Deterioration of composite insulators in power grid results in degradation of their electrical insulation performance

harming safe operation of the transmission lines.The existing working state detection methods of composite insulators are insufficient.The authors analyze the existing detection methods of composite insulators in the world

and introduce hyperspectral remote sensing technology to working state detection of composite insulator.Hyperspectral camera is adopted to acquire the hyperspectral images of silicon rubber composite insulators in different working states

and a characteristic spectrum library of composite insulators is thus constructed.The spectrum curves of the composite insulators in different working states and their clustering property in characteristic space are displayed.The characteristic spectrum library is then employed to detect the working state of the target insulator by using the probability decision function and the Bayesian discriminant function.The result shows that the present method can accurately detect working state of composite insulators.