Periodic Reporting for period 2 - PRECEPT (A novel decentralized edge-enabled PREsCriptivE and ProacTive framework for increased energy efficiency and well-being in residential buildings)
Berichtszeitraum: 2021-10-01 bis 2023-03-31
As the severity of environmental issues continuously increases, the need to reduce overall energy consumption becomes imperative. The focus is more fixated on the building sector, where the largest potential for cost-efficient energy savings has been identified. PRECEPT ambitiously aims to enable the “smooth” and near-zero operational and maintenance costs transformation of conventional residential buildings into highly efficient proactive ones. The transformation of conventional reactive residential buildings into proactive ones leads to increased energy efficiency, lower energy consumption, and optimal levels of comfort while improving the residents’ well-being, as optimal levels of temperature, humidity, moisture, air quality, and lighting are constantly maintained in a building. Moreover, PRECEPT aims for increased operational efficiency achieved through asset management, as well as predictive maintenance which leads to lower maintenance costs, less equipment downtime, fewer breakdowns, and conversely increase in the time between maintenance cycles and the overall life of the equipment.
PRECEPT strives to achieve its ambitious objective by introducing a Pred(scr)ictive and Proactive building energy management system (PP-BMS) installed locally at the building level, able to rapidly and efficiently self-manage, self-monitor, self-heal and self-optimize itself, with minimal human intervention, whenever there are system malfunctions, significant deviations in the forecasted weather conditions affecting both thermal and visual comfort as well as RES generation, occupancy schedules, flexibility forecasts, energy tariffs, loads-scheduling, and level of utilization. PRECEPT aims to provide novel techniques for supporting the collection, analysis and visualization of the expected large volume of data. Furthermore, PRECEPT will deliver and integrate building digital twins that will be continuously updated with real-life data for early prediction, automatic repair, and optimal building operation. PRECEPT also realizes the importance of user communication and user communities. To this end, PRECEPT plans to develop a social collaboration platform where users may exchange information, experiences, and best practices. PRECEPT will also introduce standardized procedures for data management and building proactiveness that will contribute to the development of the required Smart Proactiveness Indicators (SPIs) of the building. PRECEPT also aims to deliver AI algorithms dedicated to prescriptive maintenance exploiting historical building information, as well as local climatic data, and creating appliance/device behavioural models. Lastly, PRECEPT will introduce, implement, and demonstrate novel sustainable business models for transforming traditional reactive buildings into proactive buildings.
PRECEPT strives to achieve its ambitious objective by introducing a Pred(scr)ictive and Proactive building energy management system (PP-BMS) installed locally at the building level, able to rapidly and efficiently self-manage, self-monitor, self-heal and self-optimize itself, with minimal human intervention, whenever there are system malfunctions, significant deviations in the forecasted weather conditions affecting both thermal and visual comfort as well as RES generation, occupancy schedules, flexibility forecasts, energy tariffs, loads-scheduling, and level of utilization. PRECEPT aims to provide novel techniques for supporting the collection, analysis and visualization of the expected large volume of data. Furthermore, PRECEPT will deliver and integrate building digital twins that will be continuously updated with real-life data for early prediction, automatic repair, and optimal building operation. PRECEPT also realizes the importance of user communication and user communities. To this end, PRECEPT plans to develop a social collaboration platform where users may exchange information, experiences, and best practices. PRECEPT will also introduce standardized procedures for data management and building proactiveness that will contribute to the development of the required Smart Proactiveness Indicators (SPIs) of the building. PRECEPT also aims to deliver AI algorithms dedicated to prescriptive maintenance exploiting historical building information, as well as local climatic data, and creating appliance/device behavioural models. Lastly, PRECEPT will introduce, implement, and demonstrate novel sustainable business models for transforming traditional reactive buildings into proactive buildings.
