In order to solve the uncertainty of hotspot when airborne thermal imaging detects the surface of solar photovoltaic panels. The mechanism of hotspot is analyzed, and the equivalent circuit model of photovoltaic panels under shadow shading is constructed to obtain the partial shadow conditions. The mathematical expressions of the surface heat and output current of the photovoltaic panel are verified by MATLAB/ Simulink simulation and temperature measurement. Build a UAV hotspot simulation detection platform, change the input current and detection height to obtain a thermal imaging map of the photovoltaic panel surface, establish a segmentation function that characterizes the pixel statistical value at different temperatures, and use the K-means clustering algorithm to build a feature value database, which will be used to locate the subsequent faulty photovoltaic panel. The experimental results show that the method has the controllability of the current on the surface temperature of the photovoltaic panel, the infrared thermal imaging feature library constructed by the piecewise function-K-means clustering hybrid algorithm can accurately represent the hotspot features.