A new selection method of singular values using correlation coefficients for detecting weak exponential damped sinusoidal signals under strong noise

Underground magnetic resonance sounding (MRS) low signal-noise ratio (SNR) weak signal is hard to manage with traditional de-noising method such as fitting and filter.In view of this, this article makes use of both transient electromagnetic method (TEM) data and MRS data obtained from integrated detecting equipment.The goal is to extract weak MRS signal based on TEM under strong background noise.At first, underground electrical distribution can be acquired from TEM data, which is processed by equivalent conductive plane method.Then, the initial model of moisture content distribution was constructed on geological information and 3D MRS forwarding was made with finite element method.At last, the correlation coefficient series and deviations between forwarding data and measured data were calculated.The initial model would be corrected until the result is satisfying.Model calculation shows that this method can effectively extract the weak nuclear magnetic signals in tunnel under strong background noise.

In this paper, chaotic system is applied to identify and extract the weak signals of bearing early fault which are often submerged in strong background noise. Chaotic system is an effective method in weak signal detection because of its properties of noise immunity and sensitivity to the weak periodic signal. However, chaotic system is not completely immune to noise in critical chaotic state. Aiming at this problem, four indicators are used to evaluate the detection performance of Duffing oscillators. Then, the influence of Duffing oscillator parameters on the four indicators is studied in detail and a new method is proposed to improve the detection performance of Duffing oscillator. The simulation and experimental results show that the proposed method can accurately obtain the characteristic signals of early bearing fault in a lower signal-to-noise ratio (SNR) situation.

Adaptive thresholding and coordinate attention-based tree-inspired network for aero-engine bearing health monitoring under strong noise

The measurement of multiphase flow parameters is of great importance in a wide range of industries. In the measurement of multiphase, the signals from the sensors are extremely weak and often buried in strong background noise. It is thus desirable to develop effective signal processing techniques that can detect the weak signal from the sensor outputs. In this paper, two methods, i.e., lock-in-amplifier (LIA) and improved Duffing chaotic oscillator are compared to detect and process the weak signal. For sinusoidal signal buried in noise, the correlation detection with sinusoidal reference signal is simulated by using LIA. The improved Duffing chaotic oscillator method, which based on the Wigner transformation, can restore the signal waveform and detect the frequency. Two methods are combined to detect and extract the weak signal. Simulation results show the effectiveness and accuracy of the proposed improved method. The comparative analysis shows that the improved Duffing chaotic oscillator method can restrain noise strongly since it is sensitive to initial conditions.

The principle and implementation steps of the method of feature extraction of weak multi-frequency signal based on the array of modulated stochastic resonance under the background of strong noise are described in the paper. By the modulation of the known multi-frequency weak signals under the strong noise background and the carrieres with different frequency respectively; multiple signals with the same frequency of 0.01Hz were generated. Then these generated signals were as the input signals of multiple parallel non-coupled resonant units. The Runge-Kutta algorithm was used to obtain the unit outputs and to analysis the frequency spectrum. According to the SNR of the 0.01Hz to determine whether the 0.01Hz frequency components were contained in the frequency spectrum. Finally the frequency characteristic vectors of the weak signals were generated by the systemization of the detection results of the stochastic resonance units.Results show that this method has obvious effect in the extraction of the feature of the weak multi-frequencies signals, and has a very good application prospect.

Digital averaging technique is widely used in weak signal detection. The weak signal buried in strong noise is normally digitized with a multi-bit analog-to-digital converter (ADC) before the averaging. When using large number of averaging, the digitization can be accomplished by a 2-level comparator, instead of a multi-bit ADC. In this paper, we studied the statistical properties of ADCs with additive white Gaussian noise (AWGN) input. Our analysis shows that, with low signal-to-noise ratio (SNR) input, a comparator introduces 20% amplitude SNR loss compared with a fine multi-bit ADC, which can be mostly recovered with a post filter. It also shows that the comparator is linear and insensitive to input offset with low SNR input and AWGN.

The gain of the output signal-to-noise ratio (SNR) is insufficient in traditional stochastic resonance (SR) method to detect weak signals. In order to obtaining a higher SNR gain, a kind of signal detection method of adaptive stochastic resonance that use cascade system is proposed. To achieve the optimal output, stochastic resonance parameters adjustment is transformed into the multi-parameter optimization of particle swarm optimization. The optimal result in weak signal detection can be reached by adjusting the parameters of two layers subsystem. The weak signal submerged in strong noise background can be extracted through this simple method that has a fast convergence speed. The threshold of the weak signal detection is lowered by this method, and the applicable scope of stochastic resonance can be enlarged.

This paper presents a different idea that a watermarking scheme can be considered as a weak signal detection in strong noise, while the protected digital work is viewed as noise. Based on detection method of weak square signal in strong noise with a chaotic oscillator, a public-key robust watermarking scheme is proposed. This scheme can show all details except the private key. After the watermarking scheme is described, the model feasibility experiment and anti-attack ability experiments are made and discussed. Experimental and discussion results show that the scheme is robust enough against attacks of the commonly used image processing methods. Also, experimental results show that the algorithm is robust to malicious attacks.

