Abstract
Expanding the size of the power system, the complex structure, running status is
constantly changing, especially the high-voltage transmission lines is the link between power
plants and between users, an important part of the power system, undertake the task of
conveying the electrical energy, but it is also the power system most prone to failure of one of
the links. Most of the power system fault occurs in transmission lines, transmission lines, in the
event of failure to give social and economic life have a tremendous impact, whether timely and
accurate fault locate, for the power system to maintain stability and economic operation crucial.
Therefore, the accurate positioning of the failure of the transmission line is an important
measure to ensure network security, stability and economic operation technically.
Series compensation devices in transmission lines can effectively reduce the reactance of
the transmission system, improving transmission capacity, increase system stability and reduce
the project cost of the transmission system, series compensation capacitors in the transmission
line has been widely applications. However, due to series compensation capacitance exists to
change the uniformity of the transmission line, the fault location has been a problem plaguing
the industry. Fast and accurate positioning of series compensated transmission line fault on the
safe and economic operation of power system has played an important significance.
In this paper, the analysis of the existing series compensated transmission line fault
location algorithm based on differential equations of mathematical models and neural networks
will improve the integration of a new fault location method. First, by the phase -mode
transformation, the phase value into the value of the modulus, and then calculate the fault
distance modulus regions; Then, in accordance with the actual transmission line model for
transmission line fault data samples, the use of data samples for training and testing the neural
network model, and trained neural networks for series compensated transmission line fault pre-
judgment in; Transmission line mathematical model of neural network pre-processing results
and improvement of differential equations to solve the point of failure, assuming that the point
of failure occurred in the series on both sides of the compensation capacitor, by calculating the
two fault targeting solution, and then use the line at both ends to obtain the fault point voltage
value, the data were calculated according to the principle of fault point voltage value should be
equal to determine the true root to exclude pseudo - root, and thus determine the location of the
point of failure. Finally, the use of different transition resistance, variety of different fault types
or short-circuit fault simulation results show that: the method of combining neural networks
and improve differential equations in series compensated transmission line fault location has a
strong practical and accurate sex.
