抄録
In this paper, a hybrid intelligent system is proposed for estimating a load margin to the saddle node bifurcation point of voltage stability. It is based on the integration of Regression Tree (RT) and Artificial Neural Network (ANN). Voltage stability analysis is one of the main concerns in power system operating and planning. The objective of voltage stability analysis is to evaluate the saddle node bifurcation point on PV or QV curves. Thus, it is necessary to estimate a load margin to the saddle node bifurcation point of voltage stability efficiently. This paper proposes a new method for estimating the load margin with the hybrid method of RT and ANN. RT is used to classify data into terminal nodes and extract rules from each terminal node. ANN is constructed to estimate the load margin to the bifurcation points at each terminal node. Also, a new method for generating power system conditions is presented to consider the correlation of the nodal specified values. The effectiveness of proposed method is demonstrated in the IEEE 30-bus system in terms of computational accuracy and computational time.
