Abstract:In view of the slow convergence of the classical adaptive filtering algorithm in processing mechanical fault signals, an improved adaptive filtering algorithm was proposed in the framework of big data. Based on Hadoop platform, a mechanical fault big data processing framework with threelevel structure was constructed, which was used to collect and preprocess the original fault big data set. In the aspect of signal filtering, step change factor function and mean square error function were introduced to improve the convergence performance of the algorithm. Based on discrete particle swarm optimization algorithm, the filtering process of fault signal was optimized to improve the iteration speed and global optimization ability. The experimental results show that the improved filtering algorithm has obvious noise reduction effect, especially in the condition of low SNR, the convergence performance is better than the classical filtering algorithm