Probabilistic analysis and risk traceability of over-voltage for distribution networks based on global probabilistic voltage sensitivity

• A global voltage sensitivity analytical (GVSA) model based on node-shared impedance is constructed. • The probability model of voltage fluctuation is established by affine transformation of the GVSA model to forecasting error GMM. • The analytical expression of global probabilistic voltage sensitivity (GPVS) is derived. • The probability model and comprehensive contribution degree index for node voltage deviation are constructed. • The probability level of node over-voltage is analyzed, and the key risk nodes are traced based on the GPVS model. Highly penetrated distributed generations exacerbate over-voltage and increase the difficulty of voltage safety analysis in distribution networks. This study proposes a method for over-voltage probabilistic analysis and risk traceability based on global probabilistic voltage sensitivity (GPVS). First, a global voltage sensitivity analytical (GVSA) model is constructed based on the node-shared impedance. The Gaussian mixture models (GMMs) of forecasting errors are affine transformed based on the GVSA model to construct the probability models of voltage fluctuation components. Then, based on the characteristic function of GMM, the GPVS model for voltage fluctuation under the influence of source-load uncertainty is derived, which provides an analytical perspective for explaining the impact mechanism of source-load uncertainty on node voltage. Meanwhile, a comprehensive contribution degree index for node voltage deviation is constructed using the parameters of the GPVS model, which quantifies the influence weight of power uncertainty on node voltage deviation. Finally, the probability of node over-voltage is calculated using the GPVS model, and the key nodes that cause the risk of node voltage deviation are traced. The case study results show that the proposed method can characterize the non-Gaussian distribution characteristics of voltage fluctuations, quickly calculate the probability of node over-voltage, and trace the key risk nodes that cause node voltage deviation.