Deep learning AUV (Autonomous Underwater Vehicle) adaptive navigation control device under complex sea condition

Abstract

The invention provides a deep learning AUV adaptive navigation control device under complex sea conditions, and relates to the field of AUV navigation devices. The deep learning AUV adaptive navigation control device under the complex sea condition comprises environment perception, environment modeling, motion modeling, disturbance modeling, a central processing module, implementation control, result summarization, data storage, training model creation and manual control, and the environment perception is divided into a water flow sensor, a wave sensor, an ultrasonic sensor, a gyroscope and a GPS module. According to the system, the real-time learning capability is set, the system can update and optimize the control model while executing the task through the online learning technology, and the system can immediately utilize new environment data for training to adapt to constantly changing conditions, so that the accuracy and efficiency of decision making are improved, and the system is suitable for large-scale popularization and application. The continuously improved capability enables the AUV to more flexibly execute tasks in complex and dynamic sea conditions, and improves the success rate and efficiency of the overall operation.

Description

Technical Field

[0001] The present invention relates to the field of AUV navigation devices, specifically a deep learning AUV adaptive navigation control device under complex sea conditions. Background Art

[0002] Autonomous underwater vehicle navigation control devices are mainly used to achieve underwater autonomous navigation and task execution. They rely on rule-based methods, static models, and sensors to collect environmental information, so as to plan routes and adjust headings. An AUV is an unmanned submersible that can operate autonomously underwater and is widely used in fields such as marine scientific research, resource exploration, and environmental monitoring. It is usually battery-powered and navigates and collects data through preset tasks without human operation. The shape of an AUV is usually relatively compact, with a watertight shell and various sensors, capable of sensing the surrounding environment and performing complex tasks.

[0003] However, traditional AUV navigation control devices mainly rely on rule-based methods and static models, resulting in insufficient adaptability in complex sea conditions. When the marine environment changes rapidly, it is difficult for traditional systems to adjust their control strategies in real time and usually can only rely on preset navigation parameters. This fixed control method limits the flexibility and responsiveness of AUVs in dynamic environments, affecting the stability and accuracy of navigation.

[0004] Therefore, those skilled in the art have provided a deep learning AUV adaptive navigation control device under complex sea conditions to solve the problems raised in the above background art. Summary of the Invention

[0005] (1) Technical Problems to be Solved

[0006] In view of the deficiencies of the prior art, the present invention provides a deep learning AUV adaptive navigation control device under complex sea conditions, which solves the problems of poor adaptability and low environmental detection range.

[0007] (2) Technical Solutions

[0008] To achieve the above objectives, the present invention is realized through the following technical solutions: A deep learning AUV adaptive navigation control device under complex sea conditions includes environmental perception, environmental modeling, motion modeling, disturbance modeling, a central processing module, implementation control, result summary, data storage, training model creation, and manual control. The environmental perception is divided into a water flow sensor, a wave sensor, an ultrasonic sensor, a gyroscope, and a GPS module, which are used to detect and collect data on the marine environment around the AUV; The environmental modeling is divided into a data fusion module and an environmental feature extraction module, which are used to create a model reflecting the current environmental state. By integrating data from different sensors, a comprehensive environmental description is generated, including information such as flow velocity, temperature, salinity, etc. The accurate environmental model provides important background information for the movement and control of the AUV; The motion modeling is divided into a motion equation model and an AUV model analysis, which are used to describe and predict the motion behavior of the AUV during travel. The motion modeling describes the way the AUV moves in water by establishing a mathematical model, taking into account the initial position, velocity, and acceleration. These models help predict the position and heading of the AUV at a specific time, thus providing a basis for subsequent control decisions; The perturbation modeling is divided into a wave and current model and a sensor feedback model, which are used to identify and quantify the impact of the external environment on the AUV's motion. The perturbation modeling establishes an impact model of external factors such as water flow and waves on the AUV's control. By modeling these perturbations, the system can adjust the navigation strategy in real time to ensure navigation stability in complex sea conditions; The central processing module is divided into a central processor and a learning model, which are used to integrate all input data, perform analysis and decision-making, and thus generate control instructions to ensure the stable navigation of the AUV in complex sea conditions; The implementation control is divided into an automatic controller and a control algorithm module, which are used to adjust the motion state of the AUV in real time to achieve the goal. The implementation control module is responsible for generating control instructions based on the outputs of the environmental, perturbation, and motion models, and ensuring that the AUV travels along the predetermined route by controlling each drive control device on the instruction control device. This module responds to errors in real time through the central processor to optimize the heading and speed; The result summary is used to analyze the navigation results of the AUV to evaluate its performance. After the navigation is completed, the operation results are evaluated. By comparing the actual navigation trajectory with the expected trajectory, the error is calculated and potential problems are identified, providing feedback for future navigation; The data storage is used to safely store and manage navigation data and model information, and is responsible for saving sensor data, navigation records, model parameters, and training results. These data can not only be used for subsequent analysis and training, but also provide support for fault diagnosis and system maintenance to ensure data integrity and availability; The training model creation is divided into a machine learning algorithm module and a model training data set, which are used to improve the intelligent decision-making ability of the AUV by learning historical data. Machine learning algorithms are used to analyze historical data to create models that perform adaptive decision-making under multiple different environmental conditions. Through continuous training, the models improve their response ability to dynamic environments; The manual control is divided into a user interface module and a manual control interface, which are used for manual intervention and adjustment of the AUV's navigation state, providing a way for the operator to manually intervene in the control system when needed. Through a user-friendly interface, the operator can monitor the system performance and adjust the navigation strategy according to the actual situation to improve safety.

