artificial intelligence and neural network applications in power systems

Upload: bharadwaj-santhosh

Post on 09-Apr-2018

234 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    1/18

    ARTIFICIAL INTELLIGENCE AND NEURAL NETWORK

    APPLICATIONS IN POWER SYSTEMS

    Document BySANTOSH BHARADWAJ REDDY

    Email: [email protected]

    Engineeringpapers.blogspot.com

    More Papers and Presentations available on above site

    ABSTRACT:

    The electric power industry is

    currently undergoing an

    unprecedented reform, ascribable to,

    one of the most exciting and

    potentially profitable recent

    developments in increasing usage of

    artificial intelligence techniques. The

    artificial neural network approach has

    attracted number of applications

    especially in the field of power

    system since it is a model free

    estimator. Neural networks provide

    solutions to very complex and

    nonlinear problems. Nonlinear

    problems, like load forecasting that

    cannot be solved with standardalgorithms but can be solved with a

    neural network with remarkable

    accuracy. Modern interconnected

    power systems often consist of

    thousands of pieces of equipment

    each of which may have an effect on

    the security of the system. Neural

    networks have shown great promise

    for their ability to quickly and

    accurately predict the system

    security when trained with data

    collected from a small subset of

    system variables.

    The intention of this paper is to

    give an overview of application of

    artificial intelligence and neural

    network (NN) techniques in power

    systems to prognosticate load on

    power plant and contingency in case

    of any unexpected outage. In this

    paper we present the key concepts of

    artificial neural networks, its history,imitation of brain neurons

    architecture and finally the

    applications (load forecasting and

    contingency analysis). The

    applications of artificial intelligence

    mailto:[email protected]:[email protected]
  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    2/18

    in areas of load forecasting by error

    Backpropagation learning algorithm

    and contingency analysis based on

    Quality index have been

    perspicuously explained.

    INTRODUCTION:

    Modern power systems are required

    to generate and supply high quality

    electric energy to customers. To

    achieve this requirement, computers

    have been applied to power system

    planning, monitoring and control.

    Power system application programs

    for analysing system behaviours are

    stored in computers. In the planning

    stage of a power system, system

    analysis programs are executed

    repeatedly. Engineers adjust and

    modify the input data to these

    programs according to their

    experience and heuristic knowledge

    about the system until satisfactory

    plans are determined.

    For sophisticated approaches

    to system planning, development of

    methodologies and techniques to

    incorporate practical knowledge of

    planning engineers into programs

    which also include the numerical

    analysis programs are needed. In the

    area of power System monitoring and

    control, computer based Energy

    Management Systems are now

    widely used in energy control

    centers. The abnormal modes of

    system operation may be caused by

    network faults, active and reactive

    power imbalances, or frequency

    deviations. An unplanned Operation

    may lead to a mal or a complete

    system blackout. Under these

    emergency situations, power

    systems are restored back to the

    normal state according to decisions

    made by experienced operation

    engineers. There is also a need to

    develop fast and efficient methods

    for the prediction of abnormal

    system behaviour.

    Artificial intelligence (AI)

    has provided techniques for

    encoding and reasoning with

    declarative knowledge. The advent

    of neural networks (NN"s), in

    addition, provides neural network

    modules which can be executed in

    an online environment. These new

    techniques supplement conventional

    computing techniques and methods

    for solving problems of power

    system planning, operation and

    control.

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    3/18

    Areas of Applications:

    Possible applications of artificial

    intelligence in power system planning

    and operation were investigated by

    power titles and researchers. In the

    last decade, many artificial

    intelligence systems and expert

    systems have been built for solving

    problems in different areas within the

    field of power systems only. These

    areas are summarized below.

    System planning

    Transmission planning and design,

    Generation expansion, Distribution

    planning.

    System Analysis

    Loadflow engine, Transient stability

    System Operation h Monitoring

    Alarm processing, Fault diagnosis,

    Substation monitoring, System and

    network restoration, Load shedding,

    Voltage / reactive power control,

    Contingency selection, Network

    switching, Voltage collapse.

    Operational Planning

    Unit commitment, Maintenance

    scheduling, Load forecasting.

    1. Artificial Neural Networks

    1.1 What is a Neural Network?

    An

    Artificial

    Neural

    Network

    (ANN) is an

    information

    processing paradigm that is inspired

    by the way biological nervous

    systems, such as the brain, process

    information. The key element of this

    paradigm is the novel structure of

    the information processing system.