The PRECEPT activities began by laying the groundwork for the successful launch and development of the PRECEPT project. The core management procedures and tools as well as the quality assessment and data management plan were developed to ensure effective monitoring and high-quality results. An occupant-centred approach was designed and followed to define stakeholder categories and the limiting factors for the wide adoption of smart and proactive building technologies. Key performance indicators (KPIs) were defined, and smart readiness indicators (SRIs) were calculated for each of the PRECEPT pilots while existing European and national legislation was examined without losing focus on existing and upcoming standards. At the same time, a new set of key performance indicators for smart proactive buildings, the Smart Proactive Indicators (SPIs), and the methodology by which they are to be calculated were developed and calculated for each of the PRECEPT pilot sites. The conceptual architecture of PRECEPT was designed and continuously expanded upon according to the requirements of components and subcomponents and their interconnections. A novel method to generate a simplified Building Energy Model (BEM) was developed and Building Information Models (BIM) have been constructed for all pilot sites. The PRECEPT Digital Twin Platform was developed and deployed, allowing the construction of digital twins by any open standard. The digital twins of the PRECEPT pilot sites have also been enriched with Internet of Things (IoT) data and energy models.
Furthermore, the PRECEPT ecosystem and its infrastructure were successfully deployed in all pilot sites after continuous communication with the pilot tenants. The innovative PRECEPT Pred(scr)ictive and Proactive building energy management system (PP-BMS) was developed and its integration with the digital twin platform and other system components was defined. Its user recommendation system has also been built by exploiting a state-of-the-art overview of all controllable loads. Novel methodologies utilizing machine-learning techniques were explored to achieve accurate and reliable disaggregation of energy consumption at the appliance level. Monitoring data from real-life test cases in combination with other data sources such as weather APIs were leveraged and provided valuable insights into occupants’ behavioural, energy consumption and comfort patterns while state-of-the-art research was examined to predict human behaviour accurately. Disaggregation, human comfort, activity and occupancy models as well as the PRECEPT Proactive Building Operation module and the PRECEPT Proactive RES & Storage management modules were developed and validated. Additionally, innovative advancements in anomaly detection, human behaviour and comfort were made and integrated into the PRECEPT inference engine. At the same time, PRECEPT has cultivated a robust online presence and participated actively in physical events like conferences, trade fairs and workshops and continuously explores novel business models for the exploitation of the project’s results.
Furthermore, the PRECEPT ecosystem and its infrastructure were successfully deployed in all pilot sites after continuous communication with the pilot tenants. The innovative PRECEPT Pred(scr)ictive and Proactive building energy management system (PP-BMS) was developed and its integration with the digital twin platform and other system components was defined. Its user recommendation system has also been built by exploiting a state-of-the-art overview of all controllable loads. Novel methodologies utilizing machine-learning techniques were explored to achieve accurate and reliable disaggregation of energy consumption at the appliance level. Monitoring data from real-life test cases in combination with other data sources such as weather APIs were leveraged and provided valuable insights into occupants’ behavioural, energy consumption and comfort patterns while state-of-the-art research was examined to predict human behaviour accurately. Disaggregation, human comfort, activity and occupancy models as well as the PRECEPT Proactive Building Operation module and the PRECEPT Proactive RES & Storage management modules were developed and validated. Additionally, innovative advancements in anomaly detection, human behaviour and comfort were made and integrated into the PRECEPT inference engine. At the same time, PRECEPT has cultivated a robust online presence and participated actively in physical events like conferences, trade fairs and workshops and continuously explores novel business models for the exploitation of the project’s results.
Within the work toward achieving PRECEPT’s vision, there have already been significant achievements beyond state-of-the-art. A new set of key performance indicators for smart proactive buildings, the Smart Proactive Indicators (SPIs) has been developed able to provide valuable insights into the proactive potential of a building. A completely new self-adapted, -learned, -managed, -monitored, -healing and -optimized Pred(scr)ictive and Proactive building energy management system (PP-BMS) integrating with all other PRECEPT components, has been developed. Advanced federated learning (FL) frameworks were developed to test disaggregation and load forecasting models and integrated with the FogFlow framework, enabling distributed machine learning with data privacy and security preservation. Novel methodologies utilizing machine-learning techniques were explored to achieve accurate and reliable disaggregation of energy consumption at the appliance level and define thermal and visual comfort levels. Α novel approach to generate a simplified BEM was created to achieve minimal user disturbance while the digital twin platform, able to convert any open standard was also developed. Leveraging the findings of exhaustive research of the state-of-the-art, a testing framework for anomaly detection and predictive maintenance was designed. Finally, machine learning models for adaptive flexibility estimation were tested and evaluated considering factors such as user behaviour, appliance usage patterns, and external influences.