Abstract

JOSE, Oscullo,  and  JAIME, Cepeda,. Power System Stabilizer Adaptive Tuning Based on Decision Trees and a Heuristic Optimization Process. Rev Politéc. (Quito) [online]. 2023, vol.51, n.1, pp.57-66. ISSN 2477-8990.  https://doi.org/10.33333/rp.vol51n1.05.

This paper presents a novel approach for adaptively damping low-frequency electromechanical oscillations via the application of decision trees that uses as inputs frequency and power signals of generation buses monitored by the wide-area measurement system (WAMS). This methodology can be applied for adaptive tuning of conventional Power System Stabilizers (e.g., PSS1A, PSS2A, PSS2B), generally available in actual power systems. It is done by adjusting the PSS tuning parameters by analyzing critical oscillation modes for different operational scenarios using the Mean-Variance Mapping Optimization (MVMO) heuristic algorithm. An intelligent classifier selects the PSS’s optimal parameters for each scenario based on Decision Trees with the objective of adapting the PSS tuning to the operating conditions. This process uses PowerFactory of DIgSILENT and critical modes are determined using modal analysis for simulation. In real-time, binding modes are derived by modal identification algorithm established with matrix Pencil of WAMS-signals. Such a methodology is applied to the 66-Bus New York-New England test power system, showing an excellent dynamic response in adapting PSS tuning to the different power system conditions. In addition, the response is compared with the PSS4B, which is characterized by monitoring various frequencies of the oscillation modes present in the system operation, thus presenting the benefits of proposed contribution.

Keywords : Oscilatory Stability; Decision Trees; Heuristic Optimization; Machine learning.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )