Navigation emergency alarm device and method for sound source direction finding

Abstract

The present application provides a navigation emergency alarm device and method for sound source direction finding, which relates to the field of navigation alarm technology. The method includes: a hazard source direction finding accuracy configuration module for configuring the hazard source direction finding accuracy; a direction change degree calculation module for calculating the direction change degree; a hazard source distance analysis module for obtaining the hazard source distance; and a navigation emergency alarm discrimination module for obtaining the navigation emergency alarm result. This application can solve the technical problem that during navigation, due to environmental noise interference and dynamic changes in the direction of the hazard source, the hazard source cannot be accurately located and its direction and distance cannot be determined in real time, thereby affecting navigation safety. The present application can achieve the technical effect of quickly identifying the hazard source direction change and distance through dynamic direction finding and distance analysis of the audio information array, and making emergency alarm discrimination based on the hazard level classification accuracy, thereby improving the hazard source detection accuracy and alarm response speed, and ensuring navigation safety.

Description

Technical Field

[0001] The present application relates to the technical field of navigation alarms, and in particular to a navigation emergency alarm device and method for sound source direction finding. Background Art

[0002] In existing technologies, ships face complex and ever-changing ocean environments and potential hazards during navigation, such as other vessels, ocean buoys, reefs, and inclement weather. These hazards are often accompanied by various acoustic signals, such as mechanical noise, wave crashes, and emergency alarms. However, due to interference from ambient noise and the dynamic changes in the location of the hazard source, existing navigation alarm systems struggle to accurately identify and locate the direction and distance of the hazard source. This technical limitation prevents ships from promptly and accurately determining the specific location and changing trends of the hazard source in hazardous situations, thus compromising navigation safety. Existing direction-finding technologies primarily rely on single-point sound source localization or simple multi-point direction-finding equipment. However, in practice, the accuracy and reliability of direction-finding are severely impacted by the attenuation, reflection, and interference that can occur in complex environments. Furthermore, existing devices often lack the ability to analyze hazard source direction changes and assess distance in real time, making it impossible to dynamically adjust direction-finding accuracy or hazard classification accuracy based on actual conditions. This deficiency can lead to delayed alarm response or a high false alarm rate, a problem that is particularly pronounced when ships are traveling at high speeds or in drastic changes in the ocean environment. Summary of the Invention

[0003] The purpose of this application is to provide a navigation emergency alarm device and method with sound source direction finding, so as to solve the technical problem that during navigation, due to environmental noise interference and dynamic changes in the direction of the danger source, the danger source cannot be accurately located and its direction and distance cannot be determined in real time, thereby affecting navigation safety.

[0004] In view of the above problems, the present application provides a navigation emergency alarm device and method for sound source direction finding.

[0005] In the first aspect, the present application provides a navigation emergency alarm device for sound source direction finding, the device comprising: a danger source direction finding accuracy configuration module, which is used to collect an audio information array during navigation through a microphone array, and configure the danger source direction finding accuracy according to the audio change degree between each audio information array and the historical audio information array at the previous moment; a direction change degree calculation module, which is used to perform danger source direction finding identification on the audio information array according to the danger source direction finding accuracy, obtain the direction of the danger source sound source, and calculate the direction change degree in combination with the historical danger source sound source direction at the previous moment; a danger source distance analysis module, which is used to obtain the navigation speed of the ship, perform danger source distance analysis in combination with the audio change degree, and obtain the danger source distance; a navigation emergency alarm discrimination module, which is used to configure the danger level classification accuracy according to the direction change degree, perform danger level classification according to the direction change degree and the danger source distance, obtain the danger level, perform navigation emergency alarm discrimination, and obtain the navigation emergency alarm result.

[0006] In the second aspect, the present application provides a navigation emergency alarm method for sound source direction finding, which is implemented by a navigation emergency alarm device for sound source direction finding described in the first aspect, wherein the method includes: collecting an audio information array during navigation through a microphone array, and configuring the danger source direction finding accuracy according to the audio change degree between each audio information array and the historical audio information array at the previous moment; performing danger source direction finding identification on the audio information array according to the danger source direction finding accuracy to obtain the direction of the danger source sound source, and calculating the direction change degree in combination with the historical danger source sound source direction at the previous moment; obtaining the navigation speed of the ship, and performing danger source distance analysis in combination with the audio change degree to obtain the danger source distance; configuring the danger level classification accuracy according to the direction change degree, performing danger level classification according to the direction change degree and the danger source distance to obtain the danger level, performing navigation emergency alarm judgment, and obtaining the navigation emergency alarm result.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The device provided in the embodiment of the present application is based on a hazard source direction-finding accuracy configuration module. It collects audio information arrays during navigation through a microphone array, and configures the hazard source direction-finding accuracy according to the audio change degree between each audio information array and the historical audio information array at the previous moment; based on a direction change degree calculation module, it performs hazard source direction-finding identification on the audio information array according to the hazard source direction-finding accuracy, obtains the direction of the hazard source sound source, and calculates the direction change degree in combination with the historical hazard source sound source direction at the previous moment; based on a hazard source distance analysis module, it obtains the navigation speed of the ship, and performs hazard source distance analysis in combination with the audio change degree to obtain the hazard source distance; based on a navigation emergency alarm discrimination module, it According to the direction change degree, the danger level classification accuracy is configured, and according to the direction change degree and the distance to the danger source, the danger level is classified to obtain the danger level, and the navigation emergency alarm is judged to obtain the navigation emergency alarm result. This effectively solves the technical problem that the danger source cannot be accurately located and its direction and distance cannot be judged in real time due to environmental noise interference and dynamic changes in the direction of the danger source during navigation, thereby affecting navigation safety. The dynamic direction finding and distance analysis of the audio information array are used to quickly identify the direction change and distance of the danger source, and emergency alarm judgment is performed based on the danger level classification accuracy, thereby improving the danger source detection accuracy and alarm response speed, and ensuring the technical effect of navigation safety.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, which can be implemented in accordance with the contents of the description, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically listed below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in this application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and a person of ordinary skill in the art can obtain other drawings based on the provided drawings without creative work.

