LANS Informal Seminar: Jana Heckenbergerova

Seminar Title: Prediction and Optimization of Global Systems using Mathematical and Statistical Modelling
Speaker: Jana Heckenbergerova, Postdoctoral Fellow, Faculty of Electrical Engineering and Informatics, University of Pardubice

Date/Time: 2015-01-14 15:00
Location: Building 240 Room 1404-1405

Description:
The aim of this presentation is an overview of my research contributions. My research work is focused on the intersection of three subject areas: applied statistics, energy and environmental informatics and economy. It can be divided into a few keystone application topics:
1) advanced methods for dynamic thermal rating of overhead power transmission lines, aging models and optimal hydro-thermal coordination
2) spatio-temporal analyses and environmental modelling, numerical prediction models, forecast uncertainty modeling, estimation of wind direction distribution
3) detection and prediction of wind power ramp events
4) technical analysis of stock market prices, special chart pattern and graphical formation recognition
5) prediction and optimization algorithms based on nonlinear regression analysis.

Smart power transmission using advanced dynamic thermal rating methods is the largest topic. Our research group developed a system that evaluates real-time current-carrying capacity of power transmission lines. Utilization of a high-resolution numerical weather prediction model provides a complete spatio-temporal view of the thermal state of the system. A number of contributions include analyses of thermal aging and identification of hotspots, the sensitivity analysis of current-temperature calculations, the problem of economic dispatch and optimal hydro-thermal coordination and quantification of gains and risks when seasonal static thermal rating or typical meteorological year rating is utilized.

Environmental modelling and spatio-temporal analyses create the support of thermal rating model and renewable energy production forecasting. For example wind speed and direction affects both the actual ampacity of the transmission line and the production of energy in wind farms. Therefore, it is good to know the annual wind direction distribution for specific location. We have developed a new method for wind direction distribution determination using statistical model composed of a finite mixture of circular von Mises distributions, where meta-heuristics methods estimates model parameters. Other contributions contain the forecast uncertainty modeling using clustering and confidence intervals as well as characterization of a wind flutter generator, modelling of greenhouse gas concentrations and fluxes, analysis of wind power plants placement to reduce variability of generation and analyses of wind power ramp events occurrence.