[0009] Through the above technical solutions, the deep learning AUV adaptive navigation control device in complex sea conditions integrates multiple sensors. Through the sensor fusion and environmental feature extraction module, the perception ability of the complex marine environment is significantly enhanced. The device system forms a comprehensive environmental model by integrating multi-source data. This accurate environmental perception enables the AUV to more effectively identify potential risks, improving the safety and reliability during navigation. Combined with the motion modeling for azimuth modeling of the device itself, the internal and external conditions of the device during operation are detected in terms of data, thereby improving the accuracy of the device detection.

[0010] Further, the water flow sensor is used to measure the water flow speed and direction in the water body; the wave sensor is used to detect the height, frequency and direction of the surface waves; the ultrasonic sensor is used to measure the distance and position of underwater obstacles through ultrasonic ranging technology; the gyroscope is used to measure the attitude changes of the AUV, including pitch, yaw and roll, etc.; the GPS module is used to provide the accurate geographical location and navigation speed of the AUV. Through the above technical solutions, the AUV collects environmental data through multiple sensors. The water flow and wave sensors monitor the water body conditions in real time, the gyroscope provides heading and attitude information, and the GPS is used for positioning. These data are used to evaluate the current navigation environment and provide a basis for subsequent modeling and control.

[0011] Further, the data fusion module is used to integrate data from different sensors to output a more accurate and consistent environmental model; the environmental feature extraction module is used to analyze the fused data and extract the key features in the environment. Through the above technical solutions, the data from different sensors are integrated, and the accuracy of perception is improved through data fusion algorithms (such as Kalman filtering). The environmental feature extraction module analyzes the fused data and identifies important environmental features, such as flow velocity, wave height, etc., providing support for the control system.

[0012] Further, the motion equation model is used to establish a mathematical model to describe the motion laws of the AUV, including position, velocity and acceleration; the AUV model analysis is used to evaluate and analyze the dynamic behavior and performance of the AUV under specific environmental conditions. Through the above technical solution, the AUV model analysis and establishment device establishes its own motion model, and the motion equation model forms a dynamic system representation of the AUV under complex sea conditions, which ensures that the AUV can make accurate motion predictions based on its current speed, acceleration, and external disturbances.

[0013] Furthermore, the wave and current model is a mathematical model established for the characteristics of waves and currents, describing their motion laws under specific environmental conditions; the sensor feedback model is used to adjust the control strategy through real-time sensor data feedback; Through the above technical solution, waves and ocean currents are extremely complex and irregular random waves, which theoretically cannot be replaced by any regular waves. However, according to the statistical laws of waves, they can be replaced by the superposition of multiple simple regular waves with different amplitudes, frequencies, directions, and phases to establish the wave and current model. The sensor feedback model uses real-time sensor data to adjust the disturbance model to make it more suitable for the actual environment and dynamically updates to cope with sea condition changes.

