In the context of new power system construction, the randomness and volatility of photovoltaic (PV) output pose challenges to the safety and stability of the power system, efficient and accurate PV output prediction has become the key to maintain the stability of the power grid. The existing PV output prediction scheme mainly uses meteorological data and historical output series to predict the future PV output value. Due to the lack of data of some weather types, K-means, EM clustering and other methods are often used to cluster the weather types, ignoring the differences of weather characteristics between the same large categories and the differences of individual weather, resulting in errors. In this paper, the method of fuzzy clustering is used to select similar days. Through fuzzy clustering on the micro meteorological data of the whole year, similar days with similar characteristics of the day to be measured are selected as the training set, and then the key influencing factors of PV output are found as input through correlation analysis. Combining with the principal component analysis method, the micro meteorological factors are dimensioned down, and the correlation between variables is removed. The variables are brought into the wavelet-Support Vector Machine(SVM) prediction model. Experiments show that the model has better prediction ability and applicability.
Research on Photovoltaic Output Forecasting Model Based on the Selection of Similar Days by Fuzzy Clustering
2022-10-12
1262021 byte
Conference paper
Electronic Resource
English
Clustering Days with Similar Airport Weather Conditions
NTRS | 2014
|Clustering Days with Similar Airport Weather Conditions
AIAA | 2014
|Clustering Days with Similar Airport Weather Conditions
NTRS | 2014
|CLUSTERING DAYS WITH SIMILAR AIRPORT WEATHER CONDITIONS (AIAA 2014-2712)
British Library Conference Proceedings | 2014
|