Download this complete Project material titled; Short-Term Electric Power Forecast In The Nigerian Power System Using Artificial Neural Network with abstract, chapters 1-5, references, and questionnaire. Preview Abstract or chapter one below

  • Format: PDF and MS Word (DOC)
  • pages = 65

 5,000

ABSTRACT

 

This thesis is a study of short-term electric power forecasting in the Nigerian power system using artificial neural network model. The model is created in the form of a simulation program written with MATLAB tool. The model, a multilayer time-delayed feed-forward artificial neural network trained with error back propagation algorithm, was made to study the pre-historical load pattern of a typical Nigerian power system in a supervised training manner. After presenting the model with a reasonable number of training samples, the model could forecast correctly electric power supply in the Nigerian power system 24 hours in advance. An absolute mean error of 4.27% was obtained when the trained neural network model was tested on one week, daily hourly load data of a typical Nigerian power station. This result demonstrates that ANN is a powerful tool for load forecasting.

 

TABLE OF CONTENTS

Title page                                                                                                                  i

Approval page                                                                                                         ii

Certification                                                                                                            iii

Dedication                                                                                                                iii

Acknowledgment                                                                                                    iv

Table of contents                                                                                                    v

List of figures                                                                                                          ix

List of Tables                                                                                                           x

List of Abbreviations                                                                                             xi

Abstract                                                                                                                    xii

CHAPTER ONE: INTRODUCTION

  • Background to the Study                   1
  • Statement of the Problem      4

1.3       Objectives of the Study                                                                               6

1.4       Delimitation of the Study                                                                           6

1.5       Significance of the Study                                                                            8

                                                                                                                       

CHAPTER TWO: LITERATURE REVIEW

2.1       Definition of Load Forecasting                                                                 9

2.2       Importance of Load Forecasting                                                               9

2.3       Problems of Load Forecasting                                                                  11

2.4       Techniques for Load Forecasting                                                              11

2.4.1   Extrapolation Technique                                                                            14

2.4.2  End-Use Method                                                                                          15

2.4.3   Scheer’s Method                                                                                          15

2.4.4   Multiple Regression                                                                                    17

2.4.5   Exponential Smoothing                                                                               20

2.4.6   Iterative Reweighted Least-Squares                                                          21

2.4.7   Adaptive tool forecasting                                                                           22

2.4.8   Stochastic time series                                                                                  24

2.4.9   Fuzzy Logic                                                                                                  28

2.4.10 Expert Systems                                                                                            33

2.4.11 Support Vector Machines                                                                          35

2.4.12 Neural Networks                                                                                         36

CHAPTER THREE: RESEARCH DESIGN/METHODOLOGY

 

3.0       Introduction                                                                                                 51

3.1       Research data                                                                                              51

3.2       Data Pre – Processing                                                                                 51

3.3       Choice of neural network paradigm                                                         52

3.4       Construction of network architecture                                                       53

3.5       Requirement of minimum number of patterns                                       54

3.6       Selection of input Variables                                                                      55

3.7      Network Training                                                                                        56

3.7.1   Training algorithms                                                                                                 57

 

3.7.2   Back-propagation Implementation Strategy                                           58

 

3.8       Improving Generalization                                                                          64

CHAPTER FOUR:

EXPERIMENTAL RESULTS AND DISCUSSIONS

 

4.0       Introduction                                                                                                  67

4.1       Selection of Network Architecture and Parametric Values                  67

4.2       Choice of training algorithm                                                                     69

CHAPTER FIVE:

CONCLUSION AND SUGGESTIONS FOR FURTHER RESEARCH

5.1       Conclusion                                                                                                    80

5.2       Suggestions for Further Research                                                              81

REFERENCES                                                                                                        83       

APPENDIX                                                                                                               97

 

 

 

CHAPTER ONE

INTRODUCTION

1.1   Background to the Study

A great deal of effort is required to maintain an electric power supply within the requirements of the various types of customers served. Some of the requirements for power supply are readily recognized by most consumers, such as proper voltage, availability of power on demand, reliability and reasonable cost.  By availability of power on demand, we mean to say that power must be available to the consumer in any amount that he may require from time to time. Stated yet in another way, motors may be started or shut down, fans and lights may be turned on or off, without giving any advance warning or notice to the electric power supply company.

It is this random behavior of consumers coupled with nature-controlled demographic and weather factors alongside econometric factors that has posed the greatest challenges like the amount of energy to generate, the load (circuits) to switch on or off at a point in time on the part of power utility company. Hence, a power system must be well planned so as to ensure adequate and reliable power supply to meet the estimated load demand in both near and distant future.