A successive sources deletion method for enhancing the trackside acoustic weak signal of axle bearing

<abstract> <p>Rolling bear is a major critical component of rotating machinery, as its working condition affects the performance of the equipment. As a result, the condition monitoring and fault diagnosis of bearings get more and more attention. However, the strong background noise makes it difficult to extract the bearing fault features exactly. Furthermore, regular gradient disappearance and overfit appear in traditional network model training. Therefore, taking the printing press bearings as the research object, an intelligent fault diagnosis method based on strong background noise is proposed. This method integrates frequency slice wavelet transform (FSWT), deformable convolution and residual neural network together, and realizes the high-precision fault diagnosis of the printing press bearings. First, FSWT is used to preprocess the original vibration signal to obtain bearing fault features in the time and frequency domain, reconstruct the signal in any frequency band and describe local features accurately. Second, the ResNet is selected as the base network, and the two-dimensional time-frequency diagrams (TFD) obtained by preprocessing are used as input. For the model that has a poor ability to extract subtle features under strong background noise, the deformable convolution layer is introduced to reconstruct the convolution layer of ResNet, called deformable convolution residual neural network (DC-ResNet). Finally, the effectiveness of this method is verified by using the data sets collected under experimental conditions and actual working conditions for fault diagnosis of the printing press. The results show that the DC-ResNet can classify different bearing faults under strong background noise, and the accuracy and stability are greatly improved, which the accuracy meets 93.90%. The intelligent fault diagnosis with high-precision of printing press bearings under complex working conditions is realized by the proposed method.</p> </abstract>

Delayed feedback and detection of weak periodic signals in a stochastic Hodgkin–Huxley neuron

Aiming at the problem that the feature components that can reflect the gearbox system state are often interfered by noise and harmonics, when gearbox composite fault diagnosis is performed under strong background noise in large rotating machinery, which makes the extraction of fault feature signals limited. A composite fault diagnosis method based on singular value negentropy (SVN) and multipoint optimal minimum entropy deconvolution adjusted(MOMEDA) is proposed. First, the original composite fault signal is divided into different frequency bands using the 1/3 binomial tree strategy to reduce the noise impact and harmonic interference. Then the SVN of all frequency bands is calculated to construct the ordering kurtosis spectrum and the optimal frequency band is selected on the basis of considering both the periodicity and impulsivity of the fault signal. Then the multipoint kurtosis(MK) of the optimal frequency band is calculated and according to the extracted fault impulsive periodicity components using the MOMEDA algorithm deconvolute the optimal frequency band signal. Finally, the type of fault is judged by analyzing the frequency components with prominent amplitude in the mutual correlation spectrum. The analysis results of the experimental platform show that the method can effectively extract the fault characteristics disturbed by strong noise and realize the accurate diagnosis of Composite faults in gearbox.

In order to detect a weak periodic signal under the condition of intensive noise, the weak signal including strong background noise and adscititious multiplicative noise were used as input of a monostable stochastic resonance (SR) system. The adscititious noise intensity and the system parameter were adjusted adaptively with genetic algorithm by examining the SR effect on output signal-to-noise ratio (SNR). An improved numerical solution for a monostable SR model based on a fourth order Runge-Kutta algorithm was presented to enhance the SR effect. The simulation results show that the weak signal in an intensive noisy background could be successfully extracted. What is more, the output SNR was increased more than 30 dB comparing with the input SNR. It can be seen that the proposed method was superior to the traditional spectra analysis and envelope demodulation methods in detecting the weak periodic signal. Such detection approach indicates a promising prospect for mechanical fault monitoring and diagnosis.

X-ray pulsar-based navigation (XPNAV) is one of the perfect ways for autonomous deep-space navigation in the future. Due to spacecraft state model errors and strong cosmic background noises, low navigation accuracy is one of the main problems in XPNAV. This paper proposes a robust navigation filtering method to reduce the serious effect of spacecraft state model errors and strong noises on XPNAV. This method uses state model errors and pulsar observation errors to estimate and correct the state model. And then, to predict the spacecraft’ state in the next moment with high precision, the gain matrix is adjusted in quasi real-time by using the fading factor to ensure a minimized state estimation error variance in the next moment and an orthogonal residual sequence at different times. Finally, experimental results of multi-group simulations show that the proposed method had significantly improved navigation accuracy. And the accuracy of the proposed method is better than that of $\text{H}\infty $ robust filter and STUKF, especially when the state model errors and noise are great. Under the same conditions, compared with the other two methods, the proposed method has the minimum navigation filtering error and the strongest robustness.

This research work concentrates on developing new methodologies for weak signal detection, which are further used in the watermark detection application. The proposed detectors perform comparable to or better than most of the state-of-the-art techniques. Stochastic resonance plays a significant role in weak signal detection. Injection of precalculated noise is a big concern for us. Noises i.e., symmetric and asymmetric have been utilized for getting an improved version of the weak signal detector. With equality and non-equality constraints, we use the particle swarm optimization method to optimize the objective function. After that, we investigate a new detector based on the fractional operator. The noisy signal is convolved with the coefficients of fractional order filter. The proposed method has been tested with different standard performance parameters. The results of the proposed detector have been compared with state-of-the-art detectors such as a threshold detector (TD), neural network-based detector etc. The robustness of the proposed detector with respect to the parameters of signal and noise has also been explored. The problem of weak DC signal detection present in non-Gaussian noise is always a troublesome because of the non-linearity of the test statistic. We have used the concept of stochastic resonance in a conventional neural network-based detector. A predefined noise is injected during the back-propagation algorithm in order to minimize the error. The reduction in errors (in terms of enhancement in the probability of detection, $$P_D$$ at a constant value of the probability of false alarm, $$P_{FA}$$ ) and faster convergence give a right justification to use the concept of stochastic resonance in the neural network-based detector.