Functional Consistency Checking Module for Signal Validation: Topical Report

摘要

This report discusses the basic principles of a functional consistency checking module for signal validation in nuclear plants. This module will add additional features to the signal validation system currently being designed at the University of Tennessee. The goals of the functional consistency module are to provide solutions to the common mode features and single or dual sensor failures. The basic principle lies in the comparison of the actual outputs and outputs derived from reference static or dynamics models. The only requirement needed for this method is that the plant must be decomposed into subsystems with observable inputs and outputs. The reference models can be either empirical or derived from physical laws. The main feature of this module is its capability to separate component failures from sensor failures. The failures which can be isolated are bias, drifts, absormal noise or changes in the dynamic sensor characteristics. The decision-making process relies basically on truth table look-up. The truth table contents are derived from an apriori knowledge of the failure causes. More elaborate decision making processes based on pattern recognition techniques or expert systems are proposed to improve the reliability of the decisions. A modified sequential probability ratio test is suggested to handle adequately the modeling errors. The proposed functional consistency checking is valid for both steady state and transient conditions. In the latter case, the transfer functions of the instrumentation channels must be taken into account for a fast and accurate failure identification. To conclude the report, anticipated benefits of the functional consistency checking are listed. (ERA citation 12:045384)