• Nie Znaleziono Wyników

Abstract of the doctoral thesis: Forecasting of electricity usage in smart metering systems

N/A
N/A
Protected

Academic year: 2021

Share "Abstract of the doctoral thesis: Forecasting of electricity usage in smart metering systems"

Copied!
1
0
0

Pełen tekst

(1)

Abstract of the doctoral thesis:

Forecasting of electricity usage in smart metering systems

Smart metering systems are key components for creating environmental sustainability by managing energy at homes. They are supposed to play an important role in reducing overall energy consumption and increasing energy awareness of the users, through being better informed about consumption patterns. Leveraging smart metering to support energy efficiency on the individual user level poses novel research challenges in monitoring usage, providing high granularity information for the end users and accurate electricity load forecasting. Installation of smart meters opens new possibilities for advanced analytics of electricity consumption at the level of the individual household.

One of the most important tasks in various Smart Grid applications is the short-term electricity load forecasting applied at different levels, from an individual customer to a whole group of customers.

Taking into account the above considerations, the main research goal of the doctoral thesis is to develop an effective approach for the short-term electricity load forecasting based on smart meter data.

The scope of the study covers the following items: data preparation, development, computer implementation, optimization of the forecasting models and their parameters.

The thesis is divided into three empirical parts. In the first part, the study is aimed to detect the typical behavior of households (individual customers). In the second part, the dependencies between the electrical appliances were discovered and used to create dependent variables (inputs) for the forecasting models. The effectiveness of the forecasting methods developed for the individual customers were tested for the following electricity demands:

 24 hours forecasting horizon based on hourly data,

 one hour forecasting horizon based on hourly data,

 one day forecasting horizon (aggregate of the 24 hourly demands) based on daily data.

In the third part, based on the time series similarity measures and hierarchical clustering algorithm, the households were divided into homogeneous and mutually exclusive clusters. Then, the models for the aggregated demand for the whole population were developed, as well as the models for the aggregated demands for the individual households included in the previously specified homogeneous clusters. The effectiveness of the developed forecasting models (aggregated for the whole population and for the specified clusters) were checked for the following forecasting horizons:

 24 hours at the hourly demand,

 one hour at hourly demand,

 one day at a daily demand (aggregate of the 24 hourly demands).

Finally, the aggregated values of the forecasts related to the models built for different groups of individual customers (in relation to specified types of clusters), were compared with the results of the base model built for all customers (aggregate for the whole population).

The undertaken research tasks contributed cognitively and helped to confirm of the following general hypothesis:

 Clustering and pattern recognition algorithms can be good tools to identify patterns of user behavior, based on smart metering data.

 Electricity demand can be predicted effectively using the models based on advanced machine learning algorithms (compared to the other reference methods).

 It is possible to identify disjoint and homogeneous groups of households for which the forecasting will be more accurate than the forecasting applied on the entire population.

Cytaty

Powiązane dokumenty

Not only is it shown that the integration of passives into a single, multifunc- tional PCB transformer structure is feasible but also that the same integral PCB can be used

– Kodeks karny wykonawczy 14 , która przewiduje, że okres do sze- ściu miesięcy przed opuszczeniem przez odbywającego karę jednostki peni- tencjarnej (także po odbyciu całości

We suggest borehole radar measurements as a potential technique capable of monitoring fluid saturation changes in the near-field of production wells.. Electromagnetic (EM)

Wśród najważniejszych instytucji naukowych, które w znacznym stopniu stały się wzorem dla nauki arabskiej, była perska akademia w Dżundiszapurze4, która powstała jeszcze

ABSTRACT. The article is an attempt to present the analysis of a metaphor in modern texts. Having examined the update of various mechanisms and the presence of stylistic de- vices

ustanawiające Agencję Europejskiego GNSS, uchylające rozporządzenie Rady (WE) nr 1321/2004 w spra- wie ustanowienia struktur zarządzania europejskimi programami

Szkoły rządowe w Krasnymstawie utrzymywały się z funduszy miejskich (magistratu miasta) oraz w mniejszym stopniu ze składek pieniężnych zebranych od mieszkańców

The Seed Extension phase, one of the three main BWA-MEM algorithm phases that requires between 30%-50% of overall processing time, is offloaded onto the GPU.. A thorough design