A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING

ABSTRACT

This work addressed the problem of forecasting active and reactive power at a substation transformer in a distribution system. Accurate power forecast is of great importance in power  distribution  planning,  reactive  power  support  control  and  intelligent  power management.  Due to the complexity of the power system, an  intelligent and adaptive forecast algorithm based on the Adaptive Neuro-fuzzy Inference System (ANFIS) was modeled for the power forecast. For the proposed  ANFIS forecast model training and validation, historical data of active and reactive power from the Abakpa Enugu Nigeria distribution network was used. The case  study power system is modeled in MATLAB SIMULINK  with  the  proposed  neuro-fuzzy  forecast  model  integrated.  Simulation  is carried out to obtain the time series of one hour ahead and three hour ahead forecast of the active and reactive  power. Graphical  output shows that the forecasted  active and reactive  power  time  series follow  the signal profile of the actual (measured)  system active and reactive power. The evaluation of coefficient of multiple determination was used  to  determine  the accuracy  of the  forecast  model.  Result  evaluation  carried  out determined the coefficient of determination to be 0.98 and 0.72 for the one hour ahead and the three  hour ahead active power forecast  respectively.  Similarly,  the  one  hour ahead and three hour ahead reactive power forecast gave 0.82 and 0.71 respectively. For the  one   year   ahead   (long  term)   forecast   obtained,   the   coefficients   of  multiple determination are 0.54 and 0.62 for active and reactive power respectively. The results indicate very strong degree of correlation between the actual power time series and the forecasted time series. However these values show that the near real-time forecast of one hour ahead and three hour ahead, are  more accurate than the long term forecast. This shows the high degree of accuracy of the proposed neuro-fuzzy forecast model.

CHAPTER ONE

INTRODUCTION

1.0       Background of the Study

Accurate active and reactive power forecasting is of great importance for power system operation.     It  is  the  basis  of  economic  dispatch,  hydrothermal  coordination,  unit commitment, for solving optimal control problem required in minimizing power losses, enhancing  voltage  stability,  raising  general  system  reliability,  and  system  security analysis among other functions [1, 2, 3]. Short-term active, reactive and load forecasts have become increasingly important  since the rise of the competitive energy markets [4,5, 6, 7].

Forecasting is necessary in the provision of look-ahead information for effective reactive power  management  which  has  become  important  for  distribution  system  operators (DSO). This is vital in keeping voltage limits, providing quality and system reliability[8]. For  effective  reactive  support,  accurate  forecast  is necessary  in determining  suitable reactive power to be injected into the system.

Furthermore, in certain countries, DSOs must make active and reactive power forecast to provide necessary information for national transmission and generation planning studies according  to regulatory legislation.   It is reported  [2], that with  the liberation  of the electric sector, power forecasts started to play a key role with regard to investments in distribution,  planning  and  energy  management   strategies  at  regional  and  national systems.

Power forecast is very important in providing the proactive, look-ahead data for effective coordination  control required  in distribution  system  for intelligent  voltage  regulation, especially in systems involving distributed generations [9].   Voltage regulation is very

important in maintaining utility power quality and satisfactory voltage levels at customer terminals.   Therefore, the utility usually controls the main transformer  under load tap charger  (ULTC)  position and capacitor  status to improve  the  voltage  profile,  reduce system  losses  and  increase  system  efficiency  [10].  Reactive power  and  voltage  are efficiently controlled to improve the voltage quality and decrease power generation costs. However,  reactive  power  and voltage  control  devices  are operated  independently  by themselves, i.e, these devices are not coordinated.  Switched shunt capacitors have some shortcomings that include delivering reactive power to the system without following load variations and the inability to absorb superfluous reactive power from the system.  In the literature,  there  are many studies [11-13] that have been reported to solve the reactive power and voltage control problem in a distribution system.  However it is strongly held [2, 14] that accurate power forecast is required to improve on these control techniques. A model for the real-time  forecast  of active and reactive power is necessary  for  the optimal coordination of control devices to achieve the objectives which  minimize the total reactive power flow through the substation transformers, the total voltage deviation on the buses, and the total number of ULTC tap operation in the day.

System planners and operators need tools and methods to better forecast the  dynamic changes  of active,  reactive  power  demands  and  resources[15].    They  need  to better recognize significant changes of network active and reactive power demand and supply based on loading (notice the ever increasing motor and power electronics loads and their nonlinearity),  voltage profile, and topology changes due  to maintenance,  contingency, and  consequent  mitigation  measures.    It  is also  important  to  clearly  understand  the interplay  and  impact  of  switching  operations  of  reactive  power  resources,  load  tap changes of transformers, and power electronics based reactive power compensators.

