Introduction

VOLTAGE STABILITY FORECASTING USING ARTIFICIAL NEURAL NETWORK

PRESENTED BY

A.RAMAKRISHNA

N.SRIKANTH

K.VAMSHI

S.RAJESH GOUD

Introduction

Voltage stability is concerned with the ability of a power system to maintain acceptable voltages at all buses in the system under normal conditions and after being subjected to a disturbance.

Buses with values of voltage stability factors close to 1 .0 are identified as the critical buses. And after that, an Artificial Neural Network is developed for voltage stability monitoring.

Basic Structure of a Power System

CLASSIFICATION OF BUSES

Preparation Of Data For Load Flow

Let real and reactive power generated at bus- i be denoted by PGi and QGi respectively. Also let us denote the real and reactive power consumed at the i th th bus by PLi and QLi respectively. Then the net real power injected in bus- i is

Let the injected power calculated by the load flow program be Pi, calc . Then the mismatch between the actual injected and calculated values is given by

Load Flow By Newton-Raphson Method

The approach to Newton-Raphson load flow is similar to that of solving a system of nonlinear equations using the Newton-Raphson method: At each iteration we have to form a Jacobian matrix and solve for the corrections from an equation of the type given in . For the load flow problem, this equation is of the form

Load Flow Algorithm

The Newton-Raphson procedure is as follows:

Step-1: Choose the initial values of the voltage magnitudes |V| (0) of all np load buses and n 1 angles (0) of the voltages of all the buses except the slack bus.

Step-2: Use the estimated |V|(0) and (0) to calculate a total n 1 number of injected real power Pcalc(0) and equal number of real power mismatch P (0) .

Step-3: Use the estimated |V| (0) and (0) to calculate a total np number of injected reactive power Qcalc(0) and equal number of reactive power mismatch Q (0) .

Step-3: Use the estimated |V| (0) and (0) to formulate the Jacobian matrix J (0) .

Step-4: Solve for (0) and |V| (0) |V| (0).

Step-5 : Obtain the updates from

Step-6: Check if all the mismatches are below a small number. Terminate the process if yes. Otherwise go back to step-1 to start the next iteration with the updates given by above equations.

Comparision between newton raphson and fast decoupled method

The main computational effort of the fast decoupled method is the calculation at each iteration of the mismatch vector change in real power and change in reactive

Consequently the first decupled method is much faster for normal systems and for moderate accuracy

Typically a NR iteration takes around five times as long as a fast decoupled iteration. However the fast decoupled method requires more iterations than the NR method, taking in the order of two times as many iterations for normal power systems with normal loading conditions.

Artificial neural network

In most cases an ANN is an adaptive system that changes its structure based on external or internal information that flows through the network during the learning phase. Modern neural networks are non-linear stastical data modelling tools. They are usually used to model complex relationships between inputs and outputs or to find patterns in data.

An ANN is typically defined by three types of parameters:

The interconnection pattern between different layers of neurons

The learning process for updating the weights of the interconnections

The activation function that converts a neuron's weighted input to its output activation.

Principles of training multi-layer neural network using backpropagation algorithm

Newton Raphson Outputs with Change of load

Per Unit inputs To Artificial neural network

SELECTION OF INPUT AND TARGET DATA

CREATING NEURAL NETWORK

ANN ARCHITECTURE

The basic architecture consists of three types of neuron layers: input, hidden, and output.

Ann outputs:

Weights And Bias:

This is impractical and it would be easier if only one of the parameters should be adjusted. To cope with this problem a bias neuron is invented.

The bias neuron lies in one layer, is connected to all the neurons in the next layer, but none in the previous layer and it always emits 1.

Since the bias neuron emits 1 the weights, connected to the bias neuron, are added directly to the combined sum of the other weights , just like the t value in the activation functions.

GRAPHS

Comments:

1 All Buses are stable up to 110MW

2 Bus 4 is severely affected due to change in load at bus4

CONCLUSION

ANN is obtained for computing a certain quantity, its computation is simply done by an algebraic calculation using weights and coefficients. This is much simpler than the numeric methods. The most complex procedure is the training, and this can be compared to the numeric methods in terms of their complexity.

The proposed ANN architecture is very efficient as far as error control, since the ANN corresponding to each bus can be trained separately.

The simulation results showed a very good performance of the ANN for solving the problems that were posed.

FUTURE SCOPE:

Further the project can be extended from 14 bus system to 30 bus system and also the change in reactive power at load bus can also be considered. Thus stability of buses can be determined using artificial neural network.

THANK YOU