Intelligent ice-based towed oceanographic profiling buoy control system, method, and electronic device

Abstract

This invention relates to the field of marine environmental monitoring technology, and provides an intelligent ice-based towed ocean profiling buoy control system, method, and electronic equipment, comprising: a main control module, used for data acquisition according to a currently set sampling interval, and controlling the intelligent computing module to power on after acquisition to send the acquired marine environmental data; the intelligent computing module is used for inference calculation based on the received marine environmental data and locally stored historical sampling data to determine the next sampling interval and feed it back to the main control module; the main control module is also used for controlling the intelligent computing module to power off after receiving the next sampling interval, and waiting to execute the next data acquisition according to the next sampling interval. This invention, by integrating an intelligent computing module on the buoy end side and adopting a main control-triggered intermittent power supply and data interaction strategy, achieves adaptive adjustment of the sampling frequency based on real-time marine environmental perception, improves the integrity of data acquisition, and reduces system operating power consumption.

Description

Technical Field

[0001] This invention relates to the field of marine environmental monitoring technology, and in particular to an intelligent ice-based towed ocean profiling buoy control system, method, and electronic equipment. Background Technology

[0002] Global climate change has made acquiring multi-parameter data on the polar sea-ice-atmosphere interface a research hotspot. Ice-based towed ocean profiling buoy systems are a typical example of such observation systems. These systems typically consist of an ice-based meteorological buoy (located on the ice surface) and an underwater profiling buoy, connected by a high-strength towed cable. The ice-based meteorological buoy drifts with the sea ice, while the underwater profiling buoy periodically performs profiling movements, collecting data and transmitting it back to Earth via the Iridium satellite, thus enabling automatic and continuous measurement of polar environmental meteorological and hydrological elements.

[0003] To ensure long-term, low-power operation in harsh polar environments, traditional ice-based buoy control systems typically employ fixed-time-cycle observation modes, such as initiating profile observations every 12 hours and remaining in standby or dormant mode for the rest of the time. However, when the marine environment changes, such as encountering mesoscale eddies or thermohaline processes, the traditional fixed-time observation mode acquires limited marine environmental data samples, which is insufficient to meet the needs of marine scientific research. Summary of the Invention

[0004] This invention provides an intelligent ice-based towed ocean profiling buoy control system, method, and electronic equipment to address the shortcomings of existing ice-based buoys that suffer from insufficient data sampling rates due to their fixed-period observation mode.

[0005] This invention provides an intelligent ice-based towed ocean profiling buoy control system, comprising: The main control module is used to collect data according to the currently set sampling interval, and after the collection is completed, it controls the intelligent computing module to power on and sends the collected marine environmental data to the intelligent computing module. The intelligent computing module is used to start in response to the power-on control of the main control module, receive the marine environment data sent by the main control module, perform online inference calculation based on the marine environment data and locally stored historical sampling data to determine the next sampling interval, and feed the next sampling interval back to the main control module; The main control module is also used to control the intelligent computing module to power off after receiving the next sampling interval, and wait to execute the next data acquisition according to the next sampling interval.

[0006] According to the present invention, an intelligent ice-based towed ocean profiling buoy control system includes an intelligent computing module comprising: The data storage unit is used to convert the received marine environmental data into a preset format file and then store it locally. The inference calculation unit is used to read a preset number of historical sampling data stored in the data storage unit, and input the historical sampling data and the currently received marine environment data into a preset artificial neural network model for online inference calculation to obtain the current marine environment change probability value output by the artificial neural network model; The decision feedback unit is used to compare the current probability value of marine environmental change with a preset probability threshold, determine the next sampling interval based on the comparison result, and feed the next sampling interval back to the main control module through a communication interface.

[0007] According to the intelligent ice-based towed ocean profiling buoy control system provided by the present invention, the decision feedback unit is further used for: If the probability value of the current marine environment change is greater than the preset probability threshold, it is determined that the current underwater environment has changed, and the next sampling interval is set to the first preset duration corresponding to the encrypted observation mode; If the current probability value of marine environmental change is less than or equal to the preset probability threshold, it is determined that the current underwater environment has not changed, and the next sampling interval is set to the second preset duration corresponding to the normal observation mode; Wherein, the first preset duration is less than the second preset duration.

