to 53 TWh annually. To a vast extent, the differences can be explained by the application of varying rooftop utilization ratios which were extrapolated by all previous studies. Moreover, no comparison of the placement of existing PV installations to the suitability categorization from the sonnendach.ch project was yet carried out. Therefore, the aim of this master thesis was to develop and evaluate a prototype methodology to close the research gaps regarding rooftop utilization ratio and the efficiency of PV panel placement. The prototype methodology to answer these questions was developed in Python and leverages publicly available data from the Swiss government in conjunction with a Mask R-CNN for the accurate segmentation of PV panels on high resolution aerial imagery. A total of 1130 individual images of building rooftop were thereby collected in the canton of Aargau of which 974 were used to train the Mask R-CNN model. After four training iterations with varying dataset sizes, the segmentation performance of the Mask R-CNN achieved an iou_score of 0.74. Overall, the rooftop utilization ratio found in this thesis equated to 29%, suggesting that all PV potential studies systematically overestimate the extent of rooftop utilization. Moreover, the findings of this thesis suggest that the more suitable a rooftop area is, the greater its extent of utilization whereas previous studies assumed a uniform distribution of utilization ratio across all suitability categorizations. From the assessed building rooftops, 2.8% have their PV panels suboptimally placed and therefore fail to efficiently exploit solar radiation. 71% of which were successfully detected by the model. Overall, the findings of this thesis proved that an automated, large-scale assessment of PV placement efficiency is technically feasible. This information could support national energy planning as well as PV incentive decision making. However, the segmentation performance of the Mask R-CNN achieved with the resources available to this thesis is currently insufficient for detailed quantitative analyses. Consequently, further studies to improve the Mask R-CNN performance should be conducted before applying the prototype methodology on a large scale.
URI:	https://digitalcollection.zhaw.ch/handle/11475/27103
License (according to publishing contract):	CC BY-NC-ND 4.0: Attribution - Non commercial - No derivatives 4.0 International
Departement:	School of Management and Law
Appears in collections:	MSc Wirtschaftsinformatik

Files in This Item:

File	Description	Size	Format
2022_Tim_Baenziger_MSc_WI.pdf		2.5 MB	Adobe PDF	View/Open

Show full item record

Bänziger, T. C. (2022). Assessing placement efficiency of photovoltaic installations using Mask R-CNN [Master’s thesis, ZHAW Zürcher Hochschule für Angewandte Wissenschaften]. https://doi.org/10.21256/zhaw-27103

Bänziger, T.C. (2022) Assessing placement efficiency of photovoltaic installations using Mask R-CNN. Master’s thesis. ZHAW Zürcher Hochschule für Angewandte Wissenschaften. Available at: https://doi.org/10.21256/zhaw-27103.

T. C. Bänziger, “Assessing placement efficiency of photovoltaic installations using Mask R-CNN,” Master’s thesis, ZHAW Zürcher Hochschule für Angewandte Wissenschaften, Winterthur, 2022. doi: 10.21256/zhaw-27103.

BÄNZIGER, Tim Claude, 2022. Assessing placement efficiency of photovoltaic installations using Mask R-CNN. Master’s thesis. Winterthur: ZHAW Zürcher Hochschule für Angewandte Wissenschaften

Bänziger, Tim Claude. 2022. “Assessing Placement Efficiency of Photovoltaic Installations Using Mask R-CNN.” Master’s thesis, Winterthur: ZHAW Zürcher Hochschule für Angewandte Wissenschaften. https://doi.org/10.21256/zhaw-27103.

Bänziger, Tim Claude. Assessing Placement Efficiency of Photovoltaic Installations Using Mask R-CNN. ZHAW Zürcher Hochschule für Angewandte Wissenschaften, 2022, https://doi.org/10.21256/zhaw-27103.