HOTSPOT REVELATION IN SOLAR PANEL USING SPARSE RECONSTRUCTION AND EXTREME LEARNING MACHINE

In today's world, solar panel is one of the major sources for generating power directly from the sunlight by using electronic processes and there is no greenhouse emission in photovoltaic cell as it does not require any other source of fuel like coal, natural gas, oil, nuclear power systems. Hotspot is one of the main causes of photovoltaic cell which occurs due to the dissipation of power in shaded cells. In the existing literature, the hotspot in solar panel is detected by using various algorithms and techniques but it does not improve accuracy, performance, temperature distribution, problem like overfitting and underfitting also exists. To overcome that, the proposed work deals with capturing the hotspot as thermal image through an infrared camera which is mainly used for temperature distribution. For identifying hotspot, the features like shade, correlation, contrast, energy, entropy, homogeneity, prominence, sparse are extracted using sparse reconstruction and GLCM algorithms. The features are given to the classification algorithm named as Extreme Learning Machine which gives the good generalization performance and improves accuracy higher when compared to other algorithms. The overfitting and underfitting problem can also be rectified by using these algorithms. Finally using extreme learning machine, the percentage of hotspot in photovoltaic cell can be identified.