Load Profiling Approach In Clustering Automated Electrical Power Metering System

ABSTRACT

In Nigeria, customers’ complaints and lean revenue generation are the major challenges in Electricity market.  Customers’ complaints range from over billing, estimate billing to under billing. Cases of lean revenue collection with respect to energy supplied are mostly due to unstructured consumer classification. This study ushers in an established and alternative means of electricity customer classification using unsupervised classification technique. The study provides an opportunity using load profiling and clustering technique to build representative electricity consumer load profile as new electricity customer classification depicting levels of customer types which includes private customers and their representative profiles. Load data which are monthly consumers’ consumption load value (in kilowatt-hour) were mined and clustered using a combination of k-means R algorithm and python programming language, while the result was presented in k-means R. The three clusters formed by the k-means R algorithm describe uniquely main category of customer types. Other sub clusters describing different private customer groups were also realized using k-means R algorithm by breaking up further residential customer category. High qualities of cluster groups representing distinctive   load profile class were tested for convergence and stability by evaluating total percentage within sum of square error for levels of iterations. This technique was tested by clustering consumers of Enugu Electricity Distribution Company (EEDC) particularly within the state metropolis (New Heaven District) as a new electricity customer classification to generate customers’ electricity tariff to change reliance on old tariff classification generated by questionnaires. The K-means R result showed four private consumer profiles and three main   consumer profiles describing the entire customer category with Sum of Square Error  of 95.3% and 77.7% respectively. It proved the identification of consumers to different profiles including consumers within suburbs. It is on the premises that clustering technique to load profiling is deemed so important for study and recommended to Distribution companies so as to strengthen tariff and enhance revenue collection by predicting right monthly consumption based on k-means technique

 

 

TABLE OF CONTENTS

Title Page	i
Approval	ii
Certification	iii
Declaration	iv
Dedication	v
Acknowledgment	vi
Abstract	vii
List of figures	viii
List of  tables	ix
List of abbreviations	x
Table of contents	xii
CHAPTER ONE: INTRODUCTION	1
1.1 BACKGROUND	1
1.2 STATEMENT OF THE PROBLEM	4
1.3 AIM AND OBJECTIVE OF THE STUDY	5
1.4 SIGNIFICANCE OF THE STUDY	5
1.5. PROPOSED METHOD	6
1.6.  SCOPE OF THE WORK	6
1.7. PLAN OF THESIS	6
CHAPTER TWO: LITERATURE REVIEW2.1 INTRODUCTION 2.1.1 Electricity 2.1.2 Electricity Billing System 2.2 POWER METERING SYSTEM 2.2.1 Automated Electrical Metering System 2.2.2 Types of Electric Metering Technologies 2.3 ELECTRIC METER 2.3.1 Analogue Meter: Electromechanical Meter 2.3.2 Analogue Meter: Electricity Metering from Feeder Pillar 2.3.3 Automatic Meter Reading to Automatic Meter Infrastructure (AMR to AMI). 2.3.4 Digital Meter: Electronic Meters 2.4 DATA MINING 2.4.1 Data Mining Process 2.4.2 Exploring Clustering Technique 2.4.3 Hierarchical Clustering 2.4.4 Partitioning Clustering 2.4.5 K-Means 2.4.5.1 K-means Algorithm 2.4.5.2 Assigning Points to the Closest Centroid 2.5 DEFINITION OF LOAD PROFILING 2.5.1 Overview of Load Profiling 2.5.2 A Clustering Approach to Load Profiling 2.6 CUSTOMER CLASSIFICATION 2.6.1 Classifying Electricity Consumers 2.6.2 Other Related Work   CHAPTER THREE: RESEARCH METHODOLOGY 3.1 INTRODUCTION 3.1.1: Essence of K-means clustering technique 3.1.2 Clustering Method 3.2 CASE STUDY ON CONSUMPTION VALUES OF ELECTRICITY CUSTOMERS 3.2.1 Data Pre-processing 3.2.2 Software Packages 3.2.3 Centriod Initialization 3.3 CLUSTERING LOAD DATA 3.3.1 Electricity Load Profile Class Characterization 3.3.2 Coding Using Pseudo-Code   CHAPTER FOUR: EXPERIMENT, RESULTS AND DISCUSSION 4.1 INTRODUCTION 4.1.1 Data Pre-processing: Data Transformation 4.1.2 Architecture of Load Profile Using K-Means Technique 4.2 EXPERIMENT 4.2.1 The Data Segmentation Using K-Means Algorithm 4.2.2 The Pseudo Code Template 4.3 RESULT 4.3.1 Cluster Means and Iterations 4.3.2 Data Assignment via Clustering Vector 4.3.3 Cluster Formation and Data Visualization 4.3.4 Quality of Clusters by Sum of Square Error 4.4 DISCUSSION   CHAPTER FIVE: RECOMMENDATION AND CONCLUSION 5.1 SUMMARY 5.2 RECOMMENDATIONS FOR FURTHER WORK 5.3 CONTRIBUTION 5.4 CONCLUSION	77 7 8 9 10 11 12 12 13 14 15 17 18 18 21 21 24 25 28 28 30 31 31 35 36   39 39 40 41 43 43 43 44 44 45 49   51 51 52 53 54 54 55 57 57 59 61 65 67   71 71 71 72 73
REFERENCES	74
APPENDIX	83
Appendix I	84
Appendix II	86
Appendix III	87
Appendix IV	108

