基于改进级联神经网络自适应电网谐波检测

发布时间：2018-01-25 18:39

本文关键词： 电网谐波检测级联神经网络改进自适应单元步长约束条件　出处：《电力系统保护与控制》2016年20期 　论文类型：期刊论文

【摘要】：为克服电网谐波检测快速性与稳定性矛盾,基于神经网络自适应原理提出了一种级联神经网络自适应电网谐波检测的改进系统。改进级联系统初级运用大步长常规LMS(Least Mean Square)自适应神经网络单元提高检测跟随性能,次级通过嵌入均值滤波环节平滑权值波动的策略构造新的自适应神经网络单元,保证次级神经网络单元具有良好的电网谐波检测稳态精度。运用传递函数Z域变换分析嵌入均值滤波环节的电网谐波检测自适应神经网络单元的稳定性能,运算推导新的级联次级神经网络自适应单元的步长约束条件,保证改进系统既能够有效地提高电网谐波检测的跟随性能同时又可以提高检测的稳态精度。仿真实验表明改进的级联神经网络自适应系统能有效提高电网谐波检测动态性与精确性。
[Abstract]:In order to overcome the contradiction between the speed and stability of harmonic detection in power system. Based on the principle of neural network adaptation, an improved cascade neural network adaptive harmonic detection system is proposed. Least Mean Square-based adaptive neural network unit improves detection performance. A new adaptive neural network unit is constructed by embedding the mean filter to smooth the weight fluctuation. The stable performance of the adaptive neural network unit embedded in the mean filter is analyzed by using the transfer function Z domain transformation to ensure that the secondary neural network unit has a good steady-state accuracy of harmonic detection. The step size constraints of a new cascaded secondary neural network adaptive unit are derived by computing. The simulation results show that the improved cascade neural network adaptive system can effectively improve the harmonic detection performance of the power network and improve the steady-state accuracy of the detection. The simulation results show that the improved cascade neural network adaptive system can effectively improve the harmonic detection of the power network. Dynamic and accurate measurement.
【作者单位】：广东水利电力职业技术学院自动化工程系;武汉大学电气工程学院;
【分类号】：TM935;TP183
【正文快照】： 随着电能质量要求的进一步提高,谐波补偿已成为一个重要的研究课题。由于实际中有源滤波器的广泛运用,检测电网的谐波或基波具有重要意义。在数字滤波器中运用离散傅立叶变换确定谐波时,常规需要检测一个周期的数据,致使检测的计算量非常大;并且检测精度越高,滤波器的阶数要求

【相似文献】