Noise suppression method for magnetic resonance groundwater detection based on convolutional neural network
A convolutional neural network and noise suppression technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve problems such as limited noise reduction effect, poor universality, and complex operation, and achieve a complete CNN model structure, Improved work efficiency and high training accuracy
Active Publication Date: 2022-07-05
JILIN UNIV
View PDF6 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0006] The above-mentioned inventions have significant noise filtering effects on specific types of noise elimination, but there are problems of complex operation, low efficiency, and poor universality, which is not conducive to use by non-professional technicians, and the noise cancellation effect is limited, and it is impossible to extract Signals drowned in strong noise
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0072] The magnetic resonance signal denoising method based on convolutional neural network includes the following steps:
[0073] The specific steps of constructing the training data set in the step a are:
[0074] First, according to the expression Construct the simulated magnetic resonance signal, where t=(0:N-1) / fs, fs is the sampling rate, E 0 ∈(10, 4000) nV, f L ∈(1300, 3700) Hz, get X 1 (n)∈R1×N, repeat m times to obtain the NMR signal data set X(n)∈R with the number of signals M and the length of the collected data N M×N , where M=1000, N=2000;
[0075] Secondly, use the randn function and trigonometric function in MATLAB to construct different types of noise, including random noise, power frequency noise and spike noise, and get the simulated environmental noise N(n)∈R after superposition M×N ;
[0076] Third, change the amplitude, frequency, relaxation time, and phase of the signal respectively to construct multiple groups of X(n) and N(n) to obtain a nois...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention relates to a noise suppression method for magnetic resonance groundwater detection based on a convolutional neural network. First, the noisy signal and the noise signal are used as the input and output of the convolutional neural network, the transfer function gradient is calculated through the forward propagation process and the back propagation process, and the network parameters are optimized by training the network through the gradient descent method, and then the denoising model is determined. , in which the convolution layer of the network automatically extracts the features of the noisy signal and the noise signal, and the two construct a residual network to obtain a clean magnetic resonance signal. The invention not only suppresses various types of noise in the nuclear magnetic resonance signal and effectively improves the signal-to-noise ratio, but also uses big data and a high-performance computing platform to extract the effective information of the magnetic resonance signal under strong noise interference through deep learning. Compared with the traditional method , once the training is over, the prediction time is only a few seconds, which improves the work efficiency, and there is no need to manually adjust the parameters and no prior information during the training and prediction process.
Description
technical field [0001] The invention belongs to the field of nuclear magnetic resonance sounding (Magnetic Resonance Sounding, MRS) signal noise suppression methods, in particular to a magnetic resonance groundwater detection noise suppression method based on convolutional neural networks (Convolutional Neural Networks, CNN). Background technique [0002] As a geophysical method that can directly and quantitatively detect groundwater, nuclear magnetic resonance groundwater detection technology has received extensive attention in recent years and has achieved rapid development. However, the magnetic resonance signal is extremely weak, on the order of nanovolts. When the detection instrument is disturbed by environmental noise such as random noise, power frequency noise, and spike noise, the accuracy of the extracted characteristic parameters of the magnetic resonance signal is reduced, which affects the subsequent inversion interpretation results. reliability. Therefore, eff...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G01V5/00G06N3/04G06N3/08
CPCG01V5/0008G06N3/084G06N3/045G06F2218/04G06F2218/08Y02A90/30
Inventor 林婷婷韦萌张扬滕飞
Owner JILIN UNIV