A state-of-art method for solar irradiance forecast via using fisheye lens

The purpose of this investigation is to enable the solar irradiance forecast function implementing a common camera devise instead of specialized instrument thereby serve for other researches. Development of various simulated tools requires higher accuracy surrounding weather condition data. Previous studies mainly focus on the improvement of precision for professional monitor equipment i.e. total sky imager, which is limited to the scope of users. In this research, a fisheye lens graph is rectified following a particular algorithm based on the image forming principle. Moreover, solar irradiance prediction adopts the advanced BP neutral network method being proved to be valid. Final results indicate that after rectifying the special perspective images under fisheye direction, colour threshold configuration could remarkably recognize the cloud image. The conclusion shows that common camera fisheye lens coupled with BP neural network successfully predict the solar irradiance.