A new joint denoising algorithm of MMS vector hydrophone

Introduction

In ocean exploration, through the analysis and processing of the signal received by the hydrophone, the target category of the sound source and its related angle, position and other state parameters can be obtained. However, different noises and interferences will inevitably be mixed into the hydrophone during data collection. Therefore, in order to further detect, identify and locate the signal, the influence of these noise interference must be eliminated as much as possible. There are many main algorithms for underwater acoustic transducer signal denoising: traditional Fourier filtering method, wavelet transform method and empirical mode decomposition method. These algorithms will all have a certain denoising effect on noisy signals, but they also have some shortcomings. Traditional wavelet denoising has problems in how to select wavelet bases and the number of decomposition levels. A signal processing algorithm, empirical mode decomposition, is proposed. This algorithm does not need to set the basis function, but it will produce modal aliasing, which will cause the two adjacent intrinsic modal function waveforms to be aliased when reconstructing. There is still a lot of noise mixed in. U et al. proposed a collective empirical mode algorithm for this, adding auxiliary white noise to reduce the influence of modal aliasing, but it cannot guarantee that the white noise introduced in the decomposition process can completely eliminate P1. Variational modal decomposition is a new modal decomposition algorithm. The algorithm realizes the effective separation of the inherent modal functions by determining the frequency center and bandwidth of each inherent modal function [has a solid theoretical foundation and can better solve the problem of modal aliasing. According to the theory of the VMD algorithm, before using the VMD decomposition on the original denoising signal, the number of modal components fc and the penalty term factor% of the VMD decomposition need to be set in advance. The value of fc and the value of "value are directly related to the final decomposition result. If the value of c is too small, the signal decomposition will be insufficient. If the value is too large, false signal components will be generated, which will cause interference to the analysis of the useful components of the original signal. If a is too large, the bandwidth of the modal will be smaller, on the contrary, if a is smaller, the bandwidth of the modal will be larger. Therefore, the determination of the value and plays a vital role in the VMD algorithm, but most of the parameters of the VMD algorithm are set based on human experience as a comparison. [In response to the above problems, a new method based on SC A-P SO is proposed. The algorithm optimizes the V MD parameters fc and a, takes the mean square error of the reconstructed signal as the fitness function of the algorithm, and finds the optimal sum to achieve the purpose of noise reduction.

 

2 Basic principles. The principle of VM D is a non-recursive adaptive algorithm for signal processing. Bandwidth Constrained Variational Problem Corresponding to V MD Algorithm

 

 

Sine cosine algorithm

Sine and Cosine Algorithm) is a new type of swarm intelligence optimization algorithm of underwater hydrophone transducer . When using the SCA algorithm to search for optimization, it can be divided into two processes. The first is an exploration process. The optimization algorithm quickly explores the feasible region in the search space by combining a random solution among all random solutions, and the second is a parallel process. , The random solution gradually changes, and its change speed is lower than the speed of the exploration process, so its specific update.

 

Particle swarm algorithm

 

Particle swarm algorithm (is a swarm intelligence optimization algorithm. In the PSO algorithm, the direction and distance of the particle movement are determined by the speed of the particle, and the dynamic adjustment of the particle speed is carried out based on the movement experience of itself and other particles. In this way, the optimization of the particle in the solvable space is further realized so that in each iteration process, the speed and position of the particle are updated by updating the individual extremum and the global extremum. The specific update process. In the formula, it is the rth iteration. When the velocity of the i-th particle in the d-dimension is the individual optimal value of the i-th particle in the d-dimension in the ith iteration; it is the entire round of the i-th particle in the d-dimension in the ith iteration. The optimal value; is the A; the second iteration?: the position of the particle in the d dimension; w is the inertia weight; C1 and C2 are the acceleration factors, which are non-negative constants; random numbers between 0.

 

Principle of Wavelet Soft Threshold Denoising

The principle of wavelet soft threshold denoising: First, the noisy signal is orthogonally decomposed, and the wavelet coefficients are obtained after decomposition? Then set a threshold A and compare. If the magnitude of ten and A, the coefficient is mainly produced by noise; if the coefficient is mainly produced by the signal. Finally, the inverse wavelet transform is performed on the wavelet coefficients to obtain the signal warfare after denoising. The estimation formula of the soft threshold.

