基于ADALINE神经网络的电能质量监测终端设计

Design of Power Quality Monitoring Terminal Based on ADALINE Neural

谐波频谱检测是电能质量监测仪器的核心功能,其检测频谱是进行各种电能质量特征值运算的前提。传统以STFT为检测算法的电能质量终端由于时间窗固定,不具有暂态情况下的谐波分析能力。而诸如小波变换、S变换等算法则由于运算量巨大不利于谐波实时在线监测。针对这种情况,设计了一种以DSP为处理器,基于自适应线性神经网络(ADALINE)的电能质量监测终端。详细介绍信号接口电路、调理电路、PLL倍频电路、AD转换电路的硬件设计,给出了自适应线性神经算法推导和DSP数据处理框图。实验表明,所构建系统在运算量不大的情况下增强了暂态谐波测量能力,同时利用ADALINE误差信号可对电压暂降进行精确时间点定位。

The core function of Power quality terminal is to calculate the spectrum of signals in power system. The common technique for that is the short-time Fourier transform （STFT）. Its disadvantages for assessing transient signals are well known because of a fixed window and multiple new time-frequency algorithms such as wavelet transforms, S- transforms were introduced. Nevertheless, these new algorithms are too complicated to achieve for real time application. To face this issue, this paper proposes a monitoring instrument with adaptive linear（ADALINE） neuron algorithm. The deriva-tion of ADALINE is first introduced, then detailed component circuit design（including conditioning circuit, PLL circuit and ADC circuit＇and DSP’s data flow diagram are given. The experiment results show that the proposed instrument is able to calculate the spectrum of transient signals and have the ability to capture voltage dips. So this work offers a solution for continuous and online power quality monitoring terminal.