Artificial Neural Network Based Transmission Line Loss Assessment of Power Transmission System

Abstract

This paper presents the application of Artificial Neural Network for transmission line loss assessment of power transmission system. The proposed method is fast and yet accurate. Active and reactive powers of generators and loads, as well as the magnitudes of voltages at voltage-controlled buses are chosen as inputs to the ANN. System losses are chosen as the outputs. Training data are obtained by load flow studies, assuming that the state variables of the power system to be studied taken the values uniformly distributed in the ranges of their lower and upper limits. Load flow studies for different system topologies are carried out and the results are compiled to form the training set. The proposed algorithm is applied to a sample power transmission network and the numerical results are presented in this paper in order to demonstrate the effectiveness of this proposed algorithm in terms of accuracy and speed. It is concluded that the trained ANN can be utilized for both off-line simulation studies and on line calculation of demand losses. High performance has been achieved through complex mappings, by the ANN, between system losses and system topologies, operating conditions and load variations.

Keyword: Transmission Line Loss, Load Flow, Artificial Neural Network and Back Propagation Algorithm

1. Introduction

Transmission losses are inherent in power systems. Utility companies are constantly installing and upgrading their transmission lines, transformers, capacitors etc., in order to reduce the losses. In planning system expansion, for example when there is a need for transmission reinforcement to serve more load demand, costs of different transmission systems designs are often compared with savings from loss reductions. For economic operation of a power system, transmission losses are taken into account in scheduling generation. Traditionally, electric utilities have considered system losses in a broad and general manner after data is compiled by allocating a % of total system losses to each of the system components (e.g. transmission, step-up and step-down transformers, distribution and others). Accordingly, loss calculations and in particular energy loss calculations, have been approximate at best. However, utilities are operating in a new environment, which on the one hand complicates the loss calculations and on other hand demands a fair assessment of losses. Indeed, the importance of accurately and rapidly determining transmission system losses has become more recognized and emphasized in the past several years for many reasons, including; i) There is a continuous need by the electric utilities to improve system

operating efficiency. ii) In many instances, electrical losses are the controlling factor while evaluating alternative power system expansion plans. iii) The current impetus in billing and rate design is to consider actual incurred costs. iv) Losses are important considerations in the pricing of all energy transactions among inter connected utilities. Hence, importance of accurate and rapid evaluation of the capacity or demand losses (MW) and energy losses (MWh) in a power system has more recognized now. A number of approximate methods for estimating transmission system losses have been proposed and used by utilities. However, high accuracy and fast speed are two conflicting requirements in a loss evaluation methodology, as more detailed modeling is necessary for more accurate evaluation. This work presents a new approach with which satisfactory performance in terms of these two requirements can be achieved. Power system losses include those at the various levels of transmission systems (i.e. bulk transmission, transmission and sub-transmission) and in the distribution networks. Due to constant variation of the operating conditions and the complexity of power systems, an accurate and rapid computation of the losses has always been a challenge to utility engineers.

2. Review of Literatures

Depending on the purpose of the study, different approaches emphasize different aspects of the problem. For bulk transmission lines a two-level loss evaluation procedure [1] is in use where accurate capacity losses can be calculated using an optimal power flow (OPF) program if elaborate computer facilities are available and approximate results can be obtained using a simplified version of the OPF approach. Energy losses are computed by integration after identifying numerous operating modes and forming the loss versus load relationships. Doing so will sacrifice the speed, especially when computer resources are insufficient. For distribution and sub-transmission systems, where losses would be a function of load (current), losses have been calculated by applying standard empirical equivalent hour loss factor equation [2, 3]. This method cannot be applied to transmission system where losses are function of generation schedule; imports, exports, wheeling and loop flow in addition to load. As the capacity losses of a power system are complex functions of the system configuration, generation and demand pattern as well as the various voltage levels at which the system is operating, a more complicated mapping capability is needed to approximate these functions. It is noted that the energy loss is the integration of the capacity loss over time. As the capacity loss is changing constantly, energy loss evaluation can’t be accurate unless it is done on-line. To realize the complex mapping capability on-line, parallel information processing is a necessity in an algorithm for this purpose. ANN has these capabilities and is suitable to model the complicated mapping relationships.

3. Proposed Approach

The proposed approach presents the application of ANN for transmission line loss evaluation in a power transmission network. It includes problem formulation and design of ANN for loss evaluation. Results demonstrating the performance of the proposed technique for a sample power system

4. Artificial Neural Network

An Artificial Neural Network is a computing system made up of number of simple and highly interconnected processing elements which process information by its dynamic state response to external inputs. In recent times the study of ANN model has gain rapid and increasing importance because of the potential to offer solutions to some of the problems which have hitherto been intractable by standard serial computers in the areas of computer science and artificial intelligence. Instead of performing a program of instruction sequentially neural net models explore many computing hypothesis simultaneously using parallel net composed of many computational elements. No assumptions will be made because no relationships will be established. Computational elements in neural networks are non-linear models and are also faster. Hence the result comes through non-linearity due to which the result is very accurate than other methods. Because of these reasons neural networks find their applications in achieving human like performance in the fields such as speech processing, image reorganization, machine vision, robotic control etc.

4.1Multilayer Feed Forward Neural Network

The figure1 shows the schematic representation of a Feed Forward Network, which is commonly used in ANN model. Processing elements in the ANN are called neurons. These neurons are interconnected by Information channels. Each neuron can have multiple inputs but only one output as shown in figure2. Inputs to the neuron can be from external stimuli or from the output of other neurons. There is an interconnection strength called weight associated with each connection. When the weighted sum of the inputs to the neuron exceeds a certain threshold, the neuron is fired and output signal is produced. The neurons are divided into several layers; one input layer; one output layer; some hidden layers.

Please Download the Paper to read more

Attached Files:

• (186 KB)

Attached Files:

• dip13704.pdf (252 KB)