ENBIS-15 Energy Day10 September 2015; 09:00 – 17:00; The Czech Association of Scientific and Technical Societies, Prague (Czech Republic)
This special ENBIS-15 post-conference event will focus on modeling of smart grids as a challenge for stochastics and optimization.
Complex technical systems are influenced by internal characteristics, but may also depend on external (stochastic) factors and evolve over the time and/or space. The main goal of this workshop is to introduce advanced methods that are crucial for complete understanding of certain classes of complex systems, especially electricity production and distribution networks, and to use them to answer challenging questions in focused applications.
The modern energy distribution networks provided by communicative gauges (so-called smart grids) allow us to handle more information than whenever in the past. Due to the advances in capturing technologies, the complexity of the data structures is often involved: different types of data, several sources of information of different precision of the measurements on the same (sub)system, few measurements on a large number of characteristics, etc. Although diverse information is to be applauded for, extraction of the important information from these complex data becomes a real challenge.
Thus, the analysis of energy production and distribution networks based on advanced data structures faces important challenges for research, and can be translated in the following main objectives:
- How to model and optimize behavior of complex systems based on advanced data structures.
- How to select from a large, very large, or even huge set of measured characteristics of possibly different nature and preciseness those which influence crucially variable(s) of interest, and how to make short-term predictions based on such data.
- How to model changes in a model structure and/or trend of observed processes.
Electricity Consumption Prediction with Functional Linear Regression (Jaromir Antoch, Charles University in Prague, Czech Republic)
A functional linear regression model linking observations of a functional response variable with measurements of an explanatory functional variable will be considered. This model serves to analyse a real data set describing electricity consumption in Sardinia during the years 2000 - 2006. Data consists of 52,584 entries, each entry summarizing the global electricity consumption of Sardinia for one hour. The interest lies in predicting either oncoming weekends’ or oncoming weekdays’ consumption, provided actual weekdays’ consumption is known. Selected computational issues will be addressed as well. Our main objective lies in using functional regression methods to give feasible solutions for the practitioner.
Modeling Dynamics and Short Term Prediction of Complex Processes (Marek Brabec, Academy of Sciences of Czech Republic)
Modern semi parametric statistical modeling is a powerful tool useful for description of complex and nonlinear systems like those arising in energy-related fields. The approach has been successfully applied in study of many dynamic and spatial-temporal phenomena, both for theoretical and practical reasons. In particular, structured models derived e.g. as GAM (generalized additive models) have been used for short term predictors which have to respect given “physical” constraints (common in energy production and transport).
Optimization Based on Mixed Integer Nonlinear Programming (MINLP) Methods (Claudia D’Ambrosio, CNRS LIX, Ecole Polytechnique, France)
Mixed-integer nonlinear programming is a powerful modeling paradigm that has been employed by engineers, economists, and operations researchers to model a wealth of decision-making applications that involve discrete decisions and nonlinearities. Participants are expected to discuss challenges and opportunities in the field of mixed-integer nonlinear optimization, as well as to review progress made on algorithmic, computational, and applications aspects.
Round Table Discussions about Real Problems and Challenges (Alessandro Di Bucchianico, Eindhoven University of Technology, The Netherlands)
Guidance will be provided as to what kind of modelling strategy is best in different situations. Discussion will be both about the advantages of statistical methods as well as their limitations in practice.
- 9:00 – 12:30, Part 1: Introduction to the subject matters: Jaromír Antoch, Marek Brabec and Claudia D’Ambrosio
- 14:00 – 17:00, Part 2: Round tables: discussion with practitioners, individual consultations and final summary session.
Please note this special one-day ENBIS event is part of a two-day local conference. There is no registration fee but the number of available seats is limited so register here at your earliest convenience.