Selection of Sample Rate and Computer Wordlength in Digital Instrumentation andControl Systems

摘要

Digital sampling of analog signals add two types of errors, aliasing and finitewordlength error, to the sampled version of the signal. Aliasing is characterized by high frequency components misrepresented as low frequency components in the sampled signal It is greatly influenced by the sample rate, and may lead to degraded performance in monitoring, alarm, control, and protection systems. Since computer wordlengths are finite in length, digital systems are limited in their capability to represent real number values. Finite wordlength errors related to round-off, truncation, and data conversion have the potential to adversely impact the performance of digital instrumentation and control (I&C) systems. The Office of Nuclear Regulatory Research is investigating good engineering practices and review guidance regarding aliasing and finite wordlength errors in nuclear facilities. Hazards associated with these errors are minimized through proper design and selection of sample rates and computer wordlengths. This document provides the regulatory background, theoretical information, practical issues, best engineering practices, and examples associated with sample rate and computer wordlength selection. This information is used by NRC staff to identify improper treatment of aliasing and finite wordlength error in digital I&C systems.