Ship radiation noise identification method, device and equipment and readable storage medium

Abstract

The invention discloses a ship radiation noise identification method, apparatus and device, and a readable storage medium. The method comprises the steps of obtaining ship radiation noise data; inputting the ship radiation noise data into a pre-constructed ship radiation noise identification model, wherein the ship radiation noise identification model outputs a ship radiation noise category; wherein the construction of the ship radiation noise identification model is based on a network model of an improved ResNet, the network model of the improved ResNet is based on a ResNet-18 residual network, and an attention SE module is introduced. According to the invention, the identification accuracy of the ship radiation noise is improved.

Description

Technical field

[0001] The invention involves the field of water sound signal recognition technology, which specifically involves a ship radiation noise recognition method, device, equipment and readable storage medium. Background technique

[0002] Under the current international situation, the war environment is becoming more and more complex. Whether it can obtain marine security information in a timely and accurate manner, identify water sound signals, and provide accurate battlefield conditions for command systems at all levels. Essence Based on traditional machine learning methods to build a water sound signal recognition system, the main steps include feature extraction, feature selection, and classifier design. However, with the continuous increase of data volume, traditional methods are gradually difficult to meet existing identification tasks.

[0003] In recent years, with the rise of deep learning technology, deep neural networks, as the most classic deep learning alg...

Examples

Embodiment

[0065] Combine figure 2 Show, a method of ship radiation noise recognition method, specifically includes the following steps:

[0066] Step 1. The sampling of ship radiation noise datasets. Since ship radiation noise is generally mainly concentrated in the low frequency part, the reduction processing processing the calculation of subsequent audio segmentation and identifying networks while retaining ship radiation noise feature information.

[0067] Step 2. The audio segmentation of the downtime database is uniform, which makes the segmented audio fragments uniform, which is conducive to extracting consistent ship radiation noise characteristics, and distinguish the category of the ship according to the radiation noise.

[0068] Step 3. Build a network model based on improving Resnet. This model is based on the resonet-18 residual network to introduce attention SE modules to improve the useless feature information while improving the characteristics of feature expression.

[0069]...

Technical field

[0001] The invention involves the field of water sound signal recognition technology, which specifically involves a ship radiation noise recognition method, device, equipment and readable storage medium. Background technique

[0002] Under the current international situation, the war environment is becoming more and more complex. Whether it can obtain marine security information in a timely and accurate manner, identify water sound signals, and provide accurate battlefield conditions for command systems at all levels. Essence Based on traditional machine learning methods to build a water sound signal recognition system, the main steps include feature extraction, feature selection, and classifier design. However, with the continuous increase of data volume, traditional methods are gradually difficult to meet existing identification tasks.

[0003] In recent years, with the rise of deep learning technology, deep neural networks, as the most classic deep learning alg...

Examples

Embodiment

[0065] Combine figure 2 Show, a method of ship radiation noise recognition method, specifically includes the following steps:

[0066] Step 1. The sampling of ship radiation noise datasets. Since ship radiation noise is generally mainly concentrated in the low frequency part, the reduction processing processing the calculation of subsequent audio segmentation and identifying networks while retaining ship radiation noise feature information.

[0067] Step 2. The audio segmentation of the downtime database is uniform, which makes the segmented audio fragments uniform, which is conducive to extracting consistent ship radiation noise characteristics, and distinguish the category of the ship according to the radiation noise.

[0068] Step 3. Build a network model based on improving Resnet. This model is based on the resonet-18 residual network to introduce attention SE modules to improve the useless feature information while improving the characteristics of feature expression.

[0069]...

Images