Universal Smart Grid Agent for Distributed Power Generation Management
Eric MSP Veith
Diese Publikation zitieren
Eric MSP Veith, Universal Smart Grid Agent for Distributed Power Generation Management (2017), Logos Verlag, Berlin, ISBN: 9783832592233
69
Accesses
Accesses
Beschreibung / Abstract
``Somewhere, there is always wind blowing or the sun shining.'' This maxim
could lead the global shift from fossil to renewable energy sources,
suggesting that there is enough energy available to be turned into
electricity. But the already impressive numbers that are available today,
along with the European Union's 20-20-20 goal---to power 20 % of the EU energy
consumption from renewables until 2020---, might mislead us over the problem
that the go-to renewables readily available rely on a primary energy source
mankind cannot control: the weather.
At the same time, the notion of the smart grid introduces a vast array of new
data coming from sensors in the power grid, at wind farms, power plants,
transformers, and consumers. The new wealth of information might seem
overwhelming, but can help to manage the different actors in the power grid.
This book proposes to view the problem of power generation and distribution in
the face of increased volatility as a problem of information distribution and
processing.
It enhances the power grid by turning its nodes into agents that forecast
their local power balance from historical data, using artificial neural
networks and the multi-part evolutionary training algorithm described in this
book. They pro-actively communicate power demand and supply, adhering to a set
of behavioral rules this book defines, and finally solve the 0-1 knapsack
problem of choosing offers in such a way that not only solves the
disequilibrium, but also minimizes line loss, by elegant modeling in the
Boolean domain. The book shows that the Divide-et-Impera approach of a
distributed grid control can lead to an efficient, reliable integration of
volatile renewable energy sources into the power grid.
could lead the global shift from fossil to renewable energy sources,
suggesting that there is enough energy available to be turned into
electricity. But the already impressive numbers that are available today,
along with the European Union's 20-20-20 goal---to power 20 % of the EU energy
consumption from renewables until 2020---, might mislead us over the problem
that the go-to renewables readily available rely on a primary energy source
mankind cannot control: the weather.
At the same time, the notion of the smart grid introduces a vast array of new
data coming from sensors in the power grid, at wind farms, power plants,
transformers, and consumers. The new wealth of information might seem
overwhelming, but can help to manage the different actors in the power grid.
This book proposes to view the problem of power generation and distribution in
the face of increased volatility as a problem of information distribution and
processing.
It enhances the power grid by turning its nodes into agents that forecast
their local power balance from historical data, using artificial neural
networks and the multi-part evolutionary training algorithm described in this
book. They pro-actively communicate power demand and supply, adhering to a set
of behavioral rules this book defines, and finally solve the 0-1 knapsack
problem of choosing offers in such a way that not only solves the
disequilibrium, but also minimizes line loss, by elegant modeling in the
Boolean domain. The book shows that the Divide-et-Impera approach of a
distributed grid control can lead to an efficient, reliable integration of
volatile renewable energy sources into the power grid.
Inhaltsverzeichnis
- BEGINN
- 1 Introduction
- 1.1 Motivation
- 1.2 Contribution and Constraints
- 1.3 Overview
- 2 Fundamentals and Related Work
- 2.1 The Electric Power Grid
- 2.2 Simulation and Modeling
- 2.3 Computer Networks
- 2.4 Artificial Intelligence
- 2.5 Boolean Algebra
- 3 Approaching the Smart Grid by Modeling and Simulation
- 3.1 Models of the Power Grid
- 3.2 Reference Situation
- 3.3 Smart Grid Simulation Environment
- 3.4 Data Quality Assessment and its Influence on Simulation Runs
- 4 The Universal Grid Agent
- 4.1 Modular Design Principle
- 4.2 Interfaces
- 4.3 Agent Behavior
- 5 Forecasting Power Demand and Supply
- 5.1 Design of the Forecaster Universal Smart Grid Agent Module
- 5.2 The Multipart Evolutionary Training Algorithm for Artificial Neural Networks
- 5.3 Forecasting Accuracy and Efficiency
- 6 Social Component: Inter-Agent Communication
- 6.1 Motivation
- 6.2 Design Principles
- 6.3 Data Encoding
- 6.4 Analysis
- 7 Modeling and Calculating Demand and Supply for Agents
- 7.1 Agent-Local Power Balance
- 7.2 The Combinatorial Demand-Supply Problem
- 7.3 A Boolean Model of Demand and Supply
- 7.4 Evaluation of Efficiency
- 8 Conclusion