Fault diagnosis and condition surveillance for plant rotating machinery using partially-linearized neural network

Fault diagnosis and condition surveillance of rotating machinery in a plant is very important for guaranteeing production efficiency and plant safety. In a large plant, with an enormous number of rotating machines, condition surveillance and fault diagnosis for all rotating machines is not only time consuming and labor intensive, but the accuracy of condition judgment cannot be ensured. These difficulties may cause serious machine accidents and consequently great production losses. In order to improve the efficiency of condition surveillance and detect faults at an early stage, this paper proposes a method of condition surveillance and fault discrimination for rotating plant machinery using non-dimensional symptom parameters in a time domain and “Partially-linearized Neural Network” (PLNN), from which the state of a rotating machine can be discriminated automatically. The verification results of precise diagnosis for rolling bearings show that the PLNN can effectively distinguish bearing faults. The verification results for condition surveillance of rotating machinery in a real plant show that the PLNN correctly judges the machine state of the inspected rotating machine as normal or abnormal.