الفهرس 
LITERATURE REVIEWOnly 14 pages are availabe for public view
 |  | from | 32 |  |  |
| | | from | 32 | | |
Abstract
The most important incentive for electricity users to participate in a demand-side management program is the maximization of the user payoff and this can be done by rescheduling his energy consumption. Without customer participation demand response program will certainly fail to achieve their goal of reducing peak demand so customer acceptance and enrolment in the demand response program is an urgent need.
A distributed demand-side management system among users will be presented with a two way digital communication infrastructure. Then by using game theory an optimization equation is formulated. It deals with the total hourly load for each user. The players in this game are the users and their strategies are the daily schedules of their appliances and loads.
The reality of this program is proved by implementing it on a real data taken from ISO market and by using genetic algorithm with a constrain that the total daily energy consumption of each user remain approximately constant
The result show the success of this program that lead to reduce the billing tariff paid by each user and consequently the total energy cost.
Many factors affect on load forecasting studies such as previous loading temperature, weather condition, prices.
In deregulated market the price has a big influence on the load so it is useful to build a load-price forecasting models that depend on load and price time series.
Support vector machine is one of the novel machine learning techniques which enter recently power system study.
To implement a good Demand Response (DR) program, critical peak pricing (CPP) plan will be studied as an active demand response program. The economic perspective of CPP plan is the incentive of the plan conductor, or the profit of an energy service provider (ESP). The technical perspective is a method to maximize the incentive of CPP plan, or an ESP’s profit. If the electricity market price can be predicted properly, generation companies and the load service entities as main market participating entities can reduce their risks and maximize their outcomes further. This will be realized by predicting the market prices then with the help of the Support vector machine (SVM) which is one of the novel machine learning technique. It predicts the overloading conditions to make the critical peak price decision
The suggested models procedure presented in this thesis are as follows:
- Model 1 is the price forecasting model based on the data of the previous day. This model has been tested using real market data from the ISO market data
- Model 2 is the heavy load detection using support vector machine technique
- Model 3 is using game theory technique in the daily energy consumption cost optimization which means the maximization of the pay off of each user by rescheduling energy consumption Finally comparison of three algorithms is discussed and their effect on reducing the total energy cost:
- The genetic algorithm
- The harmonic search algorithm
- The backtracking search algorithm
The results show a good performance for the three algorithms as the three realize a positive profit. By comparing the profit it indicate that the harmonic search algorithm is better than the genetic because the profit is higher and the backtracking search algorithm is the best algoriyhm