چكيده به لاتين
Today, readability and safety are a vital feature of instrument and control systems that are used in high-risk situations such as aircraft, nuclear facilities, ships and submarines, and etrochemical, oil and water industries.
the purpose of the fault detection is to know the fault occurrence and use effective measures to continue to work properly after the fault occurs .
A lot of expenses have been spent to identify the fault and prevent the occurrence of the incident and the reason is damages that the fault occurring in the systems that these losses may not be compensated at some time. the consistency and performace of the system depends on the accuracy of the measurements, and this leads to the attention of some researchers with the problem of sensor fault. Research has clearly shown the effect of sensor quality measurement on system performance, since sensor information is continually required to control the closed loop system.
in this study, we intend to detect the occurrence of fault in current and position sensors of elctromechanical actuator using neural network method.
a simple model of electromechanical actuator is developed by matlab simulink.Then, in order to evaluate it, the performance of the model is compared with the characteristics of a real actuator that is extracted from its specifications from the manufacturer's data bases.Five types of fault ,(bias , drift, scales, noise and stuck at zeroe) are considered for sensors of actuator .using three neural networks, RBF, GRNN, and feedforwards faults are detected and performance of these networks are evaluated regarding fault detection.Also, use of two characteristics ,mean and standard deviation as pre-processing of collected data was selected and the effect of each on neural network performance was evaluated.