[0011] Figure 1 This is a structural diagram of a navigation emergency alarm device with sound source direction finding according to the present application.

[0012] Figure 2 This is a flow chart of a navigation emergency alarm method based on sound source direction finding in this application.

[0013] Explanation of the accompanying symbols: hazard source direction finding accuracy configuration module 1, direction change calculation module 2, hazard source distance analysis module 3, navigation emergency alarm identification module 4. DETAILED DESCRIPTION

[0014] The present application provides a navigation emergency alarm device and method with sound source direction finding, so as to solve the technical problem that the danger source cannot be accurately located and its direction and distance cannot be judged in real time due to environmental noise interference and dynamic changes in the direction of the danger source during navigation, thereby affecting navigation safety. The present application achieves the technical effect of quickly identifying the direction changes and distances of the danger source through dynamic direction finding and distance analysis of the audio information array, and making emergency alarm judgments based on the accuracy of the danger level classification, thereby improving the danger source detection accuracy and alarm response speed and ensuring navigation safety.

[0015] Below, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited to the example embodiments described herein. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. It should also be noted that, for the convenience of description, only the parts related to this application, rather than all of them, are shown in the accompanying drawings.

[0016] For example 1, please refer to the attached Figure 1 The present application provides a navigation emergency alarm device for sound source direction finding, which specifically includes the following steps:

[0017] The hazard source direction-finding accuracy configuration module is used to collect audio information arrays during navigation through a microphone array, and configure the hazard source direction-finding accuracy based on the audio change degree between each audio information array and the historical audio information array at the previous moment.

[0018] Specifically, in the hazard source direction-finding accuracy configuration module, microphone arrays are installed at multiple locations on the ship. These microphone arrays are used to collect audio information from different directions to form an audio information array. In order to improve the detection accuracy of the hazard source, the audio information array collected at the current moment and the historical audio information array at the previous moment will be combined to calculate the audio change degree. The greater the audio change degree, the more likely the ship is approaching the hazard source rapidly. Therefore, a higher hazard source direction-finding accuracy needs to be configured. This hazard source direction-finding accuracy will be used to determine the number of hazard source direction-finding paths required for subsequent hazard source direction-finding identification, thereby ensuring that the hazard source direction positioning is more accurate and reliable.

[0019] Furthermore, the microphone array is used to collect audio information arrays during navigation. Based on the audio change degree of each audio information array compared with the historical audio information array at the previous moment, the hazard source direction finding accuracy is configured, including:

[0020] Through the microphone array, an audio information array is collected during the navigation process to obtain the audio information array at the current moment and the historical audio information array at the previous moment; the audio feature information array and the historical audio feature information array are extracted; based on the audio feature information array and the historical audio feature information array, the audio change degree is calculated; based on the audio change degree, the danger source direction finding accuracy is configured.

[0021] In a preferred embodiment, the audio signal at the current moment is collected by a microphone array to form an audio information array containing multi-dimensional audio data. For example, the audio signal collected by each microphone will form an audio information data point corresponding to its position, and will be combined into the audio information array at the current moment. At the same time, the historical audio information array collected at the previous moment is loaded as the basis for analysis and comparison; then, a feature extraction operation is performed on the audio information array at the current moment and the historical audio information array, including extracting characteristic parameters of the audio signal, such as frequency, amplitude, phase difference, signal energy, etc. Specifically, a filter (such as a bandpass filter) is used to filter out background noise in the audio information array and the historical audio information array to retain the main audio signal frequency range, and then the amplitude of the denoised audio signal is normalized by maximum and minimum value normalization to form a standard audio information array and a standard historical audio information array; then, the spectral characteristics of each standard audio information in the standard audio information array and the spectral characteristics of each standard historical audio information in the standard historical audio information array, such as the main frequency, frequency band energy distribution, etc., are obtained by Fourier transform. Then, the time domain parameters of each standard audio information and each standard historical audio information are calculated, such as the signal amplitude mean, variance, peak value and signal energy. The phase difference between each standard audio information and each standard historical audio information at different microphone positions is also analyzed to extract phase information for judging the directionality of the sound source. After the extraction is completed, these feature information are organized into a current audio feature information array and a historical audio feature information array, which correspond to the audio feature data of two time points respectively. Then, by comparing the current audio feature information array with the historical audio feature information array, the audio change degree is calculated. The audio change degree is an indicator that describes the amplitude of the change in audio features between two moments. It is obtained by calculating the absolute difference between the corresponding feature values ​​in the current audio feature information array and the historical audio feature information array, and then accumulating all the calculated results. Then, based on the calculated audio change degree, combined with the average audio change degree during navigation, the hazard source direction finding accuracy is dynamically adjusted. The larger the audio change degree, the more significant the change in the hazard source position or the hazard source is rapidly approaching the ship. At this time, a higher direction finding accuracy is required to ensure the accuracy of the positioning result. Through the above process, real-time and high-precision direction finding of dangerous sources can be achieved, effectively improving navigation safety and alarm response capabilities.