 

 

 

CHAPTER ONE

INTRODUCTION

• BACKGROUND

Over these years, electric energy consumption has continued to grow in rapid proportion both in rural and urban areas. This could be attributed to the recent rise in socio-economic activities throughout the federation. Suffice to mention that the increase in developmental activities have seen a surge in people and companies requiring electric energy. According to International Energy Agency (IEA) [1], it indicates that the energy usage and consumption will top 75% from what is used today. However due to this increase, it becomes imperatively difficult to properly measure the exact consumption. Importantly, scientific techniques that will match and measure exactly the load consumption of various customer types and generate unambiguous billing to end-users in turn to ensure adequate revenue collection to utility companies has been the trend of research in Liberalized Electric market.

In Nigeria of late, according to Nwaoko in [2]-[3], the defunct National Electric Power Authority (NEPA) embarked on and commissioned the scientific control and data acquisition project (SCADA) that facilitates and enhances the distribution operation’s ability to acquire data from the grid, effect switching operation from a distance. Supply Authority went ahead to meter its transmission stations with a view to ascertaining the flow and quantity of energy. This was followed immediately by the monetization process of energy delivered to each Business Unit and the exercise created a challenge to minimize all kinds of energy losses in the system and satisfactory bills. In taking up the challenge, the Power Holdings Company of Nigeria (PHCN) has been exploring series of innovations for its present metering system.  Such innovations include Automatic Meter Reading (AMR) and the Pre-payment Meter Scheme, and load profiling.ZAccording to [1], Meters provide data that offers insight into the operation of data centre infrastructure (i.e. power and cooling systems) within a data centre. Specific types of meters exist for various reasons, from tracking the use of electricity to analysing the power quality in a facility and reporting problems such as transients and harmonics to measuring the power usage effectiveness (PUE) of the data centre. In [4], they narrowed the definition of electric meter as devices used to keep track of the quantity of usage as well as billing energy consumption charges, which are computed in kilowatt-hour (kWh). Edson Electric Institute (EEI) defined electric meters as electronic measurement devices used by utilities to communicate information for billing customers and operating the electric systems. As energy initiatives and legislation continue to increase, the necessity for more in-depth metering to better understand and optimize energy use is also increasing. From the foregoing definition of meters, it is apparent that the sole purpose of an electric meter is to provide accurate measurement of energy consumption. Electric meters in operation, adopt metal disc rotation to ascertain the quantity of energy consumed. The metal disc rotation of electromechanical meter at a defined speed is proportional to the power flowing through the meter which in-turn determines the energy usage and therefore the appropriate charge(s) that consumers pay. Both traditional and electronic metering systems have the following performance shortcomings which include low accuracy, electromagnetic interference, poor performance on change in temperature, poor interfacing facility, uneven rotation on no load condition, and failure in detecting power theft.  Due to these inefficiencies, the electromechanical systems are being phased out gradually and replaced with the Automated Meter Reading system (AMRs) which is more effective and offers a more accurate measuring device compared to the conventional electromechanical metering system though these meters are expensive. For effective metering, according to [5]-[6] they stated that electronic-meter (AMRs) gives high accuracy for nonlinear loads than conventional rotating disc type electro-mechanical meters.