[0014] Furthermore, the central processing unit is responsible for integrating the data processing and instruction generation of each module; the learning model is used to analyze historical data using machine learning algorithms to improve the intelligent decision-making ability of the system; Through the above technical solution, the central processing unit processes all the collected data, applies deep learning algorithms (such as reinforcement learning) to analyze and make decisions on motion and environment. The learning model is continuously updated based on historical data to optimize the control strategy and make it have the ability of self-adaptation.

[0015] Furthermore, the automatic controller is used to execute actual control operations according to the control instructions generated by the central processing unit; the control algorithm module is used to implement specific control algorithms to ensure that under the given target state, the error is minimized and the performance of the AUV is optimized; Through the above technical solution, according to the output signal of the central processing module, the automatic controller generates control instructions, and the control algorithm (such as a PID controller) adjusts the navigation state based on the real-time error to ensure that the AUV can move forward stably, avoid collisions, and achieve the established route in complex sea conditions.

[0016] Furthermore, the machine learning algorithm module is used to implement and apply various machine learning algorithms to process complex data sets to improve the system performance; the model training data set is used for the training of machine learning and deep learning models, including historical navigation data, sensor data, and corresponding labels; Through the above technical solution, based on the data summarized from the results, the training model creation module uses the existing data for training. Through feedback learning, the model is continuously optimized, enabling the AUV to flexibly cope with future complex situations.

[0017] Further, the user interface module is used to display status information and control options; the manual control interface is used to allow manual intervention in control instructions. Through the above technical solution, a clear user interface is provided for the operator, enabling real-time monitoring of the navigation status and environmental information. The operator can adjust the AUV through the manual control interface in case of emergency to ensure safe navigation.

[0018] The present invention provides a deep learning AUV adaptive navigation control device under complex sea conditions, having the following beneficial effects: 1. The present invention provides a deep learning AUV adaptive navigation control device under complex sea conditions. Compared with traditional AUV navigation control devices, this deep learning AUV adaptive navigation control device under complex sea conditions can continuously learn and extract complex features from a large amount of historical and real-time data by applying machine learning algorithms, establish a model of the environment where the device is located through environmental modeling, and compare it with the preset motion modeling and disturbance modeling to adjust the device in various azimuth angles, enabling the device to make different adjustments in different environments. This adaptive ability enables the AUV to more quickly adapt to the rapidly changing environmental conditions of waves and water flow. Different from traditional rule-based methods, the device can dynamically adjust its control strategy to cope with the uncertainties in the actual environment, thereby improving navigation stability and accuracy, and enhancing the adaptability of the device.

[0019] 2. The present invention provides a deep learning AUV adaptive navigation control device under complex sea conditions. Compared with traditional AUV navigation control devices, this deep learning AUV adaptive navigation control device under complex sea conditions integrates multiple sensors. Through the sensor fusion and environmental feature extraction module, the perception ability of complex marine environments is significantly enhanced. The device system forms a comprehensive environmental model by integrating multi-source data. This accurate environmental perception enables the AUV to more effectively identify potential risks, improves the safety and reliability during navigation, and cooperates with the motion modeling for azimuth modeling of the device itself to detect the internal and external data conditions of the device during operation, thereby improving the detection accuracy of the device.

[0020] 3. The present invention provides a deep learning AUV adaptive navigation control device under complex sea conditions. Compared with traditional AUV navigation control devices, this deep learning AUV adaptive navigation control device under complex sea conditions is equipped with real-time learning capabilities. Through online learning technology, the system can update and optimize the control model while performing tasks. Different from traditional AUVs that need to summarize and analyze after navigation, this system can immediately use new environmental data for training to adapt to changing conditions, thereby improving the accuracy and efficiency of decision-making. This continuous improvement ability enables the AUV to perform tasks more flexibly in complex and dynamic sea conditions, improving the overall operation success rate and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the working process of the present invention; Figure 2 It is a schematic diagram of the composition of the environmental perception and central processing module of the present invention; Figure 3 It is a schematic diagram of the composition of the environmental modeling and motion modeling of the present invention; Figure 4 It is a schematic diagram of the composition of the disturbance modeling and manual control of the present invention; Figure 5 It is a schematic diagram of the composition of the implementation control and training model creation of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention. Embodiment