[0008] According to the present invention, an intelligent ice-based towed ocean profiling buoy control system is provided, wherein the artificial neural network model is a time-series mixture model, and the artificial neural network model is configured to support quantization acceleration. The intelligent computing module is also used to adjust the input and output dimensions of the artificial neural network model, the preset probability threshold, and the preset quantity through a configuration file.

[0009] According to the present invention, a smart ice-based towed ocean profiling buoy control system is provided, wherein the main control module is specifically used for: After the intelligent computing module is powered on, a preset system startup time is waited for the system to start up in order to detect whether the intelligent computing module is ready to start. When the system ready signal of the intelligent computing module is detected, a positioning data frame and a sensor measurement data frame are sent to the intelligent computing module in sequence. After sending the data frames, a measurement end frame is sent to the intelligent computing module to trigger the intelligent computing module to start executing online inference calculation.

[0010] According to the present invention, a smart ice-based towed ocean profiling buoy control system is provided, wherein the main control module is specifically used for: Upon receiving the next sampling interval, a response frame is sent to the intelligent computing module, and a delay waiting operation is performed; After a preset delay, the intelligent computing module is powered off. The data transmission between the main control module and the intelligent computing module adopts a message format that includes a start symbol, a data segment, and a checksum.

[0011] According to the intelligent ice-based towed ocean profiling buoy control system provided by the present invention, the intelligent computing module further includes a hardware configuration unit, the hardware configuration unit being used for: In debug mode, the power supply circuits of non-core peripherals in the intelligent computing module are connected via physical jumper interfaces; Alternatively, in deployment mode, the power supply circuit of the non-core peripheral can be disconnected via the physical jumper interface; The non-core peripherals include at least one of an Ethernet controller and a universal serial bus interface.

[0012] This invention also provides a control method for an intelligent ice-based towed ocean profiling buoy, the method being applied to a main control module, the method comprising: Data is collected according to the currently set sampling interval to obtain marine environmental data; After data acquisition is completed, the intelligent computing module is powered on to trigger its startup. The collected marine environmental data is sent to the intelligent computing module to trigger the intelligent computing module to perform online inference calculations based on the received marine environmental data and locally stored historical sampling data to obtain the next sampling interval; After receiving the next sampling interval from the intelligent computing module, the system controls the intelligent computing module to power off and waits for the next data acquisition according to the next sampling interval.

[0013] This invention also provides a method for controlling an intelligent ice-based towed ocean profiling buoy, the method being applied to an intelligent computing module, the method comprising: In response to the power-on control of the main control module, a startup operation is executed. The main control module controls the intelligent computing module to power on after completing the collection of marine environmental data according to the currently set sampling interval. The system receives marine environmental data sent by the main control module and performs online inference calculations based on the marine environmental data and locally stored historical sampling data to determine the next sampling interval. The next sampling interval is fed back to the main control module, so that after receiving the next sampling interval, the main control module controls the intelligent computing module to power off and waits to execute the next data acquisition according to the next sampling interval.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the intelligent ice-based towed ocean profiling buoy control method as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent ice-based towed ocean profiling buoy control method as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the intelligent ice-based towed ocean profiling buoy control method as described above.

