Abstract
With the shortage of fossil energy, and environmental pollution caused by its combustion,
clean solar energy resources are accounted by researchers. It can be predicted, solar electrical
energy generation will become of the important energy source in the world. However
photovoltaic arrays always work in hash environment, which result that part of single
photovoltaic modules begin aging and reduce the output power of system. Hot spot
phenomenon is the most common failure in photovoltaic arrays. it will not only reduce the
output power of system, it is easy to damage the photovoltaic modules, at the same time cause
security problems. Therefore, fast and efficient detection of hot spot fault of photovoltaic
modules is very necessary.
On the analysis and summary of the existing several hot spot detection method, in view
of temperature characteristics and infrared characteristic of solar panels in the fault conditions,
the paper illustrates the feasibility of hot spot detection of photovoltaic array using infrared
image. In order to solve the faults of low resolution, low signaltonoise ratio, and the defects
of low contrast of Photovoltaic infrared images, this paper use the nonuniform stretching,
adaptive median filtering and so on for correction to increase the infrared image contrast,
filter the salt and pepper noise, which laid a foundation for the next hot spot extraction.
On the analysis and comparison of the several common image segmentation algorithm,
paper apply it to the segmentation of photovoltaic array infrared image. However, the Otsu
method and maximum differential divergence threshold method failed to meet a satisfactory
segmentation effect. In order to solve difficult segmentation problem cause by the large
bimodal peak , this paper puts forward an adaptive maximum differential divergence
threshold method. At the same time, by a large number of photovoltaic infrared image
segmentation experiments, the paper summarizes the fast access to the largest divergence
threshold method and achieves a rapid, effective solar infrared image segmentation effect.
Experiments show that this method is a practical method of adaptive threshold selection.
The paper Uses Visual Studio software platform to make experiment of photovoltaic
array hot spot infrared image processing. By comparing the Otsu method and maximum
differential divergence threshold method, the paper verifies the accuracy and quickness of
adaptive maximum differential divergence threshold method. Quick and efficient extraction of
photovoltaic array hot spot provide the basis for subsequent fault diagnosis and orientation,
and has important significance to improve the intelligent level of photovoltaic panels
detection.
Key words：Photovoltaic array， hot spot， infrared image， Otsu， Maximum scatter
万方数据