The primary pre-requisite for system planning is to arrive at realistic estimates for future demands of power. The foregoing concept is a part of load forecasting. Basically, load forecast is no more than an intelligent projection of past and present demand patterns to determine future ones with sufficient reliability [1]. The Nigerian power system today is known for its epileptic, inadequate and unreliable nature [2]. Its performance will improve if a system for accurate load forecasting is designed to aid its operation and planning. Accurate load forecasting holds a great saving potential for electric utility corporations. According to Bun and Farmer, [3] these savings are realized when load forecasting is used to control operations and decisions such as economic load dispatch, unit commitment, fuel allocation and off-line network analysis. The accuracy of load forecasts has a significant effect on power system operations, as economy of operations and control of power systems may be quite sensitive to forecasting errors [4]. Haida et al, [5] observed that both positive and negative forecasting errors resulted in increased operating costs.

Load forecasting may be applied in the long, medium, short, and very short-term time scale. Srinivasan and Lee, [6] classified load forecasting in terms of the planning horizon’s duration: up to 1 day for short-term load forecasting (STLF), 1 day to 1 year for medium-term load forecasting (MTLF), and 1-10 years for long-term load forecasting (LTLF). Short-term load forecasting (STLF) aims at predicting electric loads for a period of minutes, hours, days, or weeks [7]. STLF plays an important role in the real-time control and the security functions of an energy management system. STLF applied to the system security assessment problem, especially in the case of increased renewable energy sources (RES) penetration in isolated power grids, can provide, in advance, valuable information on the detection of vulnerable situations. Long- and medium- term forecasts are used to determine the capacity of generation, transmission, or distribution system additions, along with the type of facilities required in transmission expansion planning, annual hydro and thermal maintenance scheduling etc. [7] . Kalaitzakis et al, [7] noted that short-term load forecast for a period of 1-24 h ahead is important for the daily operations of a power utility since it is used for unit commitment, energy transfer scheduling and load dispatch.

Achieving accurate electric load forecasting is by no way a simple thing.  This is because electric load is determined largely by variables that involve “uncertainty” and whose relation with the final load is not deduced directly [8]. Some of these variables or factors include economic factors, time, day, season, weather and random effects. Electricity usage may be, therefore, predicted using data from previous history of load, temperature, humidity, luminosity, and wind speed among other factors. However, accurate models of load forecasting that use all these factors increase modeling complexity. Several methods, therefore, have been used to perform load forecasting each with its inherent shortfalls. Time series analysis is a very effective method to create mathematical models for solving a broad variety of complex problems [9]. These models are used to identify or predict the behavior of a phenomenon represented by a sequence of observations. However, creating an accurate model for a time series that represents non-linear processes or processes that have a wide variance is very difficult [9]. The trend today, however, is to solve most problems of human using Artificial Intelligence Means (AIM).  Artificial intelligence methods for forecasting give better performance in modeling of time series problems [10].  Artificial Neural Networks (ANNs) being one of the artificial intelligence means have been successfully used to solve a broad variety of systems, entailing linear and non-linear processes [9]. The application of ANNs in time series prediction is presented in [11] and in [12]. The success in the application of ANNs lies in the fact that when these networks are properly trained and configured, they are capable of accurately approximating any measurable function. The neurons learn the patterns hidden in data and make generalizations of these patterns even in the presence of noise or missing information. Predictions are performed by the ANN based on the observed data. Load forecasting is clearly a time series problem and an example of a time series problem that can be solved with ANNs is electricity load forecasting.

Artificial intelligence or AI is the general term used to describe computers or computer programs which solve problems with “intuitive” or “best-guess” methods often used by humans instead of the strictly quantitative methods usually used by computer [13].  Expert systems, neural networks, fuzzy logic and support vector machines are some of the AIs currently in use today. Programs for some problems such as image recognition, speech recognition, weather forecasting, electric load forecasting, and three dimensional modeling are not easily or accurately implemented on fixed –instruction-set computers such as 386/i486-based systems [13].  For applications such as these, new computer architecture, modeled after the human brain and which is known as Artificial Neural Network, shows considerable promise.

Hence, in this study a novel attempt is made to solve the problem of electric power forecasting in the Nigerian power system by means of artificial neuronal network.

1.2   Statement of the Problem

It is an established issue that the Nigerian power utility company is nowhere in the energy business. The utility company, PHCN, as it is called today, is yet to meet the people’s demand for electric energy satisfactorily for any known period of time. It is evident that the generated power is inadequate and so, the utility company considers load shedding and restricted demand as a way out just as the government of the federation is insisting on privatization of the energy sector as the last resort. Worst still, even under these conditions of load shedding and restricted demand, the integrity of the supplied power has always been questioned. The irony of this development is that it is happening when Nigeria is striving to attain vision 202020.

The problem, therefore, is in spite of this inadequacy in generation, is there any way we can manage what we have to satisfy our taste? Since economics is all about using limited resources to address the endless human needs, there are ways. The issue now is, what are these ways forward?

