The possibility of suppressing noise in trace long lines through the use of adaptive filtering algorithms and the use of expert systems technology to improve the reliability of the identification of the connected devices. The results of experiments confirming the effectiveness of the developed techniques.
Analysis of the current level of development and application of pulse reflectometry method for troubleshooting and identification of devices connected to a long line showed that the measurement results are greatly affected by natural and man-made noise, as well as the state of the line. As a result, the useful signal reflected from eavesdropping information devices and line discontinuities due to the presence of damaged areas can greatly distorted. As a result, errors occur in detection and identification of the fault line, and devices connected to it. The aim of the work is to improve the accuracy of pulse reflectometry long lines and the reliability of the identification of the connected devices. If the behavior studies have shown that a positive effect can be achieved through the use of adaptive filtering algorithms of the received signal and the use of expert systems technology. An experimental stand of the software and hardware that allows you to carry out modeling of identifying devices connected to the line and investigate the properties of traces. To increase the value of the signal / noise ratio applied adaptive filtering of the received signal. In the simplest case, the adaptive device comprises a programmable filter and the data processing unit that implements the adaptation algorithm, which is based on a priori information configures the programmable filter coefficients. To solve the problem selected gradient algorithm adaptive signal processing using the minimum criteria of the standard deviation.
For making decision on existence of a signal and its belonging to this or that type the expert system has been offered. It consists of the block of formation of aprioristic data, the knowledge base, the block of formation of a stream, the database, the block of entrance data. In work the special place of the mechanism of a logical conclusion in structure of expert system which realizes algorithms of a direct and return output is allocated. This mechanism can be presented by the sequence of procedures.
During the performed work the possibility of suppression of noise in the reflektogrammakh of long lines on the basis of application of algorithms of an adaptive filtration and use of technology of expert systems for increase in reliability of identification of the connected devices is shown.
Result of researches was development of expert system which will allow to diagnose not only existence of not uniformity in the long line, but also to carry out identification of the connected devices.
