基于多层前馈神经网络的并联型电能质量控制器

神经网络用于电力系统电能质量分析和控制是一个新研究领域.快速可靠地提取谐波分量决定着并联型电能质量控制器的整体性能,构造了一种和理论分析相一致的基于反向传播算法的三层前馈神经网络,离线训练收敛后可用来在线检测电力系统谐波电流.系统中逆变器补偿电流的产生对系统的补偿性能至关重要,提出了一种基于神经网络的逆变器瞬时电流PWM控制.并联型电能质量控制器投入系统后电流总畸变率由26.29%下降为5.25%.仿真实例表明,所提并联型电能质量控制器动态响应快,可改善电力系统电流波形畸变,提高电能质量.

Parallel Power Quality Controller Based on Multilayer Feedforward Neural Network

Analysis and control of power quality by neural network is a new research field in electrical power systems. A fast and reliable extraction of harmonic components is determinant for improving the overall performance of the parallel power quality controller(PPQC). A three-layer feed forward neural network based on back-propagation algorithm, which used to detect harmonic of power systems on-line after converged network in off-line way, is presented in this paper. Voltage source inverter is very important for performance of PPQC. Another neural network is proposed for instantaneous current compare PWM control of inverter. The total harmonic distortion(THD)of current is reduced from 26.29% to 5.25% after PPQC put into operation. Simulation results demonstrate that the proposed PPQC based on neural network can yield satisfactory dynamic responses, improve current waveforms, and enhance power quality.