K-means algorithm based input reduction for voltage security state classification of power distribution system

Recent literature investigates that voltage insecurity may be dangerous for the power system as it causes sudden voltage collapse. For making real-time decisions regarding the security of the power system, it is very much essential to determine the current operating state of the system. This paper recommends a combined artificial neural network scheme to classify the operating conditions of a power system into normal, alarming, or insecure states. By choosing only the important input variables and eliminating unrelated ones, greater presentation is expected with lesser computational efforts. Here the K-means clustering method reduces the number of inputs for the proposed neural network structure to calculate the voltage security state with sufficient accuracy and speed. The success of the suggested technique is verified by two standard IEEE systems and one practical 85-bus system. Results show that the proposed K-means algorithm provides a compact artificial neural network model that can effectively and correctly recognize the working state of the power system with fewer numbers of inputs.