الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
An intelligent system is proposed for the fault diagnosis of hydraulic servovalves. A dimensionless mathematical model of a single-stage servovalve is derived, where, through manipulating certain parameters, corresponding faults can be simulated. Using ftizzy and neuro-fuzzy approaches, a fault identification scheme is constructed. The severity of the detected fault is assessed. Upon implementing the fuzzy approach, the• fault identification scheme is based on trial and observation of the data patterns resulting from the model. Processes of data clustering and network training replace the human factor in the case of the neuro-flizzy approach. The fuzzy system was capable of identifying the faults through the whole tested range. However, it could not be expanded to assess the severity of the fault. The neuro-fuzzy identification scheme was able to perform both tasks. Finally, the case of two simultaneous faults was investigated. A three-stage neuro-fuzzy fault detection scheme was able to identi1’ all single and simultaneous fault cases successfully. The main advantages of the implemented approach are ease of design, efficiency and feasibility.