Genetic Algorithm and Decision Tree-Based Oscillatory Stability Assessment

Abstract

-This paper deals with a new method for eigenvalue region prediction of critical stability modes of power systems based on decision trees. The critical stability modes result from inter-area oscillations in large-scale interconnected power systems. The existing methods for eigenvalue computation are time-consuming and require the entire system model that includes an extensive number of states. However, using decision trees, the oscillatory stability can be predicted based on a few selected inputs. Decision trees are fast, easy to grow and provide high accuracy for eigenvalue region prediction. Special emphasis is hereby focused on the selection process for the decision tree inputs. In this work, a genetic algorithm is implemented to search for the best set of inputs providing the highest performance in stability assessment.

Transient Stability Assessment using Decision Trees and Fuzzy Logic Techniques

“…In [13], a new method for eigenvalue region prediction of critical stability modes of power systems based on decision trees was presented. The paper proved that DT's are fast, easy to grow, and provide high accuracy for eigenvalue region prediction.…”

Mode selective damping of power system electromechanical oscillations for large power systems using supplementary remote signals

“…Therefore, the system experiences inter-area oscillations. The system has been developed based on characteristic parameters of the European interconnected electric power system, also known as UCTE/CENTREL [23]. All generators are equipped with identical IEEE standard exciters (IEEE type DC1A excitation system).…”

A fast stability assessment scheme based on classification and regression tree

“…With respect to the input attributes of a decision tree, it is reported that different attribute combinations may result in different data mining accuracies [12]. In order to accelerate the prediction process, it is desirable to use the least number of attributes as DT inputs while keeping an acceptable level of overall prediction accuracy.…”

“…Later, in [11], the system post-disturbance stability has been analyzed by DT using its fast evaluation capability. In [12], a genetic algorithm was applied in feature selection to search for the best inputs to DT for oscillatory stability region prediction. In [13], Kamwa et al showed that there is a trade-off between a data mining model's accuracy and its transparency.…”