[0023] As Figures 1-5As shown in the figure, the embodiment of the present invention provides a deep learning AUV adaptive navigation control device under complex sea conditions, including environmental perception, environmental modeling, motion modeling, disturbance modeling, a central processing module, implementation control, result summary, data storage, training model creation, and manual control. Environmental perception is divided into a water flow sensor, a wave sensor, an ultrasonic sensor, a gyroscope, and a GPS module, which are used to detect and collect data on the marine environment around the AUV. Environmental modeling is divided into a data fusion module and an environmental feature extraction module, which are used to create a model reflecting the current environmental state. By integrating data from different sensors, a comprehensive environmental description is generated, including information such as flow velocity, temperature, and salinity. The accurate environmental model provides important background information for the motion and control of the AUV. Motion modeling is divided into a motion equation model and an AUV model analysis, which are used to describe and predict the motion behavior of the AUV during navigation. Motion modeling establishes a mathematical model to describe the way the AUV moves in water, considering the initial position, speed, and acceleration. These models help predict the position and heading of the AUV at a specific time, thus providing a basis for subsequent control decisions; Disturbance modeling is divided into a wave and current model and a sensor feedback model, which are used to identify and quantify the impact of the external environment on the motion of the AUV. Disturbance modeling establishes an impact model of external factors such as water flow and waves on the control of the AUV. By modeling these disturbances, the system can adjust the navigation strategy in real time to ensure navigation stability in complex sea conditions. The central processing module is divided into a central processor and a learning model, which are used to integrate all input data, analyze and make decisions, and thus generate control instructions to ensure the stable navigation of the AUV in complex sea conditions. Implementation control is divided into an automatic controller and a control algorithm module, which are used to adjust the motion state of the AUV in real time to achieve the goal. The implementation control module is responsible for generating control instructions based on the outputs of the environment, disturbance, and motion models, and controlling each drive control device on the device through the instructions to ensure that the AUV travels along the predetermined route. This module responds to errors in real time through the central processor to optimize the heading and speed; The result summary is used to analyze the navigation results of the AUV to evaluate the performance. After the navigation is completed, the operation results are evaluated. By comparing the actual navigation trajectory with the expected trajectory, the error is calculated and potential problems are identified to provide feedback for future navigation. Data storage is used to safely store and manage navigation data and model information, and is responsible for saving sensor data, navigation records, model parameters, and training results. These data can not only be used for subsequent analysis and training, but also provide support for fault diagnosis and system maintenance to ensure data integrity and availability; The creation of the training model is divided into a machine learning algorithm module and a model training dataset, which are used to improve the intelligent decision-making ability of the AUV by learning historical data. Machine learning algorithms are used to analyze historical data to create models for performing adaptive decisions under multiple different environmental conditions. Through continuous training, the model has improved its response ability to dynamic environments. Manual control is divided into a user interface module and a manual control interface, which are used for manual intervention and adjustment of the AUV's navigation state, providing a way for the operator to manually intervene in the control system when needed. Through a user-friendly interface, the operator can monitor the system performance and adjust the navigation strategy according to the actual situation to improve safety.

[0024] The deep learning AUV adaptive navigation control device under complex sea conditions integrates a variety of sensors. Through the sensor fusion and environmental feature extraction module, it significantly enhances the perception ability of complex marine environments. The device system forms a comprehensive environmental model by integrating multi-source data. This precise environmental perception enables the AUV to more effectively identify potential risks and improves the safety and reliability during navigation. Combined with the motion modeling that performs azimuth modeling on the device itself, it detects the data situation inside and outside the device during operation, thereby improving the accuracy of the device's detection.

[0025] The water flow sensor is used to measure the water flow speed and direction in the water body. The wave sensor is used to detect the height, frequency, and direction of surface waves. The ultrasonic sensor is used to measure the distance and position of underwater obstacles through ultrasonic ranging technology. The gyroscope is used to measure the attitude changes of the AUV, including pitch, yaw, and roll, etc. The GPS module is used to provide the precise geographical location and navigation speed of the AUV. The AUV collects environmental data through a variety of sensors. The water flow and wave sensors monitor the water body situation in real time. The gyroscope provides heading and attitude information. The GPS is used for positioning. These data are used to evaluate the current navigation environment and provide a basis for subsequent modeling and control.