 

The SCA-PSO-VMD-WT algorithm proposed in this paper is based on the analysis and theoretical basis. This paper proposes the SCA-PSO-VMD-WT algorithm for noise reduction. The noisy signal is decomposed by V MD to obtain the modal component, and whether the modal component is a noise component is determined, and the noisy modal component is selected for wavelet threshold denoising, and then the signal is reconstructed by hydrogen separation to obtain the denoised signal. The root mean square error (RMSE.) of the reconstructed signal is taken as the fitness function of SC A-P SO to find the optimal fc and a to achieve the purpose of noise reduction. The proposed SCA-P SO-VM D-WT The algorithm noise reduction is mainly divided into W steps: set the game method parameter f, the maximum number of iterations is set to 30, the population number is set to 20.2 initialization position and speed. In this paper, the VMD parameters ft and a are used as the position vector of the algorithm ．Update the position and velocity well to calculate the fitness function value. Use the formula to update the position, use the formula to update the speed, and output the optimal and global optimal fitness function values.

 

 

Simulation Experiment

The software used in the experiment in this paper is Matlab R20 14 The simulation signal is sj jM 0 sentences. In order to make the simulation experiment more realistic, random noise is added to the simulation signal. However, rtr r—In ocean exploration, the noise intensity of underwater acoustic signals is variable due to the influence of oceanic oceanography and human activities. In order to simulate this situation, this article will add-l, Gaussian white noise, the evaluation indicators of the denoising effect in this paper are root mean square error (RMSE) and signal-to-noise ratio (SNHJ. for comparison, at the same time.

 

Algorithm

 

The algorithm and the denoising result of the algorithm. Figure 1 shows the original signal and the noisy signal under different decibels. Figure. The denoising effects of different denoising algorithms. Table 1 shows the comparison of denoising evaluation indexes.

 

Comparing Figure 1 with Figure 2-Figure 6, it is found that the four algorithms can effectively remove the Gaussian and white noise in the noisy signal under different decibels, but the denoising effect of the VMD-WT denoising algorithm is poor, and the VMD- The WT denoising grate method is to perform wavelet threshold denoising after decomposing the noisy signal by VMD, which shows that the choice of VMD parameters for A: and a has a very clear effect on f signal denoising fi; compared with the VMD-WT denoising algorithm, The denoising effects of the PSO-VMD-WT and SCA-VMD-WT algorithms have been improved to a certain extent, but from Table 1, it can be seen that the SCA-PSO-VMD-WT denoising algorithm is used in SNR and RMS. E has better results. .

 

 

 Measurement

The actual measurement experiment of the MEMS vector hydrophone was carried out by researchers from the Key Laboratory of North University of China in Fenhe Second Reservoir. The hydrophone ST was fixed on the bank, the transducer was placed on the tug, and the distance between the tug and the array was Gradually increase, choose different positions to stay anchored, use the transducer to transmit the signal, and then carry out data collection. This experiment intercepts the signals of 8000 HZ and 1 0000 HZ at 1,000 points to obtain denoising. The previous measured signal. Figure 7 and Figure 8 are the measured signals and their frequency spectrums of 800 Hz and 1000 Hz, respectively, and the denoising signal and their frequency spectrums.

Observing Fig. 7 finds that: the input signal of 80 Hz has less high-frequency noise, and the waveform is smooth after denoising. The denoising effect of this algorithm is good. Observing Fig. 8 finds that the input signal of HZ has more spectral burrs, which indicates that Noise T. The noise is large, the basic characteristics of the sound source signal are retained after denoising, and the denoising effect of this algorithm is good. .

 

 

in conclusion

Aiming at the problem of random noise in the signal of the underwater acoustic sensor, this paper proposes the SC A-P SO-V MD-WT denoising method. In the simulation experiment, by comparing the evaluation indicators of the VMD-WT, PSO-VMD-WT and SCA-VMD-WT algorithms under different decibels, it is found that the SCA-PSO-VMD-WT proposed in this paper Excellent noise algorithm: PVMD-WT, PSO-VMD-WT and SCA-VMD-WT algorithms. Therefore, the SCA-PSO-VMD-WT denoising algorithm proposed in this paper can be used to denoise the measured signal data. The results show that: The denoising effect of the SCA-PSO-VMD-WT algorithm is clear, indicating that the method proposed in this paper has a denoising effect. Have a certain reference.