Practical diagnostic algorithms for run-time systems

Abstract

In a real-time system, it is plausible to implement simple and effective diagnostic algorithms for run-time health monitoring of a physical system. Since the operating environment is real-time, the periodic timing poses a critical constraint to the detection system in which the diagnostic algorithms reside. Depending on the nature and characteristics of a real-time system, this timing constraint can range between tens of microseconds to tens of milliseconds. This implies that a real-time diagnostic algorithm must be executed within the allotted time period. In addition to the timing constraint, a system designer also pays great attention to the environment-induced noise. Typically, sensors are set to work as bandpass filters and would work properly within the cut-off bandwidth. Thus, a key element in real-time health monitoring is to ensure that the sensor filtering is done properly in the presence of environmental noise. We present a few examples of diagnostic algorithms that can be embedded in a real-time system and provide a ready solution to time-critical diagnostic problems.