Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Untitled Prezi

No description
by

Ibrahem Al-harbi

on 13 January 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Untitled Prezi

Classical Regression method Time series method •ARMA
•ARIMA
•ARMAX
•ARIMAX Neural networks (ANN) Fuzzy logic Support vector machines (SVM) hybrid Artificial Intelligence Based Techniques For Short Term Load Forecasting By Supervised by: Dr.Mamdooh S. Al-Saud Ibrahem Al-harbi
Ahmed Al-fayez Important of Load Forecast: Planning
Operation of Electric Utilities
Energy Purchasing
Generation
Load Switching
Contract Evaluation
Infrastructure Envelopment Load Forecasting Categorizes:
Very Short-Term
(<1hr) Long-Term
( >1 year) Medium-Term
(1week- 1year) Short-Term
(1hr- 1week) Short term Load Foretaste (STLF) Important of STLF Generating & Purchasing Electric Power
Load Switching
Unit Commitment & Dispatch
Estimate Load Flows
Prevent Overloading
Improve Reliability
Reduced Equipment Failures & Blackouts Methods For STLF Artificial Intelligence Modes of Operation: 1- Training Mode 2- Using Mode ANN (Artificial Neural Network) AN (Artificial neural) is a model of BN (Biological neural). BN AN AN The AN collects all incoming signals, and compute a net input signal as function of the respective weight. (Artificial Neural) (Artificial Neural) AN The behavior of AN depends on : weight Architecture of ANN ANN Layers: 1- input layer
2- hidden layers
3- output layer Back Propagation This means that the artificial neurons are organized in layers and send their signals "forward", and then the errors will be propagated backwards. Back Propagation The training begins with random weights. Then the error (difference between actual and expected results) is calculated. The goal is to adjust the weights so that the error will be minimum. We provide the algorithm with examples of the inputs and outputs we want the network to compute. Back Propagation The idea of the back propagation algorithm is to reduce this error, until the ANN learns the training data. Transfer Function Why ANN for STLF !? ANN good at extracting patterns from observed past events and extrapolating them into the future. ANN are able automatically to map the relationship between input sample and output . ANN learn this relationship, and stores this learning into their parameters "weight". Design ANN For STLF There is no general rule that can be followed in this process. It depends on engineering judgment and experience and is carried out almost entirely by trials and errors. Designing ANN for STLF : 1- Data pre-processing.
2- ANN Designing.
3- ANN Implementation.
4- Validation. Application of ANN for STLF
design accurate ANN stricture for short term load forecast for Riyadh electrical network using practical data Objective data preparation History load The First step for designing an ANN architecture is the preparation of the data.




checking the validity of the data
removing outliers and missing data scaling data
Ls=(L/(Lmax)) Ls:load after scale
Ts=(T/Tmax)) Ts: max temperature
after scale For designing an ANN, a practical data has been used from the previous 5 years (2008-2012) for Riyadh city Electrical network .



the data was requested from SEC (Saudi Electric Company ).The data file contains the hourly Load from 1/1/2008 until 31/8/2012 . maximum daily load from 1/1/2008 to 31/12/2011 maximum daily load Max daily temperature there are three type for temperature:
maximum
average
minimum the source of the temperature data was from these two web sites: USU.EDU CLIMATE.GOV ANN For One-hour ahead ANN Design for designing ANN For one hour ahead we use the Matlab-2010 to design the ANN and the nntool function (Neural network tool) that built inside the toolbox in the Matlab •ANN input design for one hour ahead Trials Example 1- training the ANN for 1 month (winter)
(1/1/2011-30/1/2011) 1-Test one day at 14-1-2011 (winter) 2- Test one day at 1-8-2011 (summer) The ANN have a bad performance when it is tested in period have different condition other than we train it. 2- training the ANN for 1 month (winter) (1/3/2011)-(30/8/2011) 1-Test one week start at 1-3-2011 (winter) The ANN have good performance compare with first trial and give a trusted result. 2-Test one week start at 14-8-2011 (summer) •Validation Where:
Lt=load target
Lf=load forecasted To test the validation, the ANN has been tested for week start at (1-9-2011) on period out the range of the training period the validation output for ANN ANN error and MAPE maximum error =6.7856 %
MAPE=1.5173 % The optimal ANN design: procedure to optimize the ANN architecture in terms of the number of hidden layers the optimal ANN has three hidden layers, five neuron in the hidden layer number one, five neurons in hidden layer number two and fifteen neurons in hidden layer number three. the max error =5.75 %, MAPE=1.315 % It can be concluded that the ANN technique can be easily adopted for STLF when we have enough historical data.

In this project, an ANN for STLF has been designed to give a forecast for one hour ahead of a load, and the optimal design of the ANN architecture gave us MAPE = 1.315%. and max error = 5.75%. Due to the huge difference between winter load and summer load, we prefer to design two ANNs Architectures for each season separately. Conclusion LOAD forecasting Thank You
Full transcript