Short-Term PV Plant Power Forecasting Based on Relevance Vector Machine-Markov Chain

In order to predict the output power of PV plant accurately, an approach of the PV plant's short-term power forecasting is proposed, based on the principle of relevance vector machine - Markov chain. The original data is trained and tested by the relevance vector machine principle, and the original prediction model is obtained. The prediction error is corrected by the combining least squares method with Markov chain, and the prediction -corrected model is obtained. Based on the time scale and precision scales, the short-term power forecasting model of PV plants is established. The actual short-term power of PV plant is predicted by this model, the average relative error and average absolute error is acted as the evaluation index. The numerical example shows that the proposed prediction method is effective and practical, the proposed prediction - error correction model can be used to predict power output of PV plant. Compared with other forecasting methods, both the average relative error and the maximum relative error are small, thus the prediction accuracy of PV plant's short-term power is improved.