[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123493-en":3,"doc-seo-123493-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},123493,1099514068365,"Aurelia","https://ap-avatar.wpscdn.com/avatar/10000253d8d9f28188e?_k=1776742907772140068",8,"Research & Report","Predicting the Shading of Photovoltaic Systems Using Machine Learning","Precise knowledge of shading is crucial for operating photovoltaic power plants, enabling differentiated yield and site analyses as well as reliable monitoring and fault detection. The study presents a workflow for analyzing shading causes using GPT-4o and introduces an approach to predict shading by combining physical modeling with data-driven machine learning. An autoencoder trained on data from 1380 PV devices across diverse shading scenarios predicts shading with accurate results within a few weeks of detection time, while removing erroneous points from the physical model.","40. PV-Symposium 2025 Conference papers  \n[https://doi.org/10.52825/pv-symposium.v2i.2636](https://doi.org/10.52825/pv-symposium.v2i.2636)  \n© Authors. This work is licensed under a Creative Commons Attribution 4.0 International License Published: 27 Aug. 2025  \nPredicting the Shading of Photovoltaic Systems Using Machine Learning  \nMaximilian Schönau 1,2,* , Joseph Jachmann2  , Markus Panhuysen 1  , Alexander Schönau3  , Darwin Daume2  , Achim Schulze4  , Bernd Hüttl2  ,  \nand Dieter Landes2   \n1smartblue AG, Germany  \n2Coburg University of Applied Sciences, Germany  \n3Catholic University of Eichstätt-Ingolstadt, Germany  \n4Rosenheim Technical University of Applied Sciences, Germany  \n*Correspondence: Maximilian Schönau, [Maximilian.Schonau@smartblue.de](Maximilian.Schonau@smartblue.de)  \nAbstract. In the operation of photovoltaic power plants, precise knowledge of shading is essential in order to carry out differentiated yield and site analyses and to guarantee reliable monitoring and fault detection. A study on the causes of shading carried out with the help of GPT4-o is presented. Subsequently, an innovative approach for predicting shading using a is introduced. By combining physical and data-driven machine learning, it is possible to efficiently complete incomplete shadow analyses and eliminate erroneous data points of a physical model. The presented method utilizes data from 1380 photovoltaic devices with various shading scenarios to train an autoencoder on PV system shading. The autoencoder enables accurate prediction of shading within a detection time of only a few weeks.  \nKeywords: Operation and Maintenance (O&M) , Digital Twin, Origins of Shadows  \n1. Introduction  \nWhen monitoring photovoltaic systems, shading is a critical factor that causes an average of around 7 % of power losses [1] . Shading is caused by site-specific conditions, self-shading, and temporary factors such as snow, foliage or vegetation.  \nSite-related and self-shadowing are often accepted to increase the degree of area utilization, yet it is crucial to precisely determine their timing and extent. This enables differentiated yield analyses, allowing operators to evaluate the efficiency and planning of their photovoltaic power plants and to optimize future projects by making adjustments such as changing row spacing or module connections [2] . Temporary shading, such as grass growth, should be monitored so that it can be corrected if necessary. Last but not least , a precise evaluation of shading plays a major role in the fault monitoring of PV power plants. Shading on strings causes short-term power losses, which can lead to false error messages to the operator if these shading periods are not excluded from the error detection [3] .  \nShading of PV systems is an issue, that O&M Managers often addressed in reports. Similar to our previous work [4], these reports were analyzed using GPT4-o by OpenAI [5],[6],[7], to get statistical information about the origin of shading issues.  \nThe dataset consists of 5,089 medium to large-scale ground-mounted and rooftop PV plants constructed between April 1999 and January 2025. About 25 % were built before October 2013, with 75% completed by September 2020. These plants primarily feature traditional silicon module technologies representative of the past decade in Germany. Thin-Film modulesand tracker systems account for less than 0.3% and 3% of installations, respectively [4] .  \nO&M reports were anonymized and then filtered by using keyword identification of the word shading, which resulted in 12487 comments. Most of these comments were filtered out by GPT4-o, as they did not specify the reason of the shading issue. In addition to that, many reports containing the word shading did not specify physical and external shading issues, but problems like snow instead. The filtering process resulted in 530 comments from Asset Managers, who specified the origin of the shading issue of the monitored string.  \nAfter manua","cbCairuPcxgtzpUA","https://ap.wps.com/l/cbCairuPcxgtzpUA","pdf",2498504,1,14,"English","en",105,"# Introduction\n## Shading as a driver of power losses\n## Importance for yield analysis and fault monitoring\n# Dataset and data processing\n## Collection of O&M reports\n## Filtering and anonymization\n## Classification categories for shading origins\n# Methodology for predicting shading\n## Physical and data-driven integration\n## Autoencoder training and prediction","[{\"question\":\"Why is precise shading knowledge important for photovoltaic power plants?\",\"answer\":\"Accurate shading information supports differentiated yield and site analyses and improves reliable monitoring and fault detection by avoiding false error messages caused by shading events.\"},{\"question\":\"How does the study determine the origins of shading?\",\"answer\":\"O\\u0026M reports are anonymized and filtered using keyword identification for “shading,” then GPT-4o is used to classify comments into categories such as Trees, Grass, Self, Buildings, and Others.\"},{\"question\":\"What machine learning approach is used to predict shading?\",\"answer\":\"The method trains an autoencoder on shading data from 1380 PV devices covering varied shading scenarios, enabling prediction of shading with efficient detection time and reducing incorrect data points from a physical model.\"}]","Predicting the Shading of Photovoltaic Systems Using Machine Learning | PDF",1785816837,35,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"predicting-the-shading-of-photovoltaic-systems-using-machine-learning","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/predicting-the-shading-of-photovoltaic-systems-using-machine-learning/123493/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is precise shading knowledge important for photovoltaic power plants?","Question",{"text":75,"@type":76},"Accurate shading information supports differentiated yield and site analyses and improves reliable monitoring and fault detection by avoiding false error messages caused by shading events.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the study determine the origins of shading?",{"text":80,"@type":76},"O&M reports are anonymized and filtered using keyword identification for “shading,” then GPT-4o is used to classify comments into categories such as Trees, Grass, Self, Buildings, and Others.",{"name":82,"@type":73,"acceptedAnswer":83},"What machine learning approach is used to predict shading?",{"text":84,"@type":76},"The method trains an autoencoder on shading data from 1380 PV devices covering varied shading scenarios, enabling prediction of shading with efficient detection time and reducing incorrect data points from a physical model.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]