Renewable sources, especially the wind power, are volatile in nature which introduce non-linear uncertainties in wind power forecasting (WPF). These non-linear uncertainties hinder the way for the integration of wind power with the electric power grid. Generally, point forecast methods are used for WPF, which are less reliable as they do not offer any uncertainty-related information. In this study, a probabilistic forecasting methodology based on relevance vector machine (RVM) is used in a novel approach for WPF. Based on the forecast, wind power mismatch balancing cost (WPMBC) is formulated to offset the power balancing costs induced by wind power uncertainties (WPU). In addition to that, the probabilistic risk (PR) of failing to meet the contracted dispatch is also formulated. A realistic case study has been adopted for implementing the proposed RVM-based model. It has been found that the RVM model provides superior performance for WPF than other state-of-the-art machine learning models.

可再生能源，特别是风能，本质上是易变的，在风能预测（WPF）中引入了非线性不确定性。这些非线性不确定性阻碍了风能与电网的整合。通常，点预测方法用于WPF，可靠性较低，因为它们不提供任何与不确定性有关的信息。在这项研究中，基于相关向量机（RVM）的概率预测方法被用于WPF的一种新方法。根据预测，制定了风力发电不匹配平衡成本（WPMBC），以抵消由风力发电不确定性（WPU）引起的电力平衡成本。除此之外，还制定了不符合合同约定派遣的概率风险（PR）。一个现实的案例研究已被采用来实现建议的基于RVM的模型。已经发现，RVM模型为WPF提供了比其他最新的机器学习模型更好的性能。