Data-Driven Energy Optimization in Smart Social Housing: Weather-Aware PV Performance and Load Shifting Strategies in Planet Smart City = Data-Driven Energy Optimization in Smart Social Housing: Weather-Aware PV Performance and Load Shifting Strategies in Planet Smart City

In recent years, cities have accelerated their transition toward becoming smart by adopting digital tools to enhance connectivity, urban mobility, education and public safety. To guide this transformation various institutions and regulations have been established. Within this landscape, the Planet Smart City company holds the project with the title of Digital Districts for Flexible Energy Services (D2FX). Represents innovative model that integrates affordable housing with smart technologies and digital energy management platforms by generating smart solutions in smart cities.The core data storage, analysis, and behavioral modeling algorithm resides in the EoT software platform customized and further developed to achieve the project's stated goals.The resident-facing component of the solution is a mobile application that delivers the behavioral nudges to achieve the desired demand profiles. This functionality will be delivered within the context of a comprehensive resident application that offers a suite of services, thus encouraging the download and use of the tool daily. This thesis investigates energy efficiency and digital optimization in this environment by analyzing real-world energy and environmental data, identifying PV generation and energy flow patterns and simulating strategies to enhance self-consumption. It compare rule-based optimization with AI-driven scheduling models to assess which approach offers greater benefits. The thesis analysis focuses on the Diana house type, which includes PV panels but no mechanical cooling, relying instead on natural ventilation. Empirical data is sourced from Casa 220, which is a PV system monitored via the GoodWe portal, while Casa 163 is used to simulate conditions in a typical unmonitored home. The study begins with the analysis of real-world data collected from smart meters and the GoodWe monitoring portal. These sources provide detailed hourly information on PV generation, household consumption, in-house usage, energy sold to and purchased from the grid and income derived from solar energy. On the other hand, environmental data such as temperature, humidity, cloud cover, and global tilted irradiance (GTI) was collected from Open-Meteo to evaluate system performance. Our methodology involves integrating and cleaning this data to enable comprehensive energy flow analysis. One of the first points is analyzing the performance of the PV panels installed on the rooftops of Casa 220. By cross-referencing the actual PV output with GTI and weather conditions, the research evaluates system efficiency and identifies discrepancies caused by environmental factors, such as cloudiness or excessive heat.To assess the performance of the rooftop PV system in real-world conditions, we conducted a detailed comparison between hourly PV generation and Global Tilted Irradiance (GTI) using Power BI. By evaluating the ratio between the two, we could calculate the hourly PV system efficiency. Moreover, we introduced several classification flags, such as efficiency category, cloud cover type, GTI level, and rainy day markers to categorize days and conditions under which the system performed well or poorly. These flags were critical for correlating environmental conditions with generation outcomes and helped explain why. This comprehensive approach allowed us not only to evaluate PV system behavior under various conditions but also to use these insights in designing adaptive consumption strategies.