[0022] Furthermore, configuring the danger source direction finding accuracy according to the audio variation degree includes:

[0023] The average audio variation during navigation is obtained; the ratio of the audio variation to the average audio variation is multiplied by the number of hazard source direction finding paths and rounded to the nearest integer to obtain the current number of hazard source direction finding paths as the hazard source direction finding accuracy.

[0024] In a feasible implementation, after obtaining the audio variation degree, this audio variation degree is accumulated with all previous audio variation degrees and divided by the number of audio variation degrees to obtain the average audio variation degree during the entire navigation process; then, the ratio of the audio variation degree to the average audio variation degree is calculated to indicate the intensity of the current audio change amplitude relative to the average change degree in the historical navigation, and then the calculated ratio is multiplied by the total number of hazard source direction-finding paths (a component of the hazard source direction-finding model used for hazard source direction-finding identification), and the product is processed by rounding down or rounding off to obtain the current number of hazard source direction-finding paths. This current number of hazard source direction-finding paths is used as the current hazard source direction-finding accuracy, that is, the number of direction-finding paths used for actual analysis of the hazard source direction. It should be noted that when the audio variation degree is greater than the average audio variation degree, that is, the calculated ratio is greater than 1, the calculated ratio is reset to 1. This is because it is necessary to avoid the current number of hazard source direction-finding paths obtained exceeding the total number of hazard source direction-finding paths. Through the above process, the accuracy of hazard source direction finding can be dynamically adjusted to ensure high-precision analysis when the hazard source approaches rapidly or changes significantly, while optimizing resource usage when the environment is relatively stable.

[0025] The direction change degree calculation module is used to perform hazard source direction finding and identification on the audio information array according to the hazard source direction finding accuracy, obtain the direction of the hazard source sound source, and calculate the direction change degree in combination with the historical hazard source sound source direction at the previous moment.

[0026] Specifically, in the direction change calculation module, the direction of the hazard source is identified for the audio information array collected at the current moment based on the obtained hazard source direction finding accuracy. In this process, the corresponding number of hazard source direction finding paths are extracted from the pre-built hazard source direction finding model using the currently configured hazard source direction finding accuracy, and the audio information array is analyzed based on these hazard source direction finding paths to identify the hazard source sound direction at the current moment. Subsequently, the currently identified hazard source sound source direction is compared with the historical hazard source sound source direction identified at the previous moment to calculate the direction change. The direction change is an indicator that measures the degree of change between the current hazard source direction and its historical direction. It is usually calculated by the angle between the two directions, that is, the angle of the current hazard source direction is subtracted from the angle of the hazard source direction at the previous moment. The magnitude of the direction change can be used to infer the relationship between the ship's direction of travel and the direction of the hazard source. If the direction change is small, it means that the current hazard source direction has not changed much from the historical direction, and it is inferred that the probability that the hazard source direction is consistent with the ship's direction of travel is high, indicating that the ship is heading towards the hazard source. If the direction change is large, it indicates that the hazard source direction has changed significantly, indicating that the hazard source position has a large directional offset relative to the ship. This analysis method can combine temporal continuity information to more accurately judge the dynamic changes of hazard sources and provide a more reliable basis for direction judgment for the navigation safety of ships.

[0027] Furthermore, according to the hazard source direction finding accuracy, the audio information array is subjected to hazard source direction finding identification to obtain the hazard source sound direction, and the direction change degree is calculated based on the historical hazard source sound direction at the previous moment, including:

[0028] In the pre-trained hazard source direction-finding model, a hazard source direction-finding path is randomly selected according to the number of current hazard source direction-finding paths within the hazard source direction-finding accuracy, wherein the hazard source direction-finding model includes hazard source direction-finding paths of the number of hazard source direction-finding paths; the audio information array is input into the randomly selected hazard source direction-finding path, and a set of hazard source sound source directions is obtained as output; the mean of the set of hazard source sound source directions is calculated to obtain the hazard source sound source direction; the historical hazard source sound source direction at the previous moment is obtained, and the degree of change in direction between the hazard source sound source direction and the historical hazard source sound source direction is calculated.

