A PC-based signal validation system for nuclear power plants

In order to achieve the desired operating configuration in any process, the system conditions must be measured accurately. Examples of measurements are temperature, pressure, flow, level, motor current, vibration, etc. However, in order to operate within desired limits, it is important to know the reliability of plant measurements. Signal validation (SV) deals with this issue, and is defined as the detection, isolation and characterization of faulty signals. Also referred to as fault detection, signal validation checks inconsistencies among redundant measurements and estimates their expected values using other measurements and system models. The benefits of SV are both economic and safety related. Catastrophic signal failure can result in plant shutdown and lost revenue. Pre-catastrophic failure detection would therefore minimize plant downtime and increase plant availability. The control action taken depends primarily upon the information provided by the plant instruments. Thus, increased plant productivity and increased reliability of operator actions, would result from the implementation of such a system. The purpose of this study is to investigate some of the existing signal validation methods by incremental improvements and to develop new modules. Each of the SV modules performs a specific task. The architecture consists of four modules, an information base and a system executive integrated with a graphical user interface (GUI). All the modules are used for validation during both steady-state and transient operating conditions. The entire system was developed in the PC-framework under Microsoft WindowsTM. Some improvements were made in the structure of static data-driven models by incorporating one and two-step regression. Kalman filtering is based on the use of a physical model of plant components and was implemented for a steam generator system in a nuclear power plant. This is applicable to both steady-state and transient operations. The system executive performs several tasks: sequencing of module operation, requisition of additional data, evaluating SV information from the various modules, and displaying instrument or system status to the operator. The decision-making within the system executive was developed using a robust fuzzy logic approach. The computer display was performed by GUI objects compatible with Microsoft Windows. (author)