【作者】 梅永；

【导师】 王柏林；

【作者基本信息】 河海大学 ， 控制理论与控制工程， 2005， 硕士

【摘要】 在电力参数的测量和电网谐波分析中，交流采样代表了发展方向。理想的交流采样是同步采样，但是由于信号周期的飘移和定时器最小定时单位的限制，同步采样会出现偏差(采样同步偏差)。定时器最小定时引起的同步偏差已有众多学者研究，信号周期飘移下的同步采样则较少有人涉足。 本文针对存在周期飘移的周期信号，研究了实时最佳逼近同步采样方法和与之配套的校正误差新算法——矩阵级数法。 实时最佳同步采样法的原理可归纳为三点：(1) 用前一个周期测量(或估计)值实时计算当前采样周期；(2) 每次只采样一个信号周期的N个点；(3) 每个采样点与预计同步点的误差不大于一个定时器最小定时单位。 实时最佳同步采样法的同步偏差很小，但是采样同步偏差还是存在的，采样同步偏差直接导致谐波分析的误差。在存在采样同步偏差的情况下用特定算法减少谐波分析误差的过程称为频谱校正，本文提出了一种全新的频谱校正方法——矩阵级数法。矩阵级数法可归纳为三步：(1) 将N个实际采样时刻和采样值代入DFT公式求出谐波系数的初始向量(?)；(2) 由N个采样时刻直接计算出增益矩阵G；(3) 将向量(?)和矩阵G代入频谱校正公式求出精确的谐波系数向量θ。其中，关键的频谱校正公式是根据DFT的矩阵形式和系统辨识中的‘矩阵反演定理’首次导出的，它是一个矩阵级数形式，只有矩阵的加法和乘法运算，级数取的项数越多精度越高。矩阵级数法的主要优点是：实时性好、算法简单、精度很高。数字仿真验证了这种算法的有效性。 本文还用PSoC(可编程片上系统)制作了一台电力参数监测仪(试验装置)，在这台试验装置上原理性地实现了实时最佳同步采样法和矩阵级数法。更多还原

【Abstract】 AC sampling represents developing direction in the measurement of electric power parameters and grid harmonics analysis. Synch-sampling is an ideal AC sampling method, it may appear deviation (sampling synchronous deviation ) because of the limitation of minimum timing unit of timer and variety of signal frequence. Synchronous deviation caused by limitation of the timer has already been studied by numerous scholars, however, Synch-sampling in the instance of signal cycle instability has been less discussed.This paper proposed a novel algorithm - Matrix Series Method based on real-time sampling and related correcting error when cycle signal is instable.The theory of real-time and best Synch-sampling can be summed up for three points: (1) Sampling time is calculated real-timely by using the last cycle value. (2) N points can be sampled in one signal cycle each time; (3) The error between real sampling point and ideal Synch-sampling point is no more than minimum timing unit of timer.Although synchronous deviation of real-timely synchronous sample is very small, synchronous deviation still exists. Synchronous deviation can lead to spectrum analysis error directly. Course of reducing error with specific algorithm harmonics analysis is called frequency spectrum correcting in the situation when the synchronous deviation exists. This paper proposed a novel frequency spectrum method- Matrix Series Method. Matrix Series Method can be summarized for three steps:(1) Initiative vector of harmonics series are formed by using formula DFT basedon the obtained N sampling points and N sampling values.(2) Gain Matrix G is calculated by using N sampling points.(3) Accurate harmonics coefficient vector is calculated by frequency spectrumcorrecting formula based on the obtained matrix G and vectorThe crucial frequency spectrum correcting formula is deduced from Matrix Inversion Theorem and Matrix series of DFT, it has only addition and multiplication operation of matrix series, the more iterative times are,the higher precision is. The main advantages of Matrix Series Method are simpler algorithm, higher precision and more real-time. The effectiveness of Matrix Series Method is shown by the simulation results.An electric parameter monitoring instrument (experimental equipment) based on PSoC (Programmable System on Chip)is also studied, Synch-sampling method and Matrix Series Method are realized theorily in this instrument.更多还原