A Fusion Filtering‐Based Method for Extracting Spectral Trajectories of Non‐Repetitive Transient Weak Signals in Multiple Frequency Bands for Industrial Control Networks

ABSTRACT In response to the problem of difficulty in accurately extracting spectral trajectories of multi‐band non‐repetitive transient weak signals in industrial control networks under strong noise background, this paper proposes a method that introduces a fusion filtering algorithm to extract spectral trajectories of multi‐band non‐repetitive transient weak signals in industrial control networks. Build an XGBoost classifier for multi‐band signal states in industrial control networks, clarify the current non‐repetitive transient weak signal states in multi‐band industrial control networks, and achieve accurate identification of weak signal states. Based on the classifier results, combined with the Bloom filter, the classified signal sequence is deduplicated to eliminate redundancy caused by repeated transient signals. Further adopt a fusion filtering algorithm to filter the spectral trajectory of weak signals in multiple frequency bands of the input industrial control network, in order to restore the original signal; extract the single‐frequency spectral trajectory of multi‐band non‐repetitive transient weak signals through average sampling processing, and fuse the trajectory to obtain the complete weak signal spectral trajectory. The experimental data comparison results show that the proposed method can better extract weak signal spectral trajectory information under transient conditions, and the extraction effect is more stable and reliable, with high accuracy and anti‐interference ability.