新型自适应广义形态滤波在MOA在线监测数据处理中的应用

基于数学形态学的开－闭变换及其组合形式，综合考虑输出信号的均方根误差和信噪比，提出采用变尺度算法确定自适应的最优可变权值多级广义形态滤波器，该算法避免了自适应广义滤波器算法中最速下降法在迭代接近最优权值时收敛缓慢的缺点。将其用于MOA在线监测数据处理，以提高复杂电磁环境应力下MOA真实信号提取的精度，保证进一步分析诊断的准确性。仿真分析和实验结果表明，选取最优可变权值的广义滤波器可以抑制孤立的小点、毛刺、小桥和正负脉冲等噪声，运算简单，易于写入硬件(如嵌入式系统等)。

Application of Novel Self-adaptive Generalized Morphological Filter to MOA On-line Monitoring Data Processing

Based on the close and open transforms and their combinatorial modes, considering root-mean-square error and signal to noise ratio, this paper presents a self-adaptive multilevel generalized morphological filter, in which optimal variable weight is defined by variable scale algorithm. The algorithm overcomes the disadvantage that constriction speed is slow when iterative weight is close to the optimal value if steepest descent method is applied in the self-adaptive generalized morphological filter. This method is applied in the on-line monitoring data processing in order to improve the precision of acquired real signal of MOA in the stress of complicated electromagnetic environment. It ensures the accuracy of further analysis and diagnosis. Results of simulation study and on-site data processing show that the optimal variable weight generalized morphological filter can suppress noise such as isolated points, sentusi, small bridges, positive and negative pulses. Its calculation is simple and easily being implanted to hardware (such as embedded system).