[0029] In a feasible implementation, the pre-trained hazard source direction-finding model includes multiple hazard source direction-finding paths, each path is generated through supervised learning training, and is used to analyze the relationship between audio features and hazard source directions. According to the number of current hazard source direction-finding paths within the calculated hazard source direction-finding accuracy, a corresponding number of hazard source direction-finding paths are randomly extracted from the hazard source direction-finding model for direction analysis of the current audio information array; then, the audio information array collected at the current moment is input into multiple randomly selected hazard source direction-finding paths, and each path processes the input audio information array according to the rules obtained from its training, outputs the hazard source sound source direction under the path, and then combines the output direction results of all paths into a hazard source sound source direction set; After that, the average of all direction results in the dangerous source sound source direction set is calculated, and the calculated average is used as the final dangerous source sound source direction at the current moment; after calculating the final dangerous source sound source direction, the historical dangerous source sound source direction is obtained from the historical data of the previous moment, and the angle of the dangerous source sound source direction and the angle of the historical dangerous source sound source direction are used to calculate the absolute difference to obtain the direction change between the current direction and the historical direction. The smaller the direction change, the smaller the change in direction is between the current and the previous directions, indicating that the dangerous source position is relatively stable or the ship is heading towards the dangerous source. The larger the direction change, the larger the change in direction is, indicating that the dangerous source position has changed significantly, indicating that the dangerous source is moving rapidly or the ship's heading has changed. Through the above steps, the dangerous source direction finding accuracy, audio information array and pre-trained model can be dynamically combined to calculate the dangerous source sound source direction and its change in real time, providing ships with high-precision direction warning information and ensuring navigation safety.

[0030] Furthermore, the pre-training step of the hazard source direction finding model includes:

[0031] Based on the hazard source orientation test data records, a sample audio information array set is collected, and the hazard source directions corresponding to different sample audio information arrays are marked to obtain a sample hazard source sound source direction set; the sample audio information array set and the sample hazard source sound source direction set are divided to obtain multiple supervisory data on the number of hazard source direction finding paths; using the multiple supervisory data on the number of hazard source direction finding paths, supervised training is used respectively to train a number of hazard source direction finding paths, and the hazard source direction finding paths are combined to obtain a hazard source direction finding model.

[0032] In a feasible implementation, a series of sample data is collected from the hazard source orientation test data records obtained from the test in a controllable test environment to form a sample audio information array set. The controllable test environment includes multiple fixed hazard source positions and dynamically moving hazard sources to simulate various scenarios in actual applications. Subsequently, for each sample audio information array collected, the corresponding hazard source direction is marked according to the known position and direction of the hazard source. The hazard source direction can be represented by an angle (for example, 0° to 360°). All the marked direction information is then organized into a sample hazard source audio source direction set. Afterwards, according to the preset number of hazard source direction-finding paths (determined according to business needs and combined with expert decisions, which may be used in hazard source direction-finding identification), The maximum number of paths used) is used to divide the sample audio information array set and the corresponding sample hazardous source sound source direction set to form multiple supervision data. The number of supervision data is the same as the number of hazardous source direction-finding paths. Each supervision data includes training data, verification data and test data, wherein the training data, verification data and test data all include sample audio information arrays and corresponding sample hazardous source sound source directions; after obtaining multiple supervision data, an independent hazardous source direction-finding path is constructed for each supervision data. The construction method can be support vector machine, neural network, decision tree, etc. Taking the convolutional neural network (CNN) in the neural network as an example, CNN is used to construct a hazardous source direction-finding path, including input layer, convolution layer, pooling layer, fully connected layer and The output layer and other structures convert the sample audio information array used for training into a time-frequency graph form through short-time Fourier transform (STFT) as the input feature matrix of the direction-finding path. The input layer receives the processed time-frequency graph matrix and passes it to the convolution layer. The convolution layer uses multiple groups of filters to extract time-frequency features and enhances the nonlinear characteristics through the ReLU activation function. The pooling layer further reduces the dimension of the convolution features to reduce the computational complexity and enhance the noise resistance of the direction-finding path. After multi-layer convolution and pooling operations, the output features are expanded into a one-dimensional vector and passed to the fully connected layer for feature fusion and high-dimensional feature mapping. The output layer is set to regression mode according to the task requirements to predict the direction of the dangerous source. During the training process, the direction-finding path of the dangerous source is randomly initialized. The weight of the path is determined by the training data of the supervised data used this time, which is input into the hazard source direction-finding path for forward propagation. After being passed through the input layer, convolution layer, pooling layer, fully connected layer and output layer layer by layer, the predicted hazard source direction of the current audio information will be output. The mean square error (MSE) is used as the loss function, and the gradient of the loss with respect to the weight of each layer is calculated layer by layer through the back-propagation algorithm. The Adam optimizer is used to update the weight of the hazard source direction-finding path, and the path parameters are optimized to minimize the loss value. During the training process, the training data is input into the hazard source direction-finding path in batches. After each iteration, the verification data is used to evaluate the performance of the hazard source direction-finding path. The hyperparameters of the path (such as learning rate, batch size, etc.) are adjusted according to the results of the verification data to improve the generalization ability of the path.Training is repeated until the preset maximum number of iterations is reached or the performance of the validation data no longer improves. After training, the hazard source direction-finding path is evaluated using test data to verify the path's performance in the hazard source direction prediction task. If the hazard source direction-finding path performs well on the test data, it is saved as the final hazard source direction-finding path. If the performance does not meet the expected target, the path structure or training parameters are adjusted and retrained to optimize the effect. The same process is used for each supervised data to obtain multiple hazard source direction-finding paths that meet the required number of hazard source direction-finding paths. Finally, all trained hazard source direction-finding paths are combined in parallel to form a complete hazard source direction-finding model. This hazard source direction-finding model dynamically selects an appropriate number of direction-finding paths based on the real-time input audio information to complete hazard source direction identification. Through the above process, an efficient hazard source direction-finding model is generated based on the hazard source orientation test data. Through multi-path supervised learning, this model can dynamically select direction-finding paths according to the actual environment, improve the accuracy and robustness of hazard source direction identification, and provide accurate protection for ship navigation safety.

