单神经元变步长控制算法在谐波检测中的应用

摘要

A harmonic detection method of adaptive variable step control algorithm based on single neuron is proposed,which has the advantages of a single neuron,such as self-learning, serf-regulating, self-organizing capacity.The use of variable step size adaptive algorithm real-time online control steps. Meanwhile,with the company of TI fixed-point digital signal processor TMS320F2812 as a platform,real-time detection of power harmonic frequency,using CAN bus to transmit real-time PC analysis show that to achieve real-time monitoring.Through the experimental device testing,any number of harmonic components can be accurately analyzed, which providing an important basis for the harmonic compensation.%提出了基于单神经元的变步长自适应谐波检测方法,充分利用单神经元具有的自学习、自调节、自组织能力的优点,运用变步长自适应控制算法,通过实时在线调节步长,达到系统本身自调节以应对电网参数的突变.同时,以TMS320F2812型DSP为核心搭建了检测分析平台,实时检测分析电力谐波次数;运用CAN总线将分析结果实时传送到上位机显示,达到实时监控目的.通过实验装置检验,可准确地分析出任意次数的谐波分量,为谐波补偿提供重要依据.