Intelligent sensing: neural network based health diagnosis for sensor arrays

The paper discusses the appropriateness of local sensor health monitoring/sensor fault diagnosis and suggests an artificial intelligence (AI) based solution to the diagnosis of sensor fault, in the context of sensor arrays. In contradistinction with the "centralized" methods currently used process/system fault detection, the work presented here aims to establish a process independent and sensor specific methodology for detecting and isolating sensor failures and fault recovery. The intelligent system proposed integrates the adaptability and learning ability of artificial neural networks with effective inter-sensor communication protocols. Micromachined accelerometers are considered as a case study.