Thirty convenience stores in College Station, Texas, have been selected as the samples for an energy consumption prediction. The predicted models assist facility energy managers for making decisions of energy demand/supply plans. The models are applied to historical data for two years: 2001 and 2002. The approaches are (1) to analyze nonlinear regression models for long term forecasting of annual patterns compared with outdoor temperature, and (2) to analyze multiple regression models for the building type regardless of outdoor temperature. In the first approach, twenty four buildings are categorized as base load group and no base group. Average temperature, cooling efficiencies, and cooling knot temperature are estimated by nonlinear regression models: segment and parabola models. The adjusted r-square results in good performance up to ninety percent accuracy. In the second approach, the other selected six buildings are categorized as no trend group. This group does not respond to outdoor temperature. As the result, multiple a regression model is formed by combination of variables from the nonlinear models and physical building variables of cooling efficiency, cooling temperature, light bulbs, area, outdoor temperature, and orientation of fronts. This model explains up to sixty percent of all convenience stores' data. In conclusion, the accuracy of prediction models is measured by the adjusted r-square results. Among these three models, the multiple regression model shows the highest adjusted r-square (0.597) over the parabola (0.5419) and segment models (0.4806). When the three models come to the application, the multiple regression model is best fit for no trend data type. However, when it is used to predict the energy consumption with the buildings that relate to outdoor temperature, segment and parabola model provide a better prediction result.

This study proposes a data-driven model for prediction of the energy consumption of fleet vehicles in various missions, by characterization as the linear combination of a small set of exemplar travel segments. The model was constructed with reference to a heterogenous study group of 29 light municipal fleet vehicles, each performing a single mission, and each equipped with a commercial OBD2/GPS logger. The logger data was cleaned and segmented into 3-minute periods, each with 10 derived kinetic features and a power feature. These segments were used to define three essential model components as follows: The segments were clustered into six exemplar travel types (called "eigentrips" for brevity) Each vehicle was defined by a vector of its average power in each eigentrip Each mission was defined by a vector of annual seconds spent in each eigentrip 10% of the eigentrip-labelled segments were selected into a training corpus (representing historical observations), with the remainder held back for testing (representing future operations to be predicted). A Light Gradient Boost Machine (LGBM) classifier was trained to predict the eigentrip labels with sole reference to the kinetic features, i.e., excluding the power observation. The classifier was applied to the held-back test data, and the vehicle's characteristic power values applied, resulting in an energy consumption prediction for each test segment. The predictions were then summed for each whole-study mission profile, and compared to the logger-derived estimate of actual energy consumption, exhibiting a mean absolute error of 9.4%. To show the technique's predictive value, this was compared to prediction with published L/100km figures, which had an error of 22%. To show the level of avoidable error, it was compared with an LGBM direct regression model (distinct from the LGBM classifier) which reduced prediction error to 3.7%. / Graduate

No description available.

In this thesis, we develop an approach to characterizing the uncertainty in energy use toward development of a real-time intelligent energy monitoring system for an autonomous under- water vehicle (AUV). The purpose of the intelligent energy monitoring system is to estimate current energy onboard the AUV, estimate energy needed to complete a desired mission, and to determine if and when the AUV should terminate the current mission and return to the recovery location due low energy reserves. In this work, we examine the relationship between water currents and energy used by the AUV, and we specifically address ways to characterize the relationship between uncertainty in water currents and uncertainty in energy use. We also examine the development of a battery model for the AUV, and test this model under simulated and real world conditions. We also develop a model for predicting future energy states, and evaluate this model using real world trials. / Master of Science / In this thesis, we develop an approach to characterizing the uncertainty in energy use for an energy monitoring system for an autonomous underwater vehicle (AUV). The purpose of the energy monitoring system is to estimate current energy onboard the AUV, estimate energy needed to complete a desired mission, and to determine if and when the AUV should cancel the mission and return to the recovery location due low energy levels. In this work, we examine the relationship between water currents and energy used by the AUV, and we specifically address ways to characterize the relationship between uncertainty in water currents and uncertainty in energy use. We also examine the development of a battery model for the AUV, and test this model under simulated and real world conditions, and develop a model for predicting future energy levels.