[0033] The hazard source distance analysis module is used to obtain the navigation speed of the ship, and perform hazard source distance analysis based on the audio variation to obtain the hazard source distance.

[0034] Specifically, in the hazard source distance analysis module, the relative distance between the hazard source and the ship can be inferred based on the ship's current navigation speed and current audio variability, combined with a pre-constructed hazard source distance prediction path. This prediction path is trained using a large amount of historical navigation data and the actual distance to the hazard source, and can comprehensively consider the impact of different speed and audio variability combinations on distance. The prediction path predicts the hazard source distance based on the input data and uses it as the basic data for subsequent hazard level classification and emergency alarm discrimination. If the ship's speed is fast but the audio variability is small, it indicates that the sound signal characteristics of the hazard source do not change significantly, and it is inferred that the hazard source may be farther away. This is because fast-moving ships usually cause more significant audio feature changes for closer hazard sources. Conversely, if the ship's speed is fast and the audio variability is large, it may indicate that the hazard source is close to the ship and its audio features change rapidly as the ship approaches. In this way, the distance to the hazard source can be dynamically analyzed, helping ships to more accurately determine the location of potential threats and provide support for navigation safety.

[0035] Furthermore, the navigation speed of the ship is obtained, and combined with the audio variation, the hazard source distance analysis is performed to obtain the hazard source distance, including:

[0036] Obtain the navigation speed of the ship; input the navigation speed and audio variation into a pre-trained hazard source distance prediction path, and output the hazard source distance, wherein the hazard source distance prediction path is obtained by training using sample navigation speed, sample audio variation, and sample hazard source distance.

[0037] In a preferred embodiment, the current navigation speed of the ship is obtained through the ship navigation system or speed sensor, and then the obtained navigation speed is converted into a standardized input form through maximum and minimum value normalization to facilitate the subsequent calculation of the hazard source distance prediction path; then, the audio change degree and the standard navigation speed at the current moment are input together into the pre-trained hazard source distance prediction path, and the hazard source distance is output after the hazard source distance prediction path is calculated. The construction method of this hazard source distance prediction path is the same as the aforementioned construction of the hazard source direction-finding path, which is based on the sample navigation speed, sample audio change degree and sample hazard source distance, and is iteratively trained through forward propagation, loss calculation, back propagation, parameter optimization and other steps, which will not be repeated here.

[0038] The navigation emergency alarm identification module is used to configure the danger level classification accuracy according to the direction change degree, classify the danger level according to the direction change degree and the distance to the danger source, obtain the danger level, perform navigation emergency alarm identification, and obtain the navigation emergency alarm result.

[0039] Specifically, in the navigation emergency alarm identification module, the danger level of the danger source is comprehensively analyzed according to the distance and direction change between the ship and the danger source. By combining the direction change and the distance to the danger source, the degree of threat posed by the danger source to the ship can be judged. If the danger source is far away and the direction change is large, it means that the direction relationship between the danger source and the ship is unstable and relatively safe, and the danger level is low. On the contrary, if the danger source is close and the direction change is small, it means that the direction of the danger source is highly consistent with the direction of the ship, and the ship may be approaching the danger source. At this time, the danger level is high, indicating that the danger source poses a greater threat to the ship. During the entire analysis process, the accuracy of the danger level classification will be dynamically configured according to the current direction change. For example, When the directional change is small, higher classification accuracy is required to more accurately determine the potential threat of the hazard source. Hazard level classification is performed using a pre-trained hazard level classification model. This model is trained using sample directional change, sample hazard source distance, and sample hazard level. It outputs the corresponding hazard level based on the current distance and directional change. In actual operation, the directional change and hazard source distance are input into the hazard level classification model. Through integrated analysis, the model combines the output results of multiple classification paths to calculate a comprehensive hazard level. If the hazard level exceeds the preset hazard level threshold, an emergency alarm is triggered, prompting the ship to take appropriate safety measures. Otherwise, the monitoring state is maintained to continue tracking the dynamic changes of the hazard source. Through this analysis method, the threat level of the hazard source can be effectively identified and judged, providing accurate early warning support for the navigation safety of ships.

