An ISO/IEC/IEEE 21451 Compliant Algorithm for Detecting Sensor Faults

• Published in: IEEE sensors journal Vol. 15; no. 5; pp. 2541 - 2548
• Main Authors: Morello, Rosario, De Capua, Claudio
• Format: Journal Article
• Language: English; Japanese
• Published: IEEE 01.05.2015
• Subjects: Accuracy; Atmospheric measurements; automated testing; Calibration; Dispersion; maintenance; Reliability; Reliability assurance; Sensors; repeatability; Reliability assurance; accuracy; automated testing; maintenance; calibration; sensor fault
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2014.2361697

Preservation of metrological properties of any measuring and sensing system is an important issue to assure reliable measurements. So, in particular, calibration is the confirmation process used in the practice to verify the metrological properties of any transducer and sensor. Nevertheless, calibration needs specific procedures, measurement instrumentation, and skills. It involves the use of resources and costs to be supported. Therefore, preliminary tests are suggested to define when calibration is needed. In this sight, a transducer or sensor can be smart by verifying its status so to detect possible faults and estimate autonomously the next calibration time. As a result, the development of algorithms for automated fault detection and maintenance is an important target for developers engaged to project smart transducers and sensors. In particular, built-in algorithms can allow the same sensor to check its reliability before performing any measurement. In this paper, we propose an algorithm for detecting sensor faults. The aim is to provide a sensor-friendly algorithm for detecting failures by performing intermediate tests in view of the next calibration. The algorithm has been developed to be in compliance with ISO/IEC/IEEE 21451 standard. In detail, the algorithm allows sensor to estimate its accuracy and the repeatability of measurement so to get information on measurement reliability. Depending on the specific sensor, few points of calibration curve are checked and compared with reference values. If the estimated accuracy or repeatability are below the specification reported in the last calibration data, sensor apprises that a new calibration is required. Based on the general features, the algorithm can be used for any sensor.