Gil fault on-line monitoring system based on vibration signals and support vector machine

A GIL fault on-line monitoring system based on vibration signals and a support vector machine, comprising:

• a vibration sensor secured to a shell of a post insulator of a GIL to be monitored, and a data collector connected to an output end of the vibration sensor;
• the data collector is connected to a wireless gateway through a wireless manner and is connected to a terminal computer via the wireless gateway;
• through a support vector machine-based (SVM) fault recognition model, the terminal computer is configured to determine whether the GIL is experiencing a fault, identifying a type of the fault and locating the fault;
• and for diagnosing the GIL fault, the support vector machine is configured to perform following steps:
• 1) acquiring vibration acceleration signals:
• acquiring vibration acceleration signals under different GIL states, and classifying by using types of the GIL states as labels and using corresponding vibration acceleration signal data as corresponding data samples;
• 2) extracting feature parameters:
• extracting feature parameters of the vibration acceleration signals in a time domain, a frequency domain and a time-frequency domain;
• 3) constructing and training the SVM fault diagnosis model:
• performing feature extraction on the vibration acceleration signals at the different GIL states, and constructing and training the SVM fault diagnosis model by means of an LIBSVM toolkit, and a mathematical expression of the SVM fault diagnosis model is as follows:
• K
• 
• (
• x
• ,
• x
• i
• )
• =
• exp
• 
• (
• -
• 
• x
• -
• x
• i
• 
• 2
• 2
• 
• σ
• 2
• )
• ,
• wherein a is a kernel width parameter;
• and
• 4) based on the SVM fault diagnosis model, diagnosing the fault, identifying the type of the fault and locating the fault.