    It is composed of a large number of

    highly interconnected processing

    elements (neurons) working in

    unison to solve specific problems.

    ANNs, like people, learn by

    example. An ANN is configured for

    a specific application, such as

    pattern recognition or data

    classification, through a learning

    process. Learning in biological

    systems involves adjustments to the

    synaptic connections that exist

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    4/18

    between the neurons. This is true of

    ANNs as well.

    The major breakthrough in the

    field of ANN occurred with the

    invention of Backpropagation

    algorithm which enabled design and

    learning techniques of multilayered

    neural networks. Since then the

    development and areas of application

    in which ANN is applied has been

    thriving.

    1.3 Biological inspiration:

    The brain is principally composed

    of a very large number (circa

    10,000,000,000) of neurons,

    massively interconnected (with an

    average of several thousand

    interconnects per neuron, although

    this varies enormously). Each neuron

    is a specialized cell which can

    propagate an electrochemical signal.

    The neuron has a branching input

    structure (the dendrites), a cell body,

    and a branching output structure (the

    axon). The axons of one cell connect

    to the dendrites of another via a

    synapse. When a neuron is activated,

    it fires an electrochemical signal

    along the axon. This signal crosses

    the synapses to other neurons, which

    may in turn fire. A neuron fires only

    if the total signal received at the cell

    body from the dendrites exceeds a

    certain level (the firing threshold).

    To capture the essence of biological

    neural systems, an artificial neuron

    is defined as follows:

    It receives a number of inputs

    (either from original data, or from

    the output of other neurons in the

    neural network). Each input comes

    via a connection that has a strength

    (or weight); these weights

    correspond to synaptic efficacy in a

    biological neuron. Each neuron also

    has a single threshold value. The

    weighted sum of the inputs is

    formed, and the threshold

    subtracted, to compose the

    http://www.statsoft.com/textbook/glosn.html#Neuronhttp://www.statsoft.com/textbook/glosn.html#Neuronhttp://www.statsoft.com/textbook/glosn.html#Neuronhttp://www.statsoft.com/textbook/glosn.html#Neuronhttp://www.statsoft.com/textbook/glosn.html#Neuron
  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    5/18

    activation of the neuron (also known

    as the post-synaptic potential, or PSP,

    of the neuron).

    The activation signal is passed

    through an activation function

    (also known as a transfer

    function) to produce the output of

    the neuron.

    If a network is to be of any use, there

    must be inputs (which carry the

    values of variables of interest in the

    outside world) and outputs (which

    form predictions, or control signals).

    The input, hidden and output neurons

    need to be connected together.

    1.4 Neural networks versus

    conventional computers

    Neural networks take a different

    approach to problem solving than that

    of conventional computers.

    Conventional computers use an

    algorithmic approach i.e. the

    computer follows a set of

    instructions in order to solve a

    problem.

    Neural networks process

    information in a similar way the

    human brain does. Neural networks

    learn by example. They cannot be

    programmed to perform a specific

    task. On the other hand,

    conventional computers use acognitive approach to problem

    solving; the way the problem is to

    solved must be known and stated in

    small unambiguous instructions.

    These instructions are then

    converted to a high level language

    program and then into machine code

    that the computer can understand.

    Neural networks and conventional

    algorithmic computers are not in

    competition but complement each

    other. Even more, a large number of

    tasks, require systems that use a

    combination of the two approaches

    (normally a conventional computer

    is used to supervise the neural

    network) in order to perform at

    maximum efficiency.

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    6/18

    1.5 Features

    1. High computational rates

    due to the massive parallelism.

    2. Fault tolerance.

    3. Training the network

    adopts itself, based on the

    information received from the

    environment.

    4. Programmed rules are

    not necessary.

    5. Primitive computational

    elements.

    1.6 The Learning Process

    Supervised learning which

    incorporates an external teacher, so

    that each output unit is told what its

    desired response to input signals

    ought to be. During the learning

    process global information may be

    required.

    Unsupervised learning uses no

    external teacher and is based upon

    only local information. It is also

    referred to as self-organisation

    1.7 Transfer Function

    The behaviour of an ANN depends on

    both the weights and the input-output

    function (transfer function) that is

    specified for the units. This function

    typically falls into one of three

    categories:

    Linear (or ramp)

    Threshold

    Sigmoid

    For linear units, the output activity

    is proportional to the total weightedoutput.

