The Theory Of Information Fusion And Its Application Research On Drum Level Sensor Fault Diagnosis

Drum level sensor is very important equipment in the system of boiler, but there are many deficiencies in its reliability examining. Information fusion is a rising study, and it will be more and more popular. Through constructing the RBF network approximation facility, this paper proposes a novel fault diagnosis structure and method applying RBF network to the fault diagnosis of drum level sensor. RBF network is trained firstly by the samples of all kinds of correlative parameters determining the drum level. Then the RBF network approximation facility meeting the expectation of training error can estimate the drum level high accurately. The discrepancy between the output of the RBF network and the actual value of the sensor is compared with the predetermined threshold to detect the failures of drum level sensor; if a failure were diagnosed, through using a switch to achieve the reconfiguration for control system of drum level, the fault diagnosis system could eliminate the fault effects in time. Simulation results show this fault diagnosis method can very effectively achieve state monitoring and diagnosis of drum level sensors and has a good applied future.