Machine Learning Approaches for Sensor Fault Mitigation

Wireless Sensor Network (WSN) technology has proven great potential in industrial, commercial, and consumer applications specifically in the field of process monitoring and control. The adoption of WSN for process monitoring and control delivers prodigious rewards over the conventional wired system. However, the low-cost sensors may create faulty measurements or fail completely. This might introduce serious glitches as the sensor data obtained is crucial. In this paper, an efficient method is proposed to predict the failed sensor data in a Wireless Sensor Network setup and compensate for sensor failures using machine learning and deep learning techniques. For implementation of the proposed algorithm the dataset from an industrial sensor network setup is considered for implementation.