    Forthreshold unit, the output is set

    at one of two levels, depending on

    whether the total input is greater

    than or less than some threshold

    value.

    Forsigmoid units, the output varies

    continuously but not linearly as the

    input changes. Sigmoid units bear a

    greater resemblance to real neurons

    than do linear or threshold units, but

    all three must be considered rough

    approximations.

    2. APPLICATIONS

    App1: Power Systems Load

    Forecasting

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    7/18

    Commonly and popular problem that

    has an important role in economic,

    financial, development, expansion

    and planning is load forecasting of

    power systems. Generally most of the

    papers and projects in this area are

    categorized into three groups:

    Short-term load forecasting

    (STLF) over an interval

    ranging from an hour to a

    week is important for various

    applications such as unit

    commitment, economic

    dispatch, energy transfer

    scheduling and real time

    control. A lot of studies have

    been done for using of short-

    term load forecasting with

    different methods. One of

    these methods may be

    classified as follow:

    Regression model, Kalman filtering,

    Box & Jenkins model, Expert

    systems, Fuzzy inference, Neuro

    fuzzy models and Chaos time series

    analysis.

    Some of these methods have

    main limitations such as neglecting of

    some forecasting attribute condition,

    difficulty to find functional

    relationship between all attribute

    variable and instantaneous load

    demand, difficulty to upgrade the set

    of rules that govern at expert system

    and is ability to adjust themselves

    with rapid nonlinear system-load

    changes. The NNs can be used to

    solve these problems. Most of the

    projects using NNs have considered

    many actors such as weather

    condition, holidays, weekends and

    special sport matches days in

    forecasting model, successfully.

    This is because of learning ability of

    NNs with many input factors.

    Mid-term load

    forecasting(MTLF) that

    range from one month to five

    years, used to purchase

    enough fuel for power plants

    after electricity tariffs are

    calculated.

    Long-term load forecasting

    (LTLF), covering from 5 to

    20 years or more, used by

    planning engineers and

    economists to determine the

    type and the size of

    generating plants that

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    8/18

    minimize both fixed and

    variable costs.

    2.1.1 Overview of STLF

    Techniques:-

    A wide variety of

    techniques/algorithms for STLF have

    been reported in the literature (These

    procedures typically make use of two

    basic models peak load models and

    load shape modes.

    Standard Load Concept :- (Load

    Shape Model)

    The load forecasting is divided into

    two general parts; peak load model

    and load shape model .Former deals

    with daily or weekly peak load

    modeling & later describes load a

    discrete time series over forecasting

    intervals.

    Standard Load :-The

    standard load curve is

    produced once a day . It

    needs rescaling over time .

    The standard load

    characterizes the base load .

    It is calculated by using

    historical load data .The

    standard load calculation

    can be divided in two

    parts. The first one makes

    an average using all

    common days in the same

    period. The holidays are included

    with Saturdays and Mondays. The

    second part investigates on the

    particular characteristic for each day

    of the week, separately. For this a

    simple weighted moving average is

    made.

    Residual/Deviation Load:-The

    residual load is used to represent the

    most recent variation of the load .

    This value contains information for

    last 3 hours. Auto regressive and

    exponential smoothing are the most

    common methods used to calculate

    the deviation of load value.

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    9/18

    2.1.2 Artificial Neural network

    based short term load forecasting:

    The development of an ANN

    based STLF model is divided into two

    processes, the "learning phase" and

    the "recall phase". In learning phase,

    the neurons are trained using

    historical input &output data and

    adjustable weights are gradually

    optimized to minimize the difference

    between the computed and desired

    output. The ANN allows outputs to be

    calculated based on some form of

    experiences, rather than

    understanding the connection between

    input and output (or cause and effect).

    In recall phase the new input data is

    applied to the network & and its

    outputs are computed and evaluated

    for testing purpose. In the ANN based

    STLF model, a layered ANN

    structure (Input layer, Hidden layer,

    Output layer) is used. In this method

    the weights are calculated by a

    learning process using error

    propagation in parallel distributed

    processing. The STLF problem is

    formulated with the past data as the

    input data and the latest data are the

    desired output for training the

    network.

