Detecting leaks and sensor biases by recursive identification with forgetting factors

In industrial processes, pipes and tanks may leak and sensors may have biases since corrosion, measuring noises and instrument faults exist. In order to maintain production in normal and safe conditions, detecting possible faults of production equipment on time is crucial. In the paper, a process model is proposed to describe a boiler tube leak problem. Based on this model, least-squares methods with constant and time-varying forgetting factors are presented to detect the leakage and sensor bias. The application in a boiler system shows that the proposed methods can detect the boiler tube leakage more effectively than the method without forgetting factors.