基于矩特征傅里叶描述的示功图故障诊断研究

摘要

An indicator diagram fault diagnosis method which having torque characteristics and Fourier de-scription combined was proposed.Through optimal threshold segmentation and edge detection, the image of boundary with the largest filling area can be obtained;through calculating the image’ s torque characteristics, the torque characteristic sequence of the object shape can be reached;through discrete Fourier transform, the normalized torque characteristics’ Fourier description which boasting of features like translation,rotation and scale invariant was obtained and the fault types were classified by Euclidean geometrical distance method.Prac-tical application proves the proposed method’ s feasibility and effectiveness in diagnosing pumping unit’ s fault.%提出一种结合矩特征和傅里叶描述子的示功图故障诊断方法。示功图经过最佳阈值分割和边缘检测后，得到边界为最大区域填充后的图像，计算图像的矩特征获得表示物体形状的矩特征序列，再通过离散傅里叶变换得到具有平移、旋转和尺度不变性的归一化矩特征傅里叶描述子，采用欧氏距离法进行分类。实际应用表明：该诊断方法在油田抽油机故障诊断中具有可行性和有效性。