Effective metering and monitoring gives owners and operatives of property crucial information about how their property (whether commercial and private) are performing with a view of determining and improving them. Inefficiencies connected with improper billing can be nightmarish to the energy users. An effective metering and monitoring system has the capacity to get tenants, property managers, and owners involved in energy-efficiency measures. In African countries, complaints have been laid for irregularities in the billing system which is characterized by ambiguous estimated bills, over/under billing to no billing at all. This has made it impossible to establish a definite pattern that will regularize billing that will take into account different consumer classes.

Consumers of electricity can be classified into two major groups for modelling, category model and area model. Area model is a modelling of consumers that are not metered on time interval basis within the geographic region covered by a network and category model groups customers with a similar load pattern into categories using clustering techniques. By this consumer characterization can be fully realized. In Nigeria electricity market, before now what control and determine market price of electricity are not demand and supply. Instead distribution operators determine at what price electricity is sold. The current Nigeria Load Profile Classification method uses a historical database of loaddemand customer profiles, collected over several years, to group customers according to their load pattern, such as industrial, business, public services and residential loads.These operators developed an undefined tariff rates for their customers. The classification of customer types into different charge rate scheme is shrouded in secrecy and monopoly. The operators classified customers at random into category without due consultation, deliberation and agreement. A situation where low energy consumer class is grouped with high energy consumer class and high energy consumer is classed into low energy consumer class. Residential customer could be metered on commercial customer basis, while industrial customer metered as residential customer or commercial customer. This is done at the distribution side even as new customers are identified. Consequently, its effect is irregularities shown in billing structure, as there existed previously case of high/under estimate bills. However one of the major consequences of electricity markets liberalization is the freedom that all customers will have on the choice of their electricity supplier. This new scenario creates an environment where several retail companies compete for the electricity supply of end users. An overview about electricity retail markets is presented in [7]. To obtain well-functioning markets it is essential to define new rules and structures concerning data collection and description and the definition of communication protocols between the different participants in the market. These new structures will increase significantly the amounts of data collected by the participants in the market. It will also establish and define a customer and operator relationship as it focuses on utility rate. In this sense, what is paramount is data. This is because it plays important role in the decision support and in the definition of market behaviour. The development of frameworks and tools, able to extract useful knowledge from this data can be a competitive advantage for the participants in the market.

One of the important tools defined in these projects are different consumers’ classes represented by its load profiles. These load profiles or different consumer classes can be assigned uniquely different tariff rate depending on different level of consumption values (data). Load profiling has been a matter of research during the last years [8]–[9].

