Voltage Security Events Classification of Power System Using Machine Learning Techniques

Voltage insecurity is a most vital problem that the modern power systems are imposed off. Timely and accurate assessment of voltage security is necessary to detect post-contingency voltage problems in order to prevent a large scale blackout. The paper presents machine learning techniques based on CART algorithm and probabilistic decision tree based strategy for estimation of voltage security. To reduce redundancy and to improve efficiency certain critical contingencies are considered. The data for inputs are fuzzified and outputs target class are utilized for decision tree learning method. Probabilistic classification trees are most suitable for estimating stability problems because the objective is secure or insecure subsequent a credible contingency. The proposed method is tested using IEEE 30 bus test system. The advantage of fuzzy decision tree based exposure of voltage security is the robust classification of future samples. Also the performances of obtained results are shown through accuracy tests. The fuzzy decision tree method is also compared with CART algorithm. Key-words: CART, classification, fuzzy decision tree (FDT), power system security, voltage collapse.