The invention relates to an electric equipment thermal fault diagnosis method and system and an electronic device. The method comprises the following steps: collecting an infrared image of electric equipment and constructing a convolutional neural network model on the basis of the infrared image; inputting a to-be-detected infrared image into the convolutional neural network model; identifying a temperature scale and electric equipment in the infrared image through the convolutional neural network model; generating an RGB value and temperature reference table on the basis of RGB values of pixels points of the identified temperature scale and upper and lower bounds of the temperature scale, extracting an RGB value of the identified electric equipment, comparing the extracted RGB value withRGB values in the RGB value and temperature reference table, and obtaining a temperature result of the identified electric equipment; carrying out diagnosis on the temperature result through a power grid system diagnosis standard, and determining whether thermal faults of the electric equipment occur. The electric equipment is identified in a highly efficient and accurate manner through the convolutional neural network model, the temperature is read accurately through the RGB value, and a power grid system is improved in intelligent level.