[0017] The intelligent ice-based towed ocean profiling buoy control system, method, and electronic equipment provided by this invention effectively solves the problems of poor flexibility in traditional fixed sampling modes and excessive power consumption after introducing intelligent algorithms by dividing the hardware architecture and coordinating the control of the main control module and the intelligent computing module. Specifically, the system utilizes the intelligent computing module to perform online inference based on currently collected ocean environmental data combined with local historical data. This enables it to accurately perceive the changing trends of the ocean environment, thereby autonomously making decisions and dynamically adjusting the next sampling interval. This breaks the limitations of fixed-period sampling and ensures that dense observations can be obtained to acquire complete data during periods of drastic environmental change. At the same time, the main control module adopts an on-demand power supply strategy to control the intelligent computing module. It only wakes up the intelligent computing module within a short window when inference decisions are needed after data acquisition and cuts off its power after receiving feedback instructions. This intermittent operation mechanism avoids the power waste caused by long-term standby of high-performance computing units, thus balancing the contradiction between the high energy consumption brought by intelligent computing and the low power consumption requirements of long-term operation of ocean buoys. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1This is a schematic diagram of the intelligent ice-based towed ocean profiling buoy control system provided by the present invention; Figure 2 This is a schematic diagram of the process by which the intelligent computing module provided by the present invention performs online inference calculations; Figure 3 This is a schematic diagram of the structure of the intelligent computing module provided by the present invention; Figure 4 This is one of the flowcharts illustrating the intelligent ice-based towed ocean profiling buoy control method provided by the present invention; Figure 5 This is the second flowchart of the intelligent ice-based towed ocean profiling buoy control method provided by the present invention; Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] The accelerated changes in polar sea ice and ice sheets caused by global climate change have led to increasing public attention to the climate-sensitive Arctic and Antarctic regions. Research on polar sea ice changes and the mechanisms of sea-ice-atmosphere interactions typically requires data on multiple parameters of the sea-ice-atmosphere interface, such as temperature, salinity, and depth of the marine environment. Utilizing buoy observation systems to acquire polar ocean hydro-meteorological data is a crucial tool for studying rapid sea ice changes and, consequently, predicting global climate change. Ice-based profiling buoys are instruments specifically designed to acquire subglacial ocean data under polar sea ice conditions and harsh climatic environments. Deploying unmanned sea-ice-atmosphere observation systems in the central polar ocean regions can acquire long-term data on key environmental parameters of the air-ice interface, sea ice interior, and ice-water interface, providing crucial in-situ observational data for scientific questions such as polar sea ice changes.

[0022] Ice-based towed ocean profiling buoy systems are a typical example of such observation systems. These systems enable automatic and continuous measurement of polar marine meteorological and hydrological elements, record sea ice drift trajectories, acquire data in real time, and transmit the data back to Earth via Iridium satellite communication. The system typically consists of two parts: an ice-based meteorological buoy and an underwater profiling buoy, connected by a high-strength towed cable. The ice-based meteorological buoy is deployed on floating ice and drifts with the sea ice, while the underwater profiling buoy periodically performs profiling movements, transmitting profiling measurement data to the ice-based meteorological buoy via the high-strength towed cable during ascent. The ice-based meteorological buoy processes the data, and the collected meteorological and hydrological data, along with GPS positioning data, are transmitted back to Earth via Iridium satellite communication, thus achieving automatic and continuous measurement of polar marine meteorological and hydrological elements.

[0023] Currently, due to considerations of long-term survival in harsh polar environments and low-power operation, the control systems of ice-based meteorological buoys typically employ fixed-time-cycle observation modes. For example, the system is set to initiate a profile observation every 12 hours, with the equipment in standby or hibernation mode at other times. However, the polar marine environment is not always static. When encountering physical processes such as mesoscale eddies or violent fluctuations in thermohaline, the existing fixed long-cycle (e.g., 12-hour) observation mode leads to a severely insufficient data sampling rate, making it impossible to obtain complete process data and resulting in the loss of crucial scientific details.

[0024] To address these shortcomings, this invention provides an intelligent ice-based towed ocean profiling buoy control system and method. It should be noted that all actions involving the acquisition of signals, information, or data in this invention are performed in accordance with the relevant data protection laws and regulations of the country of origin and with authorization from the owner of the corresponding device.

[0025] Figure 1 This is a schematic diagram of the intelligent ice-based towed ocean profiling buoy control system provided by the present invention, as shown below. Figure 1 As shown, the system includes: The main control module 110 is used to collect data according to the currently set sampling interval, and after the collection is completed, control the intelligent computing module 120 to power on and send the collected marine environmental data to the intelligent computing module 120. The intelligent computing module 120 is used to start in response to the power-on control of the main control module 110, receive the marine environment data sent by the main control module 110, perform online inference calculation based on the marine environment data and locally stored historical sampling data to determine the next sampling interval, and feed the next sampling interval back to the main control module 110. The main control module 110 is also used to control the intelligent computing module 120 to power off after receiving the next sampling interval, and wait to execute the next data acquisition according to the next sampling interval.

[0026] Specifically, the control system provided in this embodiment of the invention is mainly integrated into the ice-based towed marine environmental monitoring system (i.e., the ice-based meteorological buoy). This system typically includes a meteorological buoy deployed on the ice floe (as the main control unit) and an underwater profiling buoy connected via a towed cable (as the sensor unit). The system's hardware logic mainly includes a main control module and an intelligent computing module.