[0040] Furthermore, according to the direction change degree, the danger level classification accuracy is configured, and according to the direction change degree and the distance to the danger source, the danger level is classified to obtain the danger level, and the navigation emergency alarm is determined to obtain the navigation emergency alarm result, including:

[0041] Obtain the average directional change when a hazard source appears; use the ratio of the average directional change to the directional change, multiply it by the number of hazard level classification paths, to obtain the current number of hazard level classification paths as the hazard level classification accuracy; randomly select hazard level classification paths within the pre-trained hazard level classification model, the number of hazard level classification paths within the hazard level classification accuracy, wherein the hazard level classification model includes hazard level classification paths of the number of hazard level classification paths, and each hazard level classification path is trained using different sample directional change, sample hazard source distance and sample hazard level; input the directional change and hazard source distance into the randomly selected hazard level classification path, output the hazard level set, and calculate the mean to obtain the hazard level; determine whether the hazard level is greater than the hazard level threshold, and obtain the navigation emergency alarm result.

[0042] In a preferred embodiment, the same method as the above-mentioned calculation of the average audio frequency change during navigation is used to obtain the average direction change when a danger source appears, and the ratio of the average direction change to the current direction change is calculated to indicate the intensity of the current direction change relative to the average change degree when the danger source appears in history. The calculated ratio is then multiplied by the total number of danger level classification paths (which are components of the danger level classification model used for danger level classification), and the product is processed by rounding down or rounding off to obtain the current number of danger level classification paths. This current number of danger level classification paths is used as the current danger level classification accuracy, that is, the number of classification paths used for actual analysis of the danger level. It should be noted that when the direction change is less than the average direction change, that is, the calculated ratio is greater than 1, the calculated ratio is reset to 1. This is because it is necessary to avoid the current number of danger level classification paths exceeding the total number of danger level classification paths. Subsequently, the pre-trained danger level classification model is loaded, which contains the danger level classification of the number of danger level classification paths. Classification paths. Each classification path is trained using sample data (sample direction change, hazard source distance, and hazard level) in the same way as described above. In the hazard level classification model, a hazard level classification path that meets the current number of hazard level classification paths within the hazard level classification accuracy is randomly selected from the hazard level classification paths for hazard level classification analysis at the current moment. Subsequently, the current direction change and hazard source distance are input into these selected hazard level classification paths for calculation. Each classification path outputs a hazard level value. These hazard level values ​​are combined to form a hazard level set. Then, the average of all hazard level values ​​in the hazard level set is calculated to obtain the final hazard level at the current moment. This hazard level is compared with the hazard level threshold (determined based on business needs and expert decision-making to determine whether the hazard source poses an emergency threat to the ship). If the hazard level is greater than the hazard level threshold, it is determined that the hazard level of the current hazard source has reached an emergency state and a navigation emergency alarm is triggered. Otherwise, it is considered that the hazard source threat has not yet reached an emergency state and only the hazard level result is recorded.

[0043] In summary, the navigation emergency alarm device with sound source direction finding provided by the embodiments of the present application has the following technical effects:

[0044] The hazard source direction-finding accuracy configuration module is used to collect an audio information array during navigation through a microphone array, and configure the hazard source direction-finding accuracy according to the audio change degree between each audio information array and the historical audio information array at the previous moment; the direction change degree calculation module is used to perform hazard source direction-finding identification on the audio information array according to the hazard source direction-finding accuracy, obtain the direction of the hazard source sound source, and calculate the direction change degree in combination with the historical hazard source sound source direction at the previous moment; the hazard source distance analysis module is used to obtain the navigation speed of the ship, perform hazard source distance analysis in combination with the audio change degree, and obtain the hazard source distance; the navigation emergency alarm discrimination module is used to configure the hazard level classification accuracy according to the direction change degree, perform hazard level classification according to the direction change degree and hazard source distance, obtain the hazard level, perform navigation emergency alarm discrimination, and obtain the navigation emergency alarm result. Through the above steps, the technical problem of being unable to accurately locate the danger source and its direction and distance being unable to be judged in real time due to environmental noise interference and dynamic changes in the direction of the danger source during navigation, thus affecting navigation safety, is solved. The dynamic direction finding and distance analysis of the audio information array are used to quickly identify the direction changes and distance of the danger source, and emergency alarm judgment is made according to the accuracy of the danger level classification, thereby improving the danger source detection accuracy and alarm response speed and ensuring navigation safety.

[0045] The second embodiment is based on the same inventive concept as the navigation emergency alarm device for sound source direction finding in the above embodiment. Figure 2 As shown, an embodiment of the present application provides a navigation emergency alarm method based on sound source direction finding, the method comprising:

[0046] Through the microphone array, an audio information array is collected during navigation, and the danger source direction finding accuracy is configured according to the audio change degree between each audio information array and the historical audio information array at the previous moment; according to the danger source direction finding accuracy, the audio information array is subjected to danger source direction finding identification to obtain the direction of the danger source sound source, and the direction change degree is calculated in combination with the historical danger source sound source direction at the previous moment; the navigation speed of the ship is obtained, and the danger source distance analysis is performed in combination with the audio change degree to obtain the danger source distance; according to the direction change degree, the danger level classification accuracy is configured, and the danger level classification is performed according to the direction change degree and the danger source distance to obtain the danger level, and the navigation emergency alarm is judged to obtain the navigation emergency alarm result.

