THE RELIABILITY AND ACCURACY OF THE TRIPLE MEASUREMENTS OF ANALOG PROCESS VARIABLES

Abstract

The increase in unit capacity of electric equipment as well as complication of technological processes, devices control and management of the latter in power plants and substations demonstrate the need to improve the reliability and accuracy of measurement information characterizing the state of the objects being managed. The mentioned objective is particularly important for nuclear power plants, where the price of inaccuracy of measurement responsible process variables is particularly high and the error might lead to irreparable consequences. Improving the reliability and accuracy of measurements along with the improvement of the element base is provided by methods of operational validation. These methods are based on the use of information redundancy (structural, topological, temporal). In particular, information redundancy can be achieved by the simultaneous measurement of one analog variable by two (duplication) or three devices (triplication i.e., triple redundancy). The problem of operational control of the triple redundant system of measurement of electrical analog variables (currents, voltages, active and reactive power and energy) is considered as a special case of signal processing by an orderly sampling on the basis of majority transformation and transformation being close to majority one. Difficulties in monitoring the reliability of measurements are associated with the two tasks. First, one needs to justify the degree of truncation of the distributions of random errors of measurements and allowable residuals of the pairwise differences of the measurement results. The second task consists in formation of the algorithm of joint processing of a set of separate measurements determined as valid. The quality of control is characterized by the reliability, which adopted the synonym of validity, and accuracy of the measuring system. Taken separately, these indicators might lead to opposite results. A compromise solution is therefore proposed. The quality of the evaluation of the measured signal is characterized by a single comprehensive measure that takes account of both reliability and accuracy properties of the system. This indicator is the average precision measure which is the weighted average error of the various possible states of a group of three devices.