VOLTAGE SECURITY MARGIN ENHANCEMENT USING GENERATION RESCHEDULING AND LOAD SHEDDING WITH AN ARTIFICIAL NEURAL NETWORK

The occurrence of the recent nationwide blackouts in some major power networks of the world indicates the weakness of current control and protection systems. Implementation of appropriate control and protection plans is necessary, then, to prevent such future occurrences. In this paper, an algorithm is presented which can evaluate and improve voltage stability by using an online artificial neural network (ANN). A voltage security margin (VSM) index is used to evaluate voltage stability by dividing the operating points into two groups: safe and unsafe. Generally, the algorithm consists of two stages: online and offline. Initially, though, in the offline stage, the VSM of an operating point which is labeled as unsafe is turned back to a secure level by changing the production pattern of generators and, if necessary, load shedding using the sensitivity analysis method. The ANN is then trained using different input variables with the required control actions in order to improve the voltage conditions of the system. In an online stage, the ANN can estimate the required controlling action which would return the system to a proper voltage within a safe time period. For real-time applications, network information can be obtained from Phasor Measurement Units (PMUs).