Assessment of voltage stability risks under stochastic net loads using scalable SVM classification

A novel approach is presented to identify and assess voltage stability risk in power networks that include nodes with variable loads and intermittent renewable generation. Given a power network configuration, the approach firstly assesses voltage instability boundaries at selected nodes in response to stochastic operational conditions (variable load and intermittent decentralised generation). By iteratively calculating Saddle Node Bifurcation (SNB) points one node at a time given a set of loads and intermittent generation conditions at the other nodes, those data points are used as training sets for support vector machine (SVM) classifiers. Secondly, a marginal voltage stability risk probability distribution for the intermittent buses is derived using Monte Carlo simulation methods with stochastic net load profiles within the system where instability status at each simulation is derived from the trained set of SVM classifiers. The key advantage of the proposed method is its scalability to higher dimension networks for which the SVM training date set must be calculated only once. The voltage risk probability distribution acquires a significant importance in the design of quantitative risk valuation framework for planning and expansion purposes in a context of network with decentralised and intermittent generation. This paper focuses on describing the approach to derive a voltage risk probability using SVM technique, while the construction and use of the probability distribution in a comprehensive risk valuation framework for planning is left for future work to the authors.