Performance Analysis for Predictive Voltage Stability Monitoring Using Enhanced Adaptive Neuro-Fuzzy Expert System

Abstract

Intelligent voltage stability monitoring remains an essential feature of modern research into secure operations of power system networks. This research developed an adaptive neuro-fuzzy expert system (ANFIS)-based predictive model to validate the viability of two contemporary voltage stability indices (VSIs) for intelligent voltage stability monitoring, especially at intricate loading and operation points close to voltage collapse. The Novel Line Stability Index (NLSI) and Critical Boundary Index are VSIs deployed extensively for steady-state voltage stability analysis, and thus, they are selected for the predictive model implementation. Six essential power system operational parameters with data values calculated at varying real and reactive loading levels are input features for ANFIS model implementation. The model’s performance is evaluated using reliable statistical error performance analysis in percentages (MAPE and RRMSEp) and regression analysis based on Pearson’s correlation coefficient (R). The IEEE 14-bus and IEEE 118-bus test systems were used to evaluate the prediction model over various network sizes and complexities and at varying clustering radii. The percentage error analysis reveals that the ANFIS predictive model performed well with both VSIs, with CBI performing comparatively better based on the comparative values of MAPE, RRMSEp, and R at multiple simulation runs and clustering radii. Remarkably, CBI showed credible potential as a reliable voltage stability indicator that can be adopted for real-time monitoring, particularly at loading levels near the point of voltage instability.

Keywords

• intelligent predictive analytics
• voltage instability phenomenon
• voltage stability indices
• machine learning
• adaptive neuro-fuzzy expert system (ANFIS)
• subtractive clustering tuning