Short-term photovoltaic output prediction model based on MOSMA feature selection and TCN

The prediction of photovoltaic power generation is an important basis for photovoltaic grid connection and consumption. In order to ensure the better participation of photovoltaic power generation in power grid dispatching., this paper presents a Short-term photovoltaic output prediction model based on MOSMA feature selection and TCN. Firstly, the correlation analysis is carried out, and the strong correlation meteorological factors are selected. The time series is decomposed by ICCEMDAN to reduce its volatility. and obtain a stable time series subsequence. Secondly, a multi-objective function of the minimum feature subset and the optimal prediction accuracy is established, and MOSMA is used for feature selection to obtain the feature set with the maximum fitness.. Finally, the most characteristic set is input into the TCN for photovoltaic power generation time series prediction, and the prediction results of other prediction models are evaluated and compared. Experiments show that t the TCN model can be optimized by adding modal decomposition and multi-objective feature selection,, and has higher prediction accuracy and feasibility than other prediction models.