Comment satisfaire les besoins en énergie d’une population de 9 milliards d’êtres humains en 2050, de façon économiquement viable tout en minimisant l’impact sur l’environnement. Une des réponses est l’insertion de production d’énergie propre d’origine éolienne et photovoltaïque mais leurs totales dépendances aux variations climatiques accentuent une pression sur le réseau. Les modèles prédictifs historiques centralisés et paramétriques ont du mal à appréhender les variations brutales de productions et de consommations. La révolution internet permet aujourd’hui une convergence entre le numérique et l’énergie. En Europe et depuis cinq ans, l’axe d’étude est celui de la maîtrise locale de l’électricité. Ainsi plusieurs quartiers intelligents ont été créés et les modèles utilisés de pilotage et de prédiction restent souvent la propriété des partenaires des projets. Dans cette thèse, Il s’agit de réaliser un bilan énergétique chaque heure pour prédire l’ensemble des vecteurs énergétiques d’un système. Le besoin en énergie d’un système comme une maison est décomposée en un besoin en chauffage, en un besoin en eau chaude sanitaire, en un besoin en luminaires, en besoin de ventilation et en usages spécifiques électriques utiles. Le système peut posséder une production décentralisée et un système de stockage ce qui augmentera sa capacité d’effacement. Pour le centre de pilotage, l’objectif est d’avoir une possibilité de scénarios de surproductions ou surconsommations sur un quartier donnée à court terme. Nous considérerons dans cette thèse un horizon à l’heure pour notre bilan énergétique. Cela implique une prédiction fine des différents flux énergétiques d’un système en particulier le chauffage et l’eau chaude qui représente le plus gros potentiel de flexibilité dans les bâtiments. Pour réaliser un bilan, nous devons calculer les différents flux énergétiques à l’intérieur de notre système : les déperditions par l’enveloppe et la ventilation, les gains internes solaires, des personnes et des appareils, le stockage, la production d’eau chaude sanitaire, les usages spécifiques électriques utiles. Sur certains de ces points, nous pouvons évaluer assez précisément et en fonction du temps les quantités d’énergie échangées. Pour les autres (ECS, USE, gains internes, stockage), la bibliographie nous donne que des méthodes globales et indépendantes du temps. Il n’est donc pas possible d’envisager une méthode correspondant au pas de temps souhaité. Ceci impose la mise au point d’une méthode prédictive et apprenante dont nos modèles de simulation énergétique seront le point de référence. Il n’en reste pas moins que ces modèles permettent la compréhension du comportement énergétique du système. L’outil se devra non intrusif, personnalisé, robuste et simple. Pour limiter le caractère intrusif de l’outil, il s’agit à la fois d’ajouter de l’intelligence comme par exemple l’identification des appareils utiles à partir d’un seul point de mesure mais aussi la collection et l’analyse d’informations localement. Les données privées ne sont pas transmises vers l’extérieur. Seules les informations de prédictions énergétiques sont envoyées à un niveau supérieur pour agrégation des données des quartiers. L’intelligence est également au niveau des prédictions réalisées issues de méthodes d’apprentissage comme l’utilisation des réseaux de neurones ou des arbres de décision. La robustesse est étudiée d’un point de vue technologie (plusieurs protocoles de communication ont été testés), techniques (plusieurs méthodes de collecte) et d’un point de vue du stockage de données (limiter la fréquence de collecte). La simplicité d’usage engendre une simplicité d’installation minimiser le nombre de données d’entrée tout en gardant une précision souhaitable sera notre principal axe d’optimisation. / This work presents a data-intensive solution to manage energy flux after a low transformer voltage named microgrid concept. A microgrid is an aggregation of building with a decentralized energy production and or not a storage system. These microgrid can be aggregate to create an intelligent virtual power plant. However, many problems must be resolved to increase the part of these microgrid and the renewable resource in a energy mix. The physic model can not integrate and resolve in a short time the quickly variations. The intelligent district can be integrate a part of flexibility in their production with a storage system. This storage can be electrical with a battery or thermal with the heating and the hot water. For a virtual power plant, the system can be autonomous when the price electricity prediction is low and increase the production provided on the market when the price electricity is high. For a energy supplier and with a decentralized production building distant of a low transformer voltage, a regulation with a storage capacity enable a tension regulation. Finally, the auto-consumption becomes more and more interesting combined with a low electrical storage price and the result of the COP 21 in Paris engage the different country towards the energy transition. In these cases, a flexibility is crucial at the building level but this flexibility is possible if, and only if, the locally prediction are correct to manage the energy. The main novelties of our approach is to provide an easy implemented and flexible solution to predict the consumption and the production at the building level based on the machine learning technique and tested on the real use cases in a residential and tertiary sector. A new evaluation of the consumption is realized: the point of view is energy and not only electrical. The energy consumption is decomposed between the heating consumption, the hot water consumption and the electrical devices consumption. A prediction every hour is provided for the heating and the hot water consumption to estimate the thermal storage capacity. A characterization of Electrical devices consumption is realized by a non-intrusive disaggregation from the global load curve. The heating and the hot water are identify to provide a non intrusive methodology of prediction. Every day, the heating, the hot water, the household appliances, the cooling and the stand by are identified. Every 15 minutes, our software provide a hot water prediction, a heating prediction, a decentralized prediction and a characterization of the electrical consumption. A comparison with the different physic model simulated enable an error evaluation the error of our different implemented model.

