Applied research of multilevel generalized morphological filter for ultrasonic signal of PD in processing

The multilevel generalized morphological filter (GMF) with the fixed weights, considering root-mean-square error and signal to noise ratio, is presented based on the close and open transforms and their combination modes. The methodology overcomes the steepest descent method on the adaptive generalized morphological filter when iterative weight is close to the best one because of slow constrictions. The algorithm is applied in the partial discharge (PD) processing in order to improve the precision of obtaining the real ultrasonic signal of PD in the stress of complicated electromagnetic environment. It ensures the accuracy of the further analysis and diagnosis. The results show that the best variable weight generalized morphological filter can multi-class noise such as positive and negative impulse noise effectively. And it is calculated and came true simply by hardware (embedded system).