[0027] The main control module (also known as the buoy main control board) serves as the system's logic control center, responsible for power management, satellite communication, and basic timing control. The intelligent computing module (also known as the intelligent computing board) acts as the system's high-performance computing unit, responsible for complex data processing and AI inference. The two modules are connected via physical interfaces (such as serial ports and power control lines), together forming the control system described in this embodiment of the invention. This architecture utilizes the low-power characteristics of the main control module to maintain basic system operation, and leverages the high computing power of the intelligent computing module to handle sudden or complex environmental assessment tasks, making it particularly suitable for long-term unattended observation in harsh polar environments.

[0028] Specifically, the main control module is the fundamental control center of the buoy system. It can be built using a low-power microcontroller unit (MCU) and is mainly responsible for maintaining the buoy's daily monitoring, timed wake-up, sensor data acquisition, and satellite communication with the outside world. The intelligent computing module, on the other hand, is an edge computing unit added to compensate for the insufficient computing power of the main control module. It can use a high-performance processor capable of running embedded operating systems (such as Linux) and is specifically designed to run complex artificial neural network models to achieve intelligent recognition of changes in the marine environment.

[0029] The specific workflow and principle of the control system in this embodiment of the invention are as follows: First, the main control module collects data according to the currently set sampling interval, and after the collection is completed, it controls the intelligent computing module to power on and sends the collected marine environmental data to the intelligent computing module. Here, the currently set sampling interval refers to the dormancy time determined when the buoy last performed a task. When the dormancy time ends, the main control module wakes up and controls the underwater profiling buoy to perform vertical profiling movement, collecting marine environmental data such as temperature, salinity, and depth (CTD).

[0030] In the traditional model, the main control module transmits the data via satellite immediately after acquisition. However, in this embodiment of the invention, to determine whether the observation frequency needs adjustment, the main control module connects the power supply to the intelligent computing module via hardware control circuitry (such as closing a relay or controlling the enable pin of a power chip) after acquisition. This on-demand power-on mechanism is key to the system's low power consumption; the intelligent computing module is normally completely powered off and consumes no power, only activating when data needs to be analyzed.

[0031] Secondly, the intelligent computing module will start in response to the power-on control of the main control module. Once the main control module detects that the intelligent computing module is ready, it will send the recently collected marine environmental data to the intelligent computing module via a data communication interface (such as an RS232 serial port). After receiving the marine environmental data from the main control module, the intelligent computing module will perform online inference calculations based on the marine environmental data and locally stored historical sampling data to determine the next sampling interval, and then feed this next sampling interval back to the main control module. This online inference calculation refers to processing the data directly on the buoy using a pre-trained algorithm model, without having to upload large amounts of raw data to a cloud server for processing.

[0032] Specifically, after the intelligent computing module powers on, it first receives current marine environmental data and simultaneously accesses historical sampling data stored in its internal memory. The reason for introducing historical sampling data is that marine physical processes (such as mesoscale eddies and thermohaline changes) typically exhibit temporal continuity, making it difficult to determine environmental trends based on a single data point. The intelligent computing module uses both current and historical data as inputs, feeding them into a pre-set algorithm model (such as a time series prediction model or a classification model). The model then analyzes whether the current hydrological environment has undergone significant abrupt changes relative to its historical state.

[0033] Based on the inference results, the intelligent computing module determines the most suitable next sampling interval for the current environment. For example, if the inference results indicate drastic environmental changes, a shorter sampling interval (e.g., 6 hours) is generated, i.e., denser observation to capture more details; if the environment is stable, a longer sampling interval (e.g., 12 hours) is generated, i.e., normal or sparse observation to save power. After making the decision, the intelligent computing module feeds this specific time value back to the main control module.

[0034] Finally, upon receiving the feedback command, the main control module will cut off the power supply to the intelligent computing module to minimize energy consumption, causing it to stop working. Subsequently, the main control module uses its internal timer or RTC (Real-Time Clock) to set the next wake-up time based on the received next sampling interval, and then enters a low-power sleep mode to wait.

[0035] The system provided in this invention effectively solves the problems of poor flexibility in traditional fixed sampling modes of buoys and excessive