abstract: Commercial buildings in the United States account for 19% of the total energy consumption annually. Commercial Building Energy Consumption Survey (CBECS), which serves as the benchmark for all the commercial buildings provides critical input for EnergyStar models. Smart energy management technologies, sensors, innovative demand response programs, and updated versions of certification programs elevate the opportunity to mitigate energy-related problems (blackouts and overproduction) and guides energy managers to optimize the consumption characteristics. With increasing advancements in technologies relying on the ‘Big Data,' codes and certification programs such as the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE), and the Leadership in Energy and Environmental Design (LEED) evaluates during the pre-construction phase. It is mostly carried out with the assumed quantitative and qualitative values calculated from energy models such as Energy Plus and E-quest. However, the energy consumption analysis through Knowledge Discovery in Databases (KDD) is not commonly used by energy managers to perform complete implementation, causing the need for better energy analytic framework. The dissertation utilizes Interval Data (ID) and establishes three different frameworks to identify electricity losses, predict electricity consumption and detect anomalies using data mining, deep learning, and mathematical models. The process of energy analytics integrates with the computational science and contributes to several objectives which are to 1. Develop a framework to identify both technical and non-technical losses using clustering and semi-supervised learning techniques. 2. Develop an integrated framework to predict electricity consumption using wavelet based data transformation model and deep learning algorithms. 3. Develop a framework to detect anomalies using ensemble empirical mode decomposition and isolation forest algorithms. With a thorough research background, the first phase details on performing data analytics on the demand-supply database to determine the potential energy loss reduction potentials. Data preprocessing and electricity prediction framework in the second phase integrates mathematical models and deep learning algorithms to accurately predict consumption. The third phase employs data decomposition model and data mining techniques to detect the anomalies of institutional buildings. / Dissertation/Thesis / Doctoral Dissertation Civil, Environmental and Sustainable Engineering 2017

No description available.

<p>Electric load forecasting is a critical tool for power system planning and the creation of sustainable energy systems. Precise and reliable load forecasting enables power system operators to make informed decisions regarding power generation and transmission, optimize energy efficiency, and reduce operational costs and extra power generation costs, to further reduce environment-related issues. However, achieving desirable forecasting performance remains challenging due to the irregular, nonstationary, nonlinear, and noisy nature of the observed data under unprecedented events. In recent years, deep learning and other artificial intelligence techniques have emerged as promising approaches for load forecasting. These techniques have the ability to capture complex patterns and relationships in the data and adapt to changing conditions, thereby enhancing forecasting accuracy. As such, the use of deep learning and other artificial intelligence techniques in load forecasting has become an increasingly popular research topic in the field of power systems. </p> <p>Although deep learning techniques have advanced load forecasting, the field still requires more accurate and efficient models. One promising approach is federated learning, which allows for distributed data analysis without exchanging data among multiple devices or cen- ters. This method is particularly relevant for load forecasting, where each power station’s data is sensitive and must be protected. In this study, a proposed approach utilizing Federated Deeponet for seven different power stations is introduced, which proposes a Federated Deep Operator Networks and a Lagevin Dynamics-based Federated Deep Operator Networks using Stochastic Gradient Langevin Dynamics as the optimizer for training the data daily for one-day and predicting for one-day frequencies by frequencies. The data evaluation methods include mean absolute percentage error and the percentage coverage under confidence interval. The findings demonstrate the potential of federated learning for secure and precise load forecasting, while also highlighting the challenges and opportunities of implementing this approach in real-world scenarios. </p>

No description available.

No description available.