    An initial input data set is

    presented to ANNSTLF which

    adjusts the weight values for a

    minimum error. Following a new

    input data set is presented and the

    weight values are adjusted in

    accordance .The process finishes

    when the difference between target

    output and the found output for all

    the input sets is close to zero. The

    feed forward Multilayer Perceptron

    (MLP) neural network model is used

    for implementing the STLF model

    (ANNSTLF). Fig5 shows a MLP

    with single hidden 1ayer.Tlie

    advantage of this model is that it is

    able to learn highly non-linear

    mappings. The MLP model is

    trained by standard backpropagation

    training algorithm and developed by

    Rumelhart.

    2.1.3 Multilayer Perceptron and

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    10/18

    Its application in load forecasting:-

    The multi layer Perceptron and the

    associated backpropagation algorithm

    proposed a sound method to train

    networks having more than two layers

    of neurons. The learning rule is

    known as Backpropagation which is a

    gradient decent technique with

    backward error(gradient) propagation

    is depicted in Fig.6. The back

    propagation network in essence learns

    a mapping from a set of input patterns

    (e.g. extracted features) to, a set of

    output patterns ( e.g. class

    information ) .This network can be

    designed and trained to accomplish a

    wide variety of mappings. This ability

    comes from the nodes in hidden layer

    or layer of the network which learns

    to respond to features found in input

    pattern. The features recognized or

    extracted by the hidden units ( nodes)

    correspond to the correlation of

    activity among different input units.

    As the network is trained with

    different examples, the network has

    the ability to generalize over similar

    features found in different patterns.

    The hidden unit (nodes)must be

    trained to extract a sufficient set of

    general features applicable to

    instances ; which is achieved , thus

    avoiding overloading of network ,

    by terminating Learning once a

    performance pattern has been

    reached . The backpropagalion

    network is capable of approximating

    arbitrary mappings given a net of the

    output.

    The name backpropagation comes

    from the fact that the error (gradient)

    of hidden units are derived from

    propagating backward the errors

    associated with the output Units

    since the target values for hidden

    units are not given or it is defined to

    obtain the values of the desired

    output at hidden layer.

    Error Backpropagation:-

    The back propagation (or backup)

    algorithm is a generalization of the

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    11/18

    Widrow Hoff error correction rule. In

    the Widrow-Hoff technique an error

    which is the difference between what

    the output is and what it is supposed

    to be is formed and the synaptic

    strength is changed in proportion to

    error times the input signal in a

    direction which reduces the error. The

    direction of change in weights is such

    that the error will reduce in the

    direction of the gradient (the direction

    of most rapid change of the error).

    This type of learning is also called

    gradient search. In the case of

    multilayer networks, the problem is

    much more difficult.

    Choice of activation function

    The most common activation

    function used in multilayer perceptron

    is the sigmoid. The equation of the

    sigmoid function is

    The back propagation algorithm for

    network using the sigmoid 88

    activation function is described below

    .The equation of sigmoid function is

    Written as

    2.1.4 The Application of ANN to

    STLF & Results:-

    The ANNSTLF implements

    multilayer feed forward neural

    network which was trained by using

    backpropagation training algorithm.

    Naturally 24 hours data points leads

    to 24 input nodes in MLP model.

    Here 2 hidden layers are considered.

    MSEB data for the period Oct 94 to

    June 95 i.e of 35 weeks for

    development and implantation of the

    software was utilized.

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    12/18

    The Backpropagation

    algorithm with MLP model of

    Artificial Neural Network (A") is

    developed for the problem of short

    Term Load Forecasting (STLF) with

    a lead time of at least 24 hours. The

    best performance was obtained for the

    load forecasting for the Tuesday

    which gives the maximum and

    average percentage error of 2.00%

    and 0.20% respectively. This comes

    very close to the precision obtained

    by the human forecaster. The turningof the gain, momentum terms

    selection of the weights and threshold

    values play key role in convergence

    of the network. High values of the

    weights lead to the divergence and

    generally small values of the order of

    10-2 the yield better results.

    App2: Power System Contingency

    Analysis

    Contingency analysis and risk

    assessment are important tasks for the

    safe operation of electrical energy

    networks. During the steady state

    study of an electrical network any

    one of the possible contingencies

    can have either no effect, or serious

    effect, or even fatal results for the

    network safety, depending on a

    given network operating state.