[0026] The data fusion module is used to integrate data from different sensors to output a more accurate and consistent environmental model. The environmental feature extraction module is used to analyze the fused data and extract key features in the environment. By integrating data from different sensors and using data fusion algorithms (such as Kalman filtering), the accuracy of perception is improved. The environmental feature extraction module analyzes the fused data and identifies important environmental features, such as flow velocity, wave height, etc., to provide support for the control system.

[0027] Motion equation model is: the position vector at time t, is: the initial position vector, (where \(v\) is the velocity vector, \(a\) is the acceleration vector, and \(t\) is time) is used to establish a mathematical model to describe the motion law of the AUV, including position, velocity, and acceleration. The AUV model analysis is used to evaluate and analyze the dynamic behavior and performance of the AUV under specific environmental conditions. The AUV model analysis establishes the motion model of the device itself, and the motion equation model forms the dynamic system representation of the AUV under complex sea conditions, which ensures that the AUV can make accurate motion predictions based on its current velocity, acceleration, and external disturbances.

[0028] The wave and current models are mathematical models established for the characteristics of waves and currents to describe their motion laws under specific environmental conditions. The sensor feedback model is used to adjust the control strategy through real-time sensor data feedback. Waves and currents are extremely complex and irregular random waves, and theoretically cannot be replaced by any regular waves. However, according to the statistical laws of waves, they can be replaced by the superposition of multiple simple regular waves with different amplitudes, frequencies, directions, and phases to establish the wave and current models. The sensor feedback model uses real-time sensor data to adjust the disturbance model to make it more conform to the actual environment and dynamically update to cope with sea condition changes.

[0029] The central processor is responsible for integrating the data processing and instruction generation of each module. The learning model is used to analyze historical data using machine learning algorithms to improve the intelligent decision-making ability of the system. The central processor processes all the collected data, applies deep learning algorithms (such as reinforcement learning) to analyze and make decisions on motion and environment. The learning model is continuously updated according to historical data to optimize the control strategy and make it have adaptive capabilities.

[0030] The automatic controller is used to execute actual control operations according to the control instructions generated by the central processor. The control algorithm module is used to implement specific control algorithms to ensure that under the given target state, the error is minimized and the performance of the AUV is optimized. According to the output signal of the central processing module, the automatic controller generates control instructions, and the control algorithm (such as the PID controller) adjusts the navigation state according to the real-time error to ensure that the AUV can move forward stably, avoid collisions, and achieve the established route in complex sea conditions.

[0031] Machine learning algorithm module , where is the loss function, \(m\) is the number of samples, is the prediction of the model for the input , For: real output) to implement and apply various machine learning algorithms, process complex data sets to improve system performance. The model training data set is used for the training of machine learning and deep learning models, including historical navigation data, sensor data, and corresponding labels, data summarized based on results. The training model creation module uses existing data for training, and through feedback learning, the model is continuously optimized, enabling the AUV to flexibly handle future complex situations.

[0032] The user interface module is used to display status information and control options. The manual control interface is used to allow manual intervention in control commands, providing a clear user interface for the operator, enabling them to monitor the navigation status and environmental information in real time. The operator can adjust the AUV through the manual control interface in case of emergencies to ensure safe navigation.

