Quantum-behaved particle swarm optimization -ANN based identification method for typical power quality disturbance

This paper proposes an improved algorithm based on quantum-behaved particle swarm optimization (QPSO) to improve artificial neural network (ANN) training for the identification of typical power quality disturbance (PQD). Two sub networks which are used to identify the continual and break PQ disturbance respectively form the recognizer. Characteristic of PQ disturbance acting as the input of sub networks is obtained by projection pursuit regression, dynamic computing and fractal technique. QPSO with study factors, gather and speed factor added is applied to optimize the parameters´ computing method, so as to improve the neural network training. Six types of typical spot PQ disturbance are included in the experiment. The result shows that the proposed algorithm compared with that of which is based on the standard back propagation training with momentum factor added, is superior to the other algorithm with a better astringency and stability.