Infrared image-based photovoltaic array fault grading method

The invention discloses an infrared image-based photovoltaic array fault grading method. According to the infrared image-based photovoltaic array fault grading method, infrared image information of a photovoltaic array system is collected; the infrared image information comprises a plurality of infrared images of a photovoltaic module; each image is divided into a plurality of image units; R, G and B values of each unit form an image information database tensor; fault coding is carried out on the infrared images according to obtained temperature information of the photovoltaic module; a fault grade vector is predicted through the image information database tensor by a Tensor-V hybrid partial least square method, so that a tensor-vector fault grading model is built, and has high predictive ability; rapid and effective fault grade coding on the infrared images of the photovoltaic module can be achieved; and the rapid and effective fault grade coding is supplied to photovoltaic power station operation and maintenance workers to arrange maintenance and replacement for the faulty photovoltaic module, so that an operation of the photovoltaic module in a constant optimum state is achieved; and economic benefits of a photovoltaic power plant are ensured.