Search

Haze has a significant impact on photovoltaic (PV) power generation. When the fine particulate matter reaches a certain concentration, it becomes the main factor affecting solar irradiance and seriously reduces PV power generation, but few quantitative studies on the effect caused by haze to PV power generation. This study proposes the use of the improved method of the degree of grey slope incidence to analyze the weight factors of the effects of haze on irradiance. The exponential-linear model is used to describe the impact of haze on the amount of irradiance. Furthermore, the PV system model is used to focus on the quantitative loss of PV power under the influence of haze. By modeling and analyzing the data samples of PV power generation in Hangzhou, China, it can be concluded that the losses caused by haze on PV power generation in 2017 and 2018 were 5.25 ± 1.19% and 6 ± 1.16% of the original PV power generation, respectively. We extended this analysis to other cities to analyze the PV data in Tianjin, China. From December 2018 to December 2019, the loss of PV power generation caused by haze in Tianjin was 8.77 ± 0.9%. The quantitative analysis of haze on PV power can provide an effective basis for the economic evaluation of new PV systems and also plays an important role in the prediction and scheduling of PV power generation.

In this paper, an extreme learning machine model based on the improved similar day to predict the short-term output of photovoltaic (PV) is presented. The Pearson correlation coefficient is used to analyze the influence of various meteorological factors on the output of photovoltaic power generation, so as to find out the meteorological factors that have a greater impact on photovoltaic output. And then the prediction model is to be established with combining the fuzzy clustering method by improving the similar day selection method. The multi-day photovoltaic output data with the highest correlation about the day to be tested is used to train the extreme learning machine neural network which is then used to predict the PV output of the day to be measured. Finally, the experimental results show that the proposed method has higher prediction accuracy and shorter calculation time than the traditional prediction method, and what's more, the algorithm is simple and the prediction cost is low. It has a wide application value and research space.