[0047] Furthermore, the microphone array is used to collect audio information arrays during navigation. Based on the audio change degree of each audio information array compared with the historical audio information array at the previous moment, the hazard source direction finding accuracy is configured, including:

[0048] Through the microphone array, an audio information array is collected during the navigation process to obtain the audio information array at the current moment and the historical audio information array at the previous moment; the audio feature information array and the historical audio feature information array are extracted; based on the audio feature information array and the historical audio feature information array, the audio change degree is calculated; based on the audio change degree, the danger source direction finding accuracy is configured.

[0049] Furthermore, configuring the danger source direction finding accuracy according to the audio variation degree includes:

[0050] The average audio variation during navigation is obtained; the ratio of the audio variation to the average audio variation is multiplied by the number of hazard source direction finding paths and rounded to the nearest integer to obtain the current number of hazard source direction finding paths as the hazard source direction finding accuracy.

[0051] Furthermore, according to the hazard source direction finding accuracy, the audio information array is subjected to hazard source direction finding identification to obtain the hazard source sound direction, and the direction change degree is calculated based on the historical hazard source sound direction at the previous moment, including:

[0052] In the pre-trained hazard source direction-finding model, a hazard source direction-finding path is randomly selected according to the number of current hazard source direction-finding paths within the hazard source direction-finding accuracy, wherein the hazard source direction-finding model includes hazard source direction-finding paths of the number of hazard source direction-finding paths; the audio information array is input into the randomly selected hazard source direction-finding path, and a set of hazard source sound source directions is obtained as output; the mean of the set of hazard source sound source directions is calculated to obtain the hazard source sound source direction; the historical hazard source sound source direction at the previous moment is obtained, and the degree of change in direction between the hazard source sound source direction and the historical hazard source sound source direction is calculated.

[0053] Furthermore, the pre-training step of the hazard source direction finding model includes:

[0054] Based on the hazard source orientation test data records, a sample audio information array set is collected, and the hazard source directions corresponding to different sample audio information arrays are marked to obtain a sample hazard source sound source direction set; the sample audio information array set and the sample hazard source sound source direction set are divided to obtain multiple supervisory data on the number of hazard source direction finding paths; using the multiple supervisory data on the number of hazard source direction finding paths, supervised training is used respectively to train a number of hazard source direction finding paths, and the hazard source direction finding paths are combined to obtain a hazard source direction finding model.

[0055] Furthermore, the navigation speed of the ship is obtained, and combined with the audio variation, the hazard source distance analysis is performed to obtain the hazard source distance, including:

[0056] Obtain the navigation speed of the ship; input the navigation speed and audio variation into a pre-trained hazard source distance prediction path, and output the hazard source distance, wherein the hazard source distance prediction path is obtained by training using sample navigation speed, sample audio variation, and sample hazard source distance.

[0057] Furthermore, according to the direction change degree, the danger level classification accuracy is configured, and according to the direction change degree and the distance to the danger source, the danger level is classified to obtain the danger level, and the navigation emergency alarm is determined to obtain the navigation emergency alarm result, including:

[0058] Obtain the average directional change when a hazard source appears; use the ratio of the average directional change to the directional change, multiply it by the number of hazard level classification paths, to obtain the current number of hazard level classification paths as the hazard level classification accuracy; randomly select hazard level classification paths within the pre-trained hazard level classification model, the number of hazard level classification paths within the hazard level classification accuracy, wherein the hazard level classification model includes hazard level classification paths of the number of hazard level classification paths, and each hazard level classification path is trained using different sample directional change, sample hazard source distance and sample hazard level; input the directional change and hazard source distance into the randomly selected hazard level classification path, output the hazard level set, and calculate the mean to obtain the hazard level; determine whether the hazard level is greater than the hazard level threshold, and obtain the navigation emergency alarm result.

[0059] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0060] Furthermore, the terms "first" or "second" as described above may not only represent an order relationship but may also represent a specific concept and / or refer to the selectability of multiple elements individually or collectively. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, if such modifications and variations fall within the scope of this application and its equivalents, this application is intended to include such modifications and variations.