Before we x-ray one way forward, we need ask: can prompt and proper decisions on unit commitment, fuel allocation, energy transfer scheduling, and load dispatch be of any help? Certainly, YES. Since short term load forecasting is necessary for such prompt and proper decisions on unit commitment, fuel allocation, power wheeling arrangement, load dispatch etc., knowledge of load forecasting in the Nigerian power system is one such way forward. Better still, what if this load forecasting is performed by means of artificial intelligence- Artificial Neural Network (ANN) means? In other words Man Machine Interface (MMI) can be guaranteed. The problem is more than half-way solved since management decision can now be automated.

So, this research is aimed at suggesting a solution to the ailing Nigerian power system by proposing a model which can perform 24-hours-ahead load forecasting in the Nigerian power system by means of artificial neural network.

1.3       Objectives of the Study

Although the objectives of the study can be inferred from the background to the study outlined in the previous section, it can still be clearly and concisely stated that the objectives of the study are:

  • To model artificial neural network which can forecast electric power supply for one day in advance (Short Term Load Forecasting);
  • To train the model (using back propagation algorithm) with pre-historical load data obtained from a sample of the Nigerian power company so that each input produces a desired output;
  • To Test  the model to get the values of future power supplies in the Nigerian power system ; and
  • In the light of the above, make necessary recommendations and suggestions for further research.

1.4       Delimitation of the Study

It will be clear from our objectives that even though the impetus for this study was generated by the ‘sorry’ state of the Nigerian power system, the scope of this study has been restricted to New Haven Enugu 132/33KV Transmission station. In addition, short-term load forecasting model is being proposed in this work. This restriction has been dictated by the need to attempt a reasonable depth of treatment of data collected within the time available and on the other part due to some financial limitations.

Although demographics, econometric and weather conditions need be considered during load forecasting, the ANN modeled for the purpose of this research will be trained without such factors as inputs. This became necessary so as to avoid the unnecessary model complexity that is usually associated with a model encompassing all or some of such factors. The forecasting model being proposed here does not take into account temperature, although in general it might have significant impact on model accuracy. Temperature data have been omitted simply, because the prediction is concerned with data corresponding to the territory of the whole country and since temperature changes a lot in different regions of Nigeria, it would be difficult to adjust the proper value of temperature for a particular day. Though load data from a sampled Nigerian power station will be used to test the model, the model remains for the entire Nigerian utility. From available literature still, it is primarily the behavior of low voltage consumers or residential consumers that is directly affected by weather variables [14], [15].

Furthermore, inaccuracy of weather forecasts, difficulties in weather-load relationship modeling and implementation problems limit the use of load forecasting models requiring weather data, thus several works have appeared recently omitting weather data [7],  [16], [17], [18], [19], [20].  To support this stance further, we make the following case: EUNITE (European Network of Excellence on Intelligent Technologies for Smart Adaptive Systems) had in the year 2001 organized a world-wide competition on methods to accurately predict electricity load [9]. In the contest, the average temperature and load data on half hourly basis for years 1997 and 1998 were provided. The objective of the contest was to predict daily peak demands of electricity for January 1999 based on the data from these previous years. The model proposed by Chang et al., [21] which in terms of mean absolute percentage error (MAPE) obtained the first place in the competition actually discarded the temperature data.

This shortfall is also due to non-readily availability of information on such factors at the point of need.

The above restrictions notwithstanding, however, there are strong indications from the available literature that any findings and conclusions will be generalizable to Nigerian Power System at large even at a high degree of accuracy of the model results.

1.5       Significance of the Study

The proposed study is expected:

  • To guide the operation and planning of the Nigerian Power System
  • To aid power system Engineers who may wish to design a power system newly
  • To serve as a research or academic material
  • To help validate the results obtained by other leading researchers who might have worked on this same or similar area.

GET THE COMPLETE PROJECT»

Do you need help? Talk to us right now: (+234) 08060082010, 08107932631 (Call/WhatsApp). Email: [email protected].

IF YOU CAN'T FIND YOUR TOPIC, CLICK HERE TO HIRE A WRITER»

Disclaimer: This PDF Material Content is Developed by the copyright owner to Serve as a RESEARCH GUIDE for Students to Conduct Academic Research.

You are allowed to use the original PDF Research Material Guide you will receive in the following ways:

1. As a source for additional understanding of the project topic.

2. As a source for ideas for you own academic research work (if properly referenced).

3. For PROPER paraphrasing ( see your school definition of plagiarism and acceptable paraphrase).

4. Direct citing ( if referenced properly).

Thank you so much for your respect for the authors copyright.

Do you need help? Talk to us right now: (+234) 08060082010, 08107932631 (Call/WhatsApp). Email: [email protected].

//
Welcome! My name is Damaris I am online and ready to help you via WhatsApp chat. Let me know if you need my assistance.