Photovoltaic Power Prediction Model Based on Meteorological Factors and TCN-KAN Model

In recent years, a large number of photovoltaic power generation equipment has been connected to the power system, and its intermittency and volatility have brought challenges to the prediction of power. Therefore this paper proposes a convolutional neural network based on Temporal Convolutional Network(TCN) and Kolmogorov-Arnold Network (KAN) of photovoltaic power prediction model. In order to take meteorological factors into account, this paper uses Pearson correlation coefficient to conduct correlation analysis between PV and different meteorological factors, and selects meteorological factors that have a great impact on PV power prediction, and then replaces them into the final TCN-KAN prediction model combined with historical PV data. By introducing error function for numerical comparison analysis, it is concluded that the forecast model based on meteorological factors and TCN-KAN proposed in this paper has higher prediction accuracy than TCN, LSTM and Transformer models.