Luc Dufour

Big data: the key to energy efficiency in smart buildings

Due to the high impact that energy consumption by buildings has at global scale, energy-efficient buildings to reduce

Heating and Hot Water Industrial Prediction System for Residential District

\mathrm{CO}_2

A predictive data-driven model for traffic-jams forecasting in smart santader city-scale testbed

CO2 emissions and energy consumption are needed. In this work we present a novel approach to energy saving in buildings through the identification of the relevant parameters and the application of Soft Computing techniques to generate predictive models of energy consumption in buildings. Using such models it is possible to define strategies for optimizing the day-to-day energy consumption of buildings. To verify the feasibility of this proposal, we apply our approach to a reference building for which we have contextual data from a complete year of monitoring. First, we characterize the building in terms of its contextual features and energy consumption, and then select the most appropriate techniques to generate the most accurate model of our reference building charged with estimating the energy consumption, given a concrete set of inputs. Finally, considering the energy usage profile of the building, we propose specific control actions and strategies to save energy.

Exploiting IoT-based sensed data in smart buildings to model its energy consumption

This work presents a data-intensive solution to predict Photovoltaïque energy (PV) production. PV and other renewable sources have widely spread in recent years. Although those sources provide an environmentally-friendly solution, their integration is a real challenge in terms of power management as it depends on meteorological conditions. The ability to predict those variable sources considering meteorological uncertainty plays a key role in the management of the energy supply needs and reserves. This paper presents an easy-to-use methodology to predict PV production using time series analyses and sampling algorithms. The aim is to provide a forecasting model to set the day-ahead grid electricity need. This information useful for power dispatching plans and grid charge control. The main novelties of our approach is to provide an easy implemented and flexible solution that combines classification algorithms to predict the PV plant efficiency considering weather conditions and nonlinear regression to predict weather forecasted errors in order to improve prediction results. The results are based on the data collected in the Technoples micro grid in Sierre (Switzerland) described further in the paper. The best experimental results have been obtained using hourly historical weather measures (radiation and temperature) and PV production as training inputs and weather forecasted parameters as prediction inputs. Considering a 10 month dataset and despite the presence of 17 missing days, we achieved a Percentage Mean Absolute Deviation (PMAD) of 20% in August and 21% in September. Better results can be obtained with a larger dataset but as more historical data were not available, other months have not been tested.

Test Set Validation for Home Electrical Signal Disaggregation

This work presents a data-intensive solution to predict heating and hot water consumption. The ability to predict locally those flexible sources considering meteorological uncertainty can play a key role in the management of microgrid. A microgrid is a building block of future smart grid, it can be defined as a network of low voltage power generating units, storage devices and loads. The main novelties of our approach is to provide an easy implemented and flexible solution that used a supervised learning techniques. This paper presents an industrial methodology to predict heating and hot water consumption using time series analyzes and tree ensemble algorithm. The results are based on the data collected in a building in Chamoson(Switzerland) and simulations. Considering the winter season 2012-2013 for the training, the heating and hot water predictions is correctly estimated 90% +/- 1.2 for the winter season 2013-2014.

Economic Interest of Heating and Hot Water Prediction System for Residential District

The building energy consumption represent 60% of total primary energy consumption in the world. In order to control the demand response schemes for residential users, it is crucial to be able to predict the different components of the total power consumption of a household. This work provide a non intrusive identification model of devices with a sample frequency of one hertz. The identification results are the inputs of a model to predict the flexible energy. This corresponds at the different devices could be shift in a predetermined time. In a residential building, the heating and the hot water represent this flexible energy. The Support Vector Machine (SVM) enable an identification around 95% of heating, hot water, household electrical and a ensemble of decision tree provide the prediction for the next 15 minutes.

I-BAT: A Data-Intensive Solution Based on the Internet of Things to Predict Energy Behaviors in Microgrids

In this paper, a model for traffic jam prediction using data about traffic, weather and noise is presented. It is based on data coming from a Smart City in Spain called Santander. The project in this city is called ”Smart Santander” and provides a platform for large-scale experiment based on realtime data. This paper demonstrates the possibility of predicting traffic jams and is a basis to integrate in projects to improve the quality of services. In this work, a cross validation method to ratify our training set is proposed. Data intelligence analysis techniques are used for the prediction with an implementation of Neural Network and Decision Tree algorithms. These algorithms are using different parameters coming from Smart Santander and other external sources. Furthermore, a cross validation process is also integrated to improve the final result. The traffic jam prediction for the next 15 minutes reached an accuracy of 99.95%.

Outil de prédiction solaire basé sur un calcul d'erreur météorologique

Due to the high impact that energy consumption of buildings has at global scale, it has been stated the need of achieving energy-efficient buildings to reduce CO2 emissions and energy consumption at global scale. In this work we propose to model the energy consumption associated with services provided in buildings to help select the best strategies to save energy. To verify the feasibility of the proposed approach using measurements of relevant parameters affecting, we carry out some analysis in a reference building of which we have contextual data. Firstly, we provide a complete characterization of this building in term of its energy consumption and generate accurate building models able to predict its energy consumption given a concrete set of inputs. Finally, considering the generated energy usage profile of the building, we propose some concrete control actions and strategies to save energy.