F-P fiber optic ultrasonic sensor with high sensitivity

strong anti-interference ability

good electrical in-sulation and many other advantages

can be detected in the oil discharge of the ultrasonic signal. In order to identifythe fault type of the partial discharge signal measured by the F-P fiber ultrasonic sensor

four typical partial dis-charge models of oil wedge discharge model

suspension discharge model

tip discharge model and quay dischargemodel are established

and the related experimental study is carried out. In this paper

aiming at the ultrasonic pulsewaveform

a method of extracting characteristic parameters of signal is studied

and the extraction of characteristicparameters is realized. Then

the probabilistic neural network is used to identify the ultrasonic signal and analyze therecognition effect. The four kinds of partial discharge models can generate stable ultrasonic signals

meet the test re-quirements and extract the characteristic parameters of ultrasonic pulse waveforms. The pattern recognition and anal-ysis of the probabilistic neural networks are carried out. The results show that the recognition effect is good and therecognition is high correct rate.