Research On The Baseline Drift Removal Algorithm For Hydroacoustic Signals Based On Cumulative Sum-Downsampling

In underwater detection,especially vector underwater detection,the collected underwater signals often suffer from baseline drift caused by the signals themselves,the receiving devices and acquisition circuits,as well as environmental factors such as waves and platform motion.Baseline drift refers to the fact that the mean of the target signal and high-frequency noise components in the underwater signal,which should be zero,actually fluctuates over time.This fluctuation itself constitutes a very low frequency signal component,which is difficult to remove without distortion using traditional time or frequency domain methods such as filtering,because the frequency of the underwater target signal is also low and close to the frequency of the baseline drift.Although the baseline drift can be effectively removed by variational mode decomposition,the process is cumbersome and computationally intensive.This paper proposes a low-complexity algorithm to address the non-stationary and difficult-to-solve baseline drift problem in underwater signals,which achieves the same level of performance as variational mode decomposition and has important practical significance.First,this paper analyzes the causes and characteristics of baseline drift and explains its non-stationary nature,starting from actual underwater signals that contain baseline drift.The paper introduces the intrinsic mode function model and empirical mode decomposition and variational mode decomposition,which are used to process non-stationary signals.Second,the paper analyzes the method of down-sampling baseline fitting and its problems,and proposes a new algorithm called "sum-down-sampling baseline fitting",which constructs the algorithm by adding and differentiating before and after the down-sampling baseline fitting algorithm.The new algorithm only requires four steps: summing,down-sampling,interpolation fitting,and differentiating,to fit the baseline drift in the signal,with excellent fitting performance.The paper also provides a detailed analysis of the frequency domain interpretation of this algorithm and preliminarily determines a parameter selection method.Finally,to verify the effectiveness of the algorithm,the paper compares it with the widely used variational mode decomposition algorithm.Firstly,the paper verifies the simplicity of the algorithm’s implementation in terms of theoretical time complexity.Secondly,the paper uses simulation to demonstrate that the proposed algorithm has smaller errors than VMD.Finally,the paper verifies the algorithm’s superior ability to separate baseline drift through underwater experiments.