Fault detection and recovery for full range of hydrogen sensor based on relevance vector machine

In order to improve the reliability of hydrogen sensor, a novel strategy for full range of hydrogen sensor fault detection and recovery is proposed in this paper. Three kinds of sensors are integrated to realize the measurement for full range of hydrogen concentration based on relevance vector machine (RVM). Failure detection of hydrogen sensor is carried out by using the variance detection method. When a sensor fault is detected, the other fault-free sensors can recover the fault data in real-time by using RVM predictor accounting for the relevance of sensor data. Analysis, together with both simulated and experimental results, a full-range hydrogen detection and hydrogen sensor self-validating experiment is presented to demonstrate that the proposed strategy is superior at accuracy and runtime compared with the conventional methods. Results show that the proposed methodology provides a better solution to the full range of hydrogen detection and the reliability improvement of hydrogen sensor.