[0033] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A deep learning AUV adaptive navigation control device under complex sea conditions, including environment perception, environment modeling, motion modeling, disturbance modeling, central processing module, implementation control, result summary, data storage, training model creation and manual control, characterized by: The environmental sensing is divided into water flow sensor, wave sensor, ultrasonic sensor, gyroscope and GPS module, which are used to detect and collect data on the marine environment around the AUV; The environmental modeling is divided into a data fusion module and an environmental feature extraction module, which are used to create a model that reflects the current environmental state. By integrating data from different sensors, a comprehensive environmental description is generated, including information such as flow rate, temperature, and salinity. The accurate environmental model provides important background information for the movement and control of the AUV. The motion modeling is divided into the motion equation model and AUV model analysis, which are used to describe and predict the motion behavior of the AUV while traveling. Motion modeling describes the way the AUV moves in the water by establishing mathematical models. Taking into account the initial position, velocity and acceleration, these models help predict the position and heading of the AUV at a specific time, thereby providing a basis for subsequent control decisions; The disturbance modeling is divided into wave and current models and sensor feedback models, which are used to identify and quantify the impact of the external environment on the movement of the AUV. The disturbance modeling establishes a model of the impact of external factors such as water currents and waves on the control of the AUV. By modeling these disturbances, the system can adjust the navigation strategy in real time to ensure navigation stability in complex sea conditions. The central processing module is divided into a central processing unit and a learning model, which are used to integrate all input data, analyze and make decisions, and thus generate control instructions to ensure that the AUV can navigate stably in complex sea conditions; The implementation control is divided into an automatic controller and a control algorithm module, which are used to adjust the motion state of the AUV in real time to achieve the goal. The implementation control module is responsible for generating control instructions based on the output of the environment, disturbance and motion model, and controls the various drive control devices on the command control device to ensure that the AUV travels along the predetermined route. This module responds to errors in real time through the central processor to optimize the heading and speed; The result summary is used to analyze the navigation results of the AUV to evaluate the performance. After the navigation is completed, the operation results are evaluated to provide feedback for future navigation by comparing the actual navigation trajectory with the expected trajectory, calculating errors and identifying potential problems; The data storage is used to safely store and manage navigation data and model information, and is responsible for preserving sensor data, navigation records, model parameters and training results. These data can not only be used for subsequent analysis and training, but also provide support for fault diagnosis and system maintenance, ensuring data integrity and availability; The training model creation is divided into a machine learning algorithm module and a model training data set, which is used to improve the intelligent decision-making ability of the AUV by learning historical data, using machine learning algorithms to analyze historical data, and creating a model that performs adaptive decision-making under multiple different environmental conditions. Through continuous training, the model improves its responsiveness to dynamic environments; The manual control is divided into a user interface module and a manual control interface, which are used to manually intervene and adjust the navigation status of the AUV, providing a way for the operator to manually intervene in the control system when necessary. Through a user-friendly interface, the operator can monitor the system performance and adjust the navigation strategy according to the actual situation to improve safety.

2. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1 is characterized by: The water flow sensor is used to measure the speed and direction of water flow in the water body; the wave sensor is used to detect the height, frequency and direction of surface waves; the ultrasonic sensor is used to measure the distance and position of underwater obstacles through ultrasonic ranging technology; the gyroscope is used to measure the attitude changes of the AUV, including pitch, yaw and roll, etc.; the GPS module is used to provide the precise geographic location and navigation speed of the AUV.

3. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1 is characterized by: The data fusion module is used to integrate data from different sensors to output a more accurate and consistent environmental model; the environmental feature extraction module is used to analyze the fused data and extract key features in the environment.

4. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1, characterized in that: The motion equation model is used to establish a mathematical model to describe the motion law of the AUV, including position, velocity and acceleration; the AUV model analysis is used to evaluate and analyze the dynamic behavior and performance of the AUV under specific environmental conditions.

5. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1, characterized in that: The wave and current model is a mathematical model established for the characteristics of waves and currents to describe their movement laws under specific environmental conditions; the sensor feedback model is used to adjust the control strategy through real-time sensor data feedback.

6. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1, characterized in that: The central processing unit is responsible for integrating the data processing and instruction generation of each module; the learning model is used to apply machine learning algorithms to analyze historical data, thereby improving the system's intelligent decision-making capabilities.

7. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1, characterized in that: The automatic controller is used to perform actual control operations according to the control instructions generated by the central processing unit; the control algorithm module is used to implement a specific control algorithm to ensure that the error is minimized and the performance of the AUV is optimized under a given target state.

8. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1, characterized in that: The machine learning algorithm module is used to implement and apply various machine learning algorithms and process complex data sets to improve system performance; the model training data set is used for training machine learning and deep learning models, and includes historical navigation data, sensor data and corresponding labels.

9. The deep learning AUV adaptive navigation control device under complex sea conditions according to claim 1, characterized in that: The user interface module is used to display status information and control options; the manual control interface is used to allow manual intervention in control instructions.

Images