    Load flow analysis can be used

    as a crisp technique for contingency

    risk assessment. However

    performing at run time the necessary

    load flow analysis studies is a

    tedious and time consuming

    operation. An alternative solution is

    the off-line training and the run-time

    application of artificial neural

    networks. This article aims at

    describing how artificial neural

    networks can be used to bypass the

    traditional load flow cycle, resulting

    in significantly faster computation

    times for online contingency

    analysis. A discussion over the

    efficiency of the proposed

    techniques is also included.

    2.2.1 What is contingency in

    power system?

    system contingency is defined

    as a disturbance that can occur in the

    network and can result in possible

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    13/18

    loss of parts of the network like

    buses, lines, transformers, or power

    units in any of the network areas.

    Load flow analysis is an adequate

    ans for studying the effect of a

    possible contingency on a given

    operating point of the network. It is

    often the case that experienced

    engineers, involved in operation of a

    given system, can guess effectively

    contingency without the support of

    numerical computations. This

    intuition of the operators is useful in

    supporting the initial selection of a

    list of possible contingencies, which

    then will be analysed using the

    described here technique.

    2.2.2. System Architecture

    A suitable way of studying the effects

    of contingencies on an electrical

    network is through the definition of

    representative operating points,

    creation of a relevant data base, in

    which parameters relating to these

    operating points is stored as these

    have been measured directly through

    network snapshots. Once a number of

    operating points is simulated, a list of

    contingencies to be studied upon is

    formed. Each contingency is applied

    on all operating points found in the

    database and then a power flow

    solution is attempted on the

    network. According to the results of

    the power flow solution the

    contingency applied of the specific

    operating point can be ranked as

    innocent, violating, or

    diverging / serious. The pre-

    contingency operating point

    parameters, various operating point

    indices and metrics, the contingency

    and the power flow result are next

    stored in a table per contingency.

    This contingency table constitutes a

    set of features and tuples that can be

    considered as suitable neural

    network input layer data elements if

    selected in any combination and

    after being statistically normalized.

    The power flow solution classifying

    any contingency for any operating

    point, is the output layer value of the

    neural network.

    Neural network training is a

    computer intensive work that needs,

    however, to be done only once. As

    soon as the neural network is trained

    for a contingency, the predictions

    about the effects of a contingency on

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    14/18

    any operating point can easily be

    deduced. The efficiency of the

    predictions depends on various

    factors such as the quality and the

    quantity of the training features, the

    type, complexity and connectivity of

    the neural network.

    2.2.4 Neural Network Input

    Feature Selection

    A wide range of electrical network

    parameters can be used for describing

    the network state. Some of them can

    be the network load level expressed

    as a percentage of the maximal

    network load, the number of lines, the

    cumulative rating of all lines, the

    cumulative active load, active

    generation, reactive load, reactive

    generation, apparent power etc. In

    recent bibliography there are

    references in more elaborate

    aggregates that yield better results

    when applied, such as the active

    apparent power margin index

    (expressed as the fraction of the

    flowing aggregate apparent power,

    over the aggregate MVA line

    transmission limits) and the voltage

    stability index. The voltage stability

    index is computed as the

    sensitivities of the total reactive

    power generation to a reactive

    power consumption, known as

    reactive power dispatch

    coefficients .

    Neural networks can be

    trained with any number of input

    features. The neural network

    training process can selectively

    overweight the most salient features

    and underweight the least significant

    ones. However, the selection

    procedure is time consuming for the

    training of the neural network, while

    after the training is complete, it is

    not always obvious which of the

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    15/18

    input nodes are of greater importance.

    Further more, the least important

    input layer nodes may add noise to

    the neural network training process.

    Bearing this in mind, a pre-selection

    of the neural network input nodes is

    of great use. This can be achieved

    through the use of statistical methods.

    The statistical methods that apply in

    the procedure of the selection of

    features are used in the classification

    theory. The classification of a set of

    training examples by two features in

    two classes is considered to be better

    when the sub-populations look

    different. the simplest test proposed is

    the test of separating two classes

    using just the means. A feature

    selection test from Means and

    Variances is also proposed:

    A and B are of the same feature

    measured for the classes 1 and 2 n1

    and n2 are the corresponding numberof cases sig is a significance level. In

    [4] the following measure for filtering

    features separating two classes is also

    proposed:

    where M1 and M2 are the vectors of

    feature means for class 1 and class

    2, C1 -1 and C2 -1 are the inverse ofthe covariance matrix for class 1 and

    class 2 respectively. For reasons of

    simplicity, a combination of bus and

    line losses only has been considered

    as a constituent element of a

    contingency under study. The four

    most salient features found were the

    aggregate reactive power generation,

    the voltage stability index, the

    aggregate MVA power flow and the

    real power margin index. This set of

    selected features has been used for

    the training and testing of the neural

    networks subsequently built.