There are several methodologies for electric power and load forecasting reported in the references  [3,  16,  17].    The  most  common  models  reported  are  the  Autoregressive Moving Average (ARMA) models, possibly with exogenous inputs (ARMAX), the naïve forecast model, and statistical models. The main shortcomings of these forecast models, as reported [16, 17], is that they are based on the assumption that the dynamics of the power system is linear.   It is found that these deterministic forecast models might not perform  well under dynamically  changing  operating conditions  of the power system, hence operational forecast using these  models might not be reliable.   Furthermore,  as showed by reference [14], that the  forecast of reactive power cannot be modeled by a simple  regression  of  active  power.    Due  to  these  shortcomings  of  these  forecast algorithms,  a  more  intelligent  and  adaptive  algorithm  is  required  for  the  real-time forecast of active and reactive powers in a power system.

1.2       Statement of the Problem

One  of  the  key challenges  facing  the  power  system  engineering  community  is  the development  of  accurate  forecast  model  for  look-ahead  information  for  distribution system coordination control required for ensuring voltage stability, effective distribution planning and energy management strategies.  There is the demand for accurate real-time forecasting  tool  required  to  provide  real-time  data  for  complex  computation  in  the optimal  minimization  of  power  losses  and  for  the  coordination  of  reactive  support equipment:  the shortcomings  of shunt  capacitor can be reduced or eliminated  by the provisions of real-time forecast data that enables it to proactively follow load variations in delivering reactive power to the system.  Furthermore, considering the shortcomings of the conventional power  forecast models as briefly noted in the background of this study,  the  problem  to  be  tackled  by  this  work  is  the  development  of  a  real-time

intelligent   forecast   model,   that  can  use  as  input  the  non-linear,   imprecise   and dynamically  changing  system  operational  data  (measured  from  station  feeders)  to forecast the active and reactive powers at a substation transformer.

1.3       Objectives of the Study

The aim or main objective of this study is to develop an intelligent model for the real- time forecasting of active and reactive powers at a substation transformer, that provides necessary  real-time   data  for  the  coordination   control  of  reactive   power  support equipment, distribution planning and effective energy management.   To this end, this study is to realize the following specific objectives

(i).    To develop an adaptive neuro-fuzzy inference model for the real-time forecast of   active   and   reactive   power   at   substation   transformers   using   system operational data measured from feeders.

(ii).   To model the integration of the proposed forecast model with a power system using MATLAB

(iii).  To setup and carry out simulation studies of the operations of the  proposed intelligent forecast model

(iv).   To  carry  out  evaluation  of  the  accuracy  of  the  proposed  forecast  model.

Evaluation includes evaluating the correlation of  the forecasted and actual(i.e. measured) time series of the active and reactive power.

1.4         Significance of the Study

The study has a number of significance for the industry and academia.

The  realization  of an  intelligent  power  forecast  model  as proposed  in this  study  is important industrially for the vital task of economic dispatch and load frequency control in power systems.  This is carried out to adjust the area generation such that it matches the  area  load.   Since  energy  demand  forecast  is vital in carrying  out  this task,  the development of the intelligent model capable of accurate forecast of active and reactive power becomes very essential.

The immediate value of this study is its realization of an intelligent forecast model for the accurate, active and reactive power forecast in real-time and over a wide  range of time.  More knowledge of active and reactive power demand allows system planners to optimally  allocate  reactive  resources  to  minimize   the  reactive   power  flow  over transmission  lines  and  through  substation  transformers  and  therefore,  reduce  power losses.  Better active and reactive power forecast allow system operators to control the reactive  power  to  improve  voltage  profile  and  reduce  power  losses.    Reduction  of reactive  power  flow  over  transmission  system  enables  more  utilization  of  existing transmission capacity.

It is known  that voltages  are better  controlled  by predicting  and  correcting  reactive power demands from loads [14].  A study such as this, is an important contribution in the development of forecast models for the coordination of control devices (transformer tap changes, capacitor switches, SVC etc.) and power compensation devices such as FACTS. The realization of an intelligent forecast  system is vital for the provision of real-time look-ahead  data,  to  give the proper  dispatching  strategy for  capacitor  switching  and transformer tap movement for control of voltage in a distribution system.

The significance of this is furthermore emphasized considering that the development of a versatile forecast algorithm would be useful in the forecast of peak power of substations and  to  support  the  planning  of new  investments  on  grid  expansion,  reactive  power support and energy purchasing.

1.5. Scope of the Study

This work covers the forecast  of active and reactive power at distribution  substation transformer. It includes real time and non real time forecast of active and reactive power using substation historic data. Though forecast data for active and  reactive powers are used for economic dispatch and reactive control planning and  configuration,  the work does not delve into such areas.

₦5,000.00 – DOWNLOAD FULL PROJECT DOWNLOAD FULL PROJECT Added to cart

---

---

---

---

---

Related Project Topics :