Claims

1. A navigation emergency alarm device with sound source direction finding, characterized in that: The device comprises: The hazard source direction finding accuracy configuration module is used to collect audio information arrays during navigation through a microphone array. The hazard source direction finding accuracy is configured based on the audio change between each audio information array and the historical audio information array at the previous moment. The module includes: The microphone array is used to collect audio information arrays during navigation to obtain the current audio information array and the historical audio information array of the previous moment; Extracting an audio feature information array and a historical audio feature information array from the audio information array and the historical audio information array; Calculating and obtaining an audio change degree based on the audio feature information array and the historical audio feature information array; According to the audio variation, the hazard source direction finding accuracy is configured, including: Get the average audio frequency variation during navigation; The ratio of the audio variation to the average audio variation is multiplied by the number of hazard source direction finding paths and rounded to an integer to obtain the current number of hazard source direction finding paths as the hazard source direction finding accuracy. A direction change degree calculation module is used to perform hazard source direction finding and identification on the audio information array according to the hazard source direction finding accuracy, obtain the direction of the hazard source sound source, and calculate the direction change degree based on the historical hazard source sound source direction at the previous moment; A hazard source distance analysis module is used to obtain the navigation speed of the ship and perform hazard source distance analysis based on the audio variation to obtain the hazard source distance; The navigation emergency alarm discrimination module is used to configure the danger level classification accuracy according to the direction change degree, classify the danger level according to the direction change degree and the distance to the danger source, obtain the danger level, perform navigation emergency alarm discrimination, and obtain the navigation emergency alarm result, including: Obtain the average degree of direction change when a hazard source occurs; The ratio of the average direction change degree to the direction change degree is multiplied by the number of hazard level classification paths to obtain the current number of hazard level classification paths as the hazard level classification accuracy; In the pre-trained hazard level classification model, hazard level classification paths of the number of current hazard level classification paths within the hazard level classification accuracy are randomly selected, wherein the hazard level classification model includes hazard level classification paths of the number of hazard level classification paths, and each hazard level classification path is trained using different sample direction change degrees, sample hazard source distances and sample hazard levels.

2. The navigation emergency alarm device with sound source direction finding according to claim 1, characterized in that: According to the hazard source direction finding accuracy, the audio information array is subjected to hazard source direction finding and identification to obtain the hazard source sound direction. Combined with the historical hazard source sound direction at the previous moment, the direction change degree is calculated, including: In a pre-trained hazard source direction finding model, randomly selecting a hazard source direction finding path according to the number of current hazard source direction finding paths within the hazard source direction finding accuracy, wherein the hazard source direction finding model includes hazard source direction finding paths of the number of hazard source direction finding paths; Input the audio information array into a randomly selected hazard source direction finding path, and output a set of hazard source sound direction; Calculating the mean of the set of dangerous sound source directions to obtain the dangerous sound source direction; The direction of the sound source of the historical hazard source at the previous moment is obtained, and the degree of change between the direction of the sound source of the hazard source and the direction of the historical hazard source sound is calculated.

3. The navigation emergency alarm device with sound source direction finding according to claim 2, characterized in that: The pre-training steps of the hazard source direction finding model include: Based on the hazard source orientation test data record, a sample audio information array set is collected, and the hazard source directions corresponding to different sample audio information arrays are marked to obtain a sample hazard source sound source direction set; Dividing the sample audio information array set and the sample hazard source sound source direction set to obtain multiple sets of supervisory data on the number of hazard source direction-finding paths; Using multiple supervision data of the number of hazardous source direction finding paths, supervised training is used to train the hazardous source direction finding paths of the number of hazardous source direction finding paths, and the hazardous source direction finding model is obtained by combining them.

4. The navigation emergency alarm device with sound source direction finding according to claim 1, characterized in that: Obtaining the ship's navigation speed, combining it with the audio variation, and performing hazard source distance analysis to obtain the hazard source distance, including: Get the ship's sailing speed; The navigation speed and audio variation are input into a pre-trained hazard source distance prediction path, and the hazard source distance is output, wherein the hazard source distance prediction path is obtained by training using sample navigation speed, sample audio variation and sample hazard source distance.

5. The navigation emergency alarm device with sound source direction finding according to claim 1, characterized in that: According to the direction change degree, the danger level classification accuracy is configured, and according to the direction change degree and the distance to the danger source, the danger level is classified to obtain the danger level, and the navigation emergency alarm is judged to obtain the navigation emergency alarm result, including: Input the direction change degree and the distance to the hazard source into a randomly selected hazard level classification path, output a hazard level set, and calculate the mean to obtain the hazard level; Determine whether the danger level is greater than the danger level threshold and obtain a navigation emergency alarm result.

6. A navigation emergency alarm method based on sound source direction finding, characterized in that: The navigation emergency alarm method using sound source direction finding is performed by a navigation emergency alarm device using sound source direction finding according to any one of claims 1 to 5, comprising: The microphone array collects audio information arrays during navigation, and configures the hazard source direction finding accuracy based on the audio change degree of each audio information array compared to the historical audio information array at the previous moment; According to the hazard source direction finding accuracy, the audio information array is subjected to hazard source direction finding and identification to obtain the hazard source sound direction, and the direction change degree is calculated based on the historical hazard source sound direction at the previous moment; Obtaining the navigation speed of the ship, and combining it with the audio variation, performing hazard source distance analysis to obtain the hazard source distance; According to the direction change degree, the danger level classification accuracy is configured, and according to the direction change degree and the distance to the danger source, the danger level is classified to obtain the danger level, and the navigation emergency alarm is judged to obtain the navigation emergency alarm result.

Images