Method to identify acoustic sources for anti-submarine warfare

A method of detecting an underwater source of an acoustic signal acquired by a listening device, the method comprising:

• a. decomposing the acoustic signal into one or more resolution levels;
• b. integrating the acoustic signal at each resolution level resulting in a denoised signal;
• c. enhancing the denoised signal by autocorrelation;
• d. processing the denoised signal with a convolutional neural network (CNN), the CNN having one or more convolutional levels, one or more activation functions levels, and one or more pooling levels;
• e. training said CNN with a training data set;
• f. assigning weights and biases to each convolution level;
• g. measuring an error of classification of the each convolution level by comparing the classification by the CNN to the training data set;
• h. adjusting the weights and biases;
• i. repeating processing by the CNN with the training data set until an error rate stabilizes;
• and
• j. detecting the underwater source by the CNN.

A method, comprising:

• decomposing a waveform signal in a signal domain into a plurality of resolution levels;
• selecting, with a trained model, a resolution level from the plurality of resolution levels to be reconstructed back to the signal domain;
• reconstructing, using a scaling coefficient, the waveform signal in a selected resolution level back to the signal domain through each previous resolution level;
• and
• discerning a source of the waveform signal from a background noise.

A method of training a model, the method comprising:

• decomposing a waveform signal in a signal domain into a plurality of resolution levels;
• selecting, with a trained model, a resolution level from the plurality of resolution levels to be reconstructed back to the signal domain;
• reconstructing, a denoised waveform signal in selected resolution level back to the signal domain through each previous resolution level to discern a source of the denoised waveform signal from a background noise;
• processing the denoised waveform signal with a convolutional neural network (CNN), the CNN having one or more convolutional levels, one or more activation functions levels, and one or more pooling levels;
• training the CNN with a training data set;
• assigning weights and biases to each artificial neuron of the CNN;
• adjusting the weights and biases by measuring an error of classification;
• and
• repeating processing by the CNN with the training data set until an error rate stabilizes.

A method, comprising:

• removing, with a wavelet analysis, a random noise in an acoustic signal, resulting in a denoised signal;
• enhancing a signal source of the acoustic signal by autocorrelating the denoised signal;
• isolating the signal source, comprising:
• extracting features with the wavelet analysis,
• matching a waveform by overlapping a mask onto a testing waveform and calculating a difference between the mask and the testing waveform, and
• processing the denoised signal with a convolutional neural network (CNN);
• and
• identifying the signal source of the acoustic signal from a background noise.

A method, comprising:

• decomposing a waveform signal in a signal domain into a plurality of resolution levels;
• selecting, with a trained model, a resolution level from the plurality of resolution levels to be reconstructed back to the signal domain;
• reconstructing the waveform signal in a selected resolution level back to the signal domain through each previous resolution level;
• discerning a source of the waveform signal from a background noise;
• and
• acquiring the waveform signal as an underwater waveform signal from underwater listening devices and identifying an underwater source of the underwater waveform signal with an autocorrelation.

A method, comprising:

• decomposing a waveform signal in a signal domain into a plurality of resolution levels, the decomposing including at least:
• dividing the waveform signal into a first resolution level of two groups, a first scaling coefficient group and a first detail coefficient group,
• deconstructing successive resolution levels by further decomposing the first scaling coefficient group into at least a second scaling coefficient subgroup and a second detail coefficient group, and
• further decomposing previous resolution level at each successive resolution level;
• selecting, with a trained model, a resolution level from the plurality of resolution levels to be reconstructed back to the signal domain;
• reconstructing the waveform signal in a selected resolution level back to the signal domain through each previous resolution level;
• and
• discerning a source of the waveform signal from a background noise.

A method, comprising:

• decomposing a waveform signal in a signal domain into a plurality of resolution levels;
• selecting, with a trained model, a resolution level from the plurality of resolution levels to be reconstructed back to the signal domain;
• reconstructing the waveform signal in a selected resolution level back to the signal domain through each previous resolution level;
• and
• discerning a source of the waveform signal from a background noise, the discerning including at least overlapping a mask onto a testing waveform and calculating a difference between the mask and the testing waveform.