Development of An Intelligent Forecasting System for Pakistan WAPDA using Machine Learning Techniques

Principal Investigator’s Organization (PIO):

Al-Khawarizmi Institute of Computer Science, UET Lahore

Principal Investigator (PI):

Dr. Zubair Ahmed Khan, Engr. Nisar Ahmad Bazmi, Dr. Waqar Mahmood

Summary

Power industry provides backbone for a country’s growth and development and is one of the most important utilities that underpin the survival of a nation’s critical industrial infrastructures and services. It is, therefore, very important to plan generation, distribution and transmission of electric power for current and also for future needs. The current imbalance between available energy supplies and increasing energy demand in Pakistan has necessitated rolling blackouts and renewed requests for conservation. With the increase in energy demand, Independent Power Plants (IPP’s) were introduced few years ago. The birth of a deregulated power industry in Pakistan brought new challenges for Pakistan WAPDA in terms of managing the IPP’s for their power generation, distribution, overall billing and invoice management. WAPDA has established a Power Purchase Agreement (PPA) with these private energy producers to handle energy purchase related issues. According to the PPA, IPP’s have to declare their available capacity 12 months, 3 months, 3 days and 24 hours in advance for medium term and short term available generation capacity. This project proposes to develop an intelligent decision making system that can forecast accurately short-term, long-term and real-time energy requirements and facilitate in most economic unit commitment and in planning efficient and least cost power generation; that applies to multiple sites in different seasons and is suitable for customized requirements. Load forecasting emerged as a natural requirement for IPP’s and there is also a direct need at National Transmission & Dispatch Company (NTDC) for such a system. Electric load forecasting (LF) involves the projection of peak demand levels and overall energy consumption patterns to support an electric utility’s future system and business operations. Accurate electricity LF is first and foremost important step in the decision-making process of any electric utility. This project developed an intelligent and accurate LF system for Pakistan Power System. This system is based on the intelligent modeling techniques of machine learning for regression and classification. Such models are applied on data collected from national and local power supply companies. Software user interface is simple and user friendly. A large variety of mathematical methods and ideas have been used for load forecasting. The development and improvements of appropriate mathematical tools lead to the development of more accurate load forecasting techniques. The accuracy of load forecasting depends not only on the load forecasting techniques, but also on the accuracy of forecasted weather scenarios and historical data quality. Noisy data can generate erroneous results. This software puts the most recent mathematical algorithms into this task and provides reliable pre-processing of input features to detect outliers, fill missing values and to remove duplicated values

Start Date 01-Nov-2011

Duration 15 months

Budget PKR 7.5 million

Status  Closed Project

Progress Report View Progress Report

Publications   N/A

Thematic Area  Engergy

Project Website 
http://kics.edu.pk/pro