Load profile is different forms data can take to be able to do data prediction. In [8] data mining techniques are used on the determination of load profiles for different type of consumers. Customers are classed in groups and each group has a representative load group. In [10] and [11] statistical and clustering techniques are used on the determination of load profiles to support the development of tariff offer and market strategies. In [9] a knowledge discovery in databases (KDD) process, to extract useful knowledge from electricity consumers’ data, is described. In this process data mining techniques are applied, on different stages of the process, to find the different consumption patterns. These patterns are represented by their load profiles and each of the patterns represents a consumer class. This knowledge is useful to develop a decision support system to support the definition of adequate contract options and market strategies. This research presents a framework developed to support the retail and distribution companies on the extraction of knowledge from electricity consumption data. For the start, in order to identify and classify customers from the varying scales of consumption values available, core clustering techniques are investigated while k-means clustering technique is adopted. Main clusters(customer scales attributes) viz: residential, commercial, and industrial customers were formed using K-means algorithm. Other sub clusters were also developed. The clusters formed by the miner’s software describe uniquely category of customer types. With this technology, more or if not all customers will be identified, categorized and billed translating to robust revenue collection.This proposed technique can be used by EEDC-Enugu Electricity Distribution Company as a new electricity customer classification to generate customers’ electricity tariff. The results demonstrate that the proposed method is efficient for assigning Load Profile to the consumers and also shows that the energy consumption can be clustered not only based on the load pattern but also load value.The framework is able to treat different data sets in an easy and efficient way and provides results like consumer classes, represented by its load profiles, and classification models. These results can be updated as new data is collected.

• STATEMENT OF THE PROBLEM

Developing an efficient alternative measure for electricity customer classification and robust revenue collection are necessary for billing customers accurately. Building up a representative load profile using clustering technique is the most reliable means of billing customers that cannot afford smart metering systems. With this efficient technique, all levels of customers will be captured for billing including low electricity customers in suburbs and rural areas.

1.3       AIM AND OBJECTIVE OF THE STUDY

The aim of the research is to develop a load profiling technique in clustering automated electrical power metering system. The specific objectives are:

• To build a category model to support the retail and distribution companies by extraction of knowledge from electricity consumption data.
• Building a data sample for Enugu Electricity Distribution Company EEDC where electricity consumers can be monitored for load management distribution system planning.
• To provide a template that ensures consumers are properly identified, grouped into representative load profiles that uniquely determines their tariff rate.

1.4       SIGNIFICANCE OF THE STUDY

Developing load profiling using unsupervised learning methodology is very significant to electricity distribution companies as well as its customers in that it will enable various participants to have a complete newer perspective about billing system geared towards ensuring that unstructured tariff rates are eliminated in electricity market. Secondly, this research work is important since it will give distribution companies the requisite skill and tools to sift through large data (formally unexploited) accruing from automated (consumption values) power metering system for market strategic behaviour. By this power theft, estimate/under billing and no billing of electric consumers as the case may be, are adequately addressed.

This work can be said to be of significant to both business, government and as well as the academic community. In the business area, the work can be used to improve and enhance the manner by which many distribution companies appropriate their tariff charges at satisfactory services. The work can also be useful to the government agencies in building intelligence especially in area of surveillance and load planning. Finally, this work presents sufficient challenges to the academia of Business Intelligence by increasing their research interest in this field of knowledge from many other dimensions.

1.5       PROPOSED METHOD

The study presents and ushers in new analytical method for electricity customer classification as reliance on historical database collected over several years for tariff structure does not entirely justify load pattern of this age and time. In this research load profiling approach in clustering load values realized using k-means(R analytical software tool) technique is developed.

1.6       SCOPE OF THE WORK

This work studies automated electric metering system, load profiling, clustering techniques particularly k-means algorithm i.e. unsupervised learning technique. This study is all encompassing as it is bent to review metering system, automated metering system, data mining as well as clustering techniques and their impact as it groups electricity consumers into right tariff and billing for customers.  Consumption values of different consumption charges will be used to construct different patterns of classes of consumers and unsupervised learning technique will be applied to gain insight into the internal underlying structure of the datasets for knowledge.

1.7       PLAN OF THESIS

This work is further organized as follows: Chapter two reviewed all the electricity customer classification with emphasis on tariff rate particularly for Nigeria electricity market. Attention is also drawn to concept of load profiling, clustering techniques and their related works. Chapter three detailed the research methodology and the proposed technique developed. In Chapter four, the result of the research was discussed and validated while in Chapter five conclusions were drawn and recommendations made. Finally the work ended with References and Appendixes

 

 

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].