    2.2.5 Quality index

    The quality index is a qualitative

    measure of the classification power

    of the neural network. It is an index

    that has been calculated for all

    simulations and applies on the idea

    that within the three classes ofcontingency states, the major

    difference can be considered to

    occur between two possible

    categories of contingencies:

    innocent and non-innocent

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    16/18

    contingencies. In order to compute

    this quality index (QA) the

    following formula has been used:

    where ai,j is the I-th element of the j-

    th column of the confusion matrix

    Ai,j. The confusion matrix Ai,j is a

    matrix of frequencies. For each

    element of the matrix ai,j the i index

    refers to predicted values, while the j

    index refers to real values. The values

    range from one to three denoting the

    three possible contingency cases: one

    in case of nonconvergence /

    potentially serious contingency, two

    in case of MVA and voltage

    violations and three in case of an

    innocent contingency.

    This technique is involves a

    tedious training phase, where a set of

    neural networks is created,

    corresponding to a given set of

    possible contingencies. The resulting

    set of ANNs demonstrate satisfactory predictive power in classifying the

    contingencies correctly at run time.

    The run time performance of the

    system is very good in terms of

    computational time and recourses

    requirements.

    2.5.6 Result

    Seven ANNs have been trained for

    predicting the severity of

    contingencies for the network of the

    island of Crete, the testing set

    performance of these ANNs was in

    the range of 57% to 96%, this

    performance did not seem to be

    affected by the split between the

    training and testing cases, as for

    both a 70-30 and a 85-15 split the

    results were simillar.

    while sensitivity analysis in terms of

    the ANN architecture demonstrated

    that the number of hidden nodes

    seem to have a serious effect on the

    performance of the network,

    suggesting use of more complex

    ANNs.

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    17/18

    The promising results of this study

    suggest application of similar

    techniques in other areas of security

    assessment of power systems and

    other industrial processes.

    4. BIBILIOGRAPHY:

    Neural networks

    1.PatrickK. Simpson, Artificial

    Neural Systems, Pergamon Press,

    Elmsford, N. Y., 1990.

    2.Special Issue on Neural Networks

    I: Theory and Modeling.

    Proceedings of the IEEE, September

    1990.

    Load forecasting

    3.Jacques de Villiers and Etienne

    Barnard, Backpropagation Neural

    Nets with One and Two Hidden

    Layers, IEEE Transactions on

    Neural Netcuorks, Volume 4, January

    1993, pages 13G144.

    4.Special Issue on Neural Networks

    11: Analysis, Techniques, and

    Applications, Proceedings of the

    IEEE, October 1990.

    D.C. Park, et al., Electric Load

    Forecasting Using an artificial

    Neural Network, IEEE

    Transactions on Power Systems,

    Volume 6, Number 2,May 1991,

    pages 442-449.

    5.Load Forecasting by ANN-IEEE

    computer applications in power

    systems. Duane D. Highley" and

    Theodore J. Hilmes "

    Contingency analysis:

    6.Mitchell T. M., Machine Learning,

    McGraw-Hill

    Series in Computer Science, 1997,

    p.81

    7 Grainger J. J., W D. Stevenson,

    Jr., Power System

    Analysis, McGraw-Hill, 1994, chap.

    9

    8 Wehenkel L.A., Automatic

    Learning Techniques in

    Power Systems, Kluwer Academic

    Publ., 1998, (p.

    210)

  • 8/7/2019 Artificial Intelligence and Neural Network Applications in Power Systems

    18/18

    3. Keywords:

    Artificial neural networks, contingency analysis, load forecasting, applications

    of ANN in power system, Artificial intelligence training and testing.

    Document By

    SANTOSH BHARADWAJ REDDY

    Email: [email protected]

    More Papers and Presentations available on above site

    mailto:[email protected]:[email protected]