IEE Proceedings - Generation Transmission and Distribution · 1996 · 33 citations · 5 references
EngineeringComputer ScienceElectrical EngineeringArtificial IntelligenceEnergy SystemsMachine LearningControl SystemsControl TheoryNeural NetworksPower System DesignPower System StabilityPower SystemsElectrical Power SystemsNeural Network ControlPower Oscillation DampingPower OscillationsPower System ControlPower System OperationPower System StabilizationTraining
The paper presents an artificial neural network (ANN) power system stabiliser (NNPSS). The neural network in the proposed NNPSS is trained by an improved BP algorithm. The main difference between the proposed BP algorithm and the conventional BP algorithm is that two variable factors, a learning rate factor ɛ and a momentum factor α, are used. This significantly improves the convergence of the ANN's training. A four layer (7–7–4–1) ANN is used to design the NNPSS. The NNPSS is trained by samples obtained from power systems controlled by nonlinear power system stabilisers. The ability of the trained NNPSS to handle unknown disturbances using measurable variables has been investigated in two power systems, a single machine to infinite bus power system and a three machine power system. Test results show that the NNPSS is effective in damping out power system oscillations and is robust to the variations of both the system parameters and the system operating conditions.
5