Fault Location in 22 kV Power Network by Using the Elements of Artificial Intelligence
 
 
  Project
   State-of-art
   Project outcomes
   Project objectives
   Originality of objectives
  Research team
  Contact
   
 
 
Current state-of-art in the field

There are several methods that have been developed for fault location in power networks, nevertheless only few of them can be used for fault location in 22 kV networks. The reason is that these networks have radial topology and they are operated as impedance earthed systems. These networks are mostly protected by overcurrent and earth fault protection relays, which are placed at the beginning of power lines and cooperate with automatic reclosing. Both types of used protection relays are not suitable for fault location in regard of their working principle.

The complexity of fault location in 22 kV networks is caused also by fallowing factors:

  • the measurement valid for whole substation is often used for fault location (busbar voltages and transformer currents),

  • if the current of faulty line is not directly measured, an assumption that transformer current during a fault is a current of the faulty line brings certain error into fault location calculation, moreover it is not possible to compensate accurately for the pre-fault load current of the faulty line,

  • the radial topology of 22 kV network does not allow, due to used one-end fault location method, to define single fault position, because several alternatives appear as possible ones,

  • the loads located between the fault point and the busbar change in time, what presents the problems in fault location calculations,

  • 22 kV lines consists of both cable and overhead sections what complicates the formation of adequate representation of equivalent scheme for fault location calculation.

Above mentioned problems are the main reasons, why it is not possible to use fault location method commonly used by higher voltage levels for fault location in 22 kV network.

The fault location in Slovak power distribution systems is based on bisecting method. At the beginning of fault location the dispatcher bisects the faulty line into two disconnected segments by the help of remote controlled circuit breakers in power substations and remote controlled section switches placed in the line. Then he tries to feed one of these segments, at first the one closer to the power substation. If the fault is situated in selected segment, protection relays turn off the segment again. If the segment stays to be fed than the bisection is applied to other segment. And so it continues till precise fault localization. If the dispatcher does not manage to locate the fault by the help of remote controlled section switches, there are only few of them located in the line, he is obliged to send maintenance crews to look for the fault personally. The crews switch on or off locally controlled section switches according to instructions obtained form dispatcher via radio communication. The time periods, which the crews need for traveling between locally controlled section switches, can cause that the average time needed for fault localization takes up to 70 % of total time from fault detection to electricity supply restoration.

A few other fault location methods have been developed in scientific world, but their function and accuracy was tested only by simulations. These methods are based on:

  • Symmetrical components – the localization is based on measurement of voltages and currents during the fault. They are used for the calculation of fault impedance, which is decomposed to symmetrical components - positive sequence, negative sequence and zero sequence. Symmetrical components are then applied to form a fault circuit equivalent scheme that is used in particular algorithm to calculate the fault position.

  • Steady state measurement evaluation - this method tries to eliminate the influence of transients’ measurement and fault resistance on fault location. It uses symmetrical components of voltages, currents and line impedance before and after the fault. The fault location calculation is based on the mathematical expression of voltage drop across fault impedance, which is possible to calculate from the difference of voltages at the begging and at the end of the line, by taking the voltage drop across the line into calculation. The advantage of this method is that it is possible to repeat the measurement needed for fault location.

  • Using H matrices - a power line is divided into n sections, which are substituted by quadripole equivalents during the calculation (e.g. T- network). For each quadripole equivalent the corresponding H matrix is expressed. H matrices representing line sections can be easily connected in series. The line consisting of n sections could be imaginary divided into first m-1 sections without fault, m-th section with fault and n-m sections without fault. The fault location calculation is based on the fact that a matrix product is not commutative. Then it is possible to calculate the H matrix, from known voltages and currents at the beginning and the end of line, which is consequently used for calculating the fault location.

  • Transients’ analysis - a fault in power network generates a transient – an impulse that is traveling from the place of the fault to power system. This impulse can be measured and used for fault location. The other variation of this method is sending high frequency impulse in the line affected by the fault. The impulse travels to the place of the fault, where it bounces back due to the change of line impedance. The fault location can be estimated from the time between sent and received impulse, from the change of signal’s polarity and signal’s absorption.

Expert systems represent one field of artificial intelligence. Their task is to use stored knowledge and acquired facts to come to the same conclusions as human experts in specific area. Their application is useful by problems, which solution depends to a great extent on good knowledge of solved problem. Expert systems are in power electrical engineering mainly used for device diagnostic. Other fields of artificial intelligence, e.g. neural networks or genetic algorithms, are applied in area of protection relays or in optimization of power system operation (voltage regulation in nodal points).

 
[home] [project] [research team] [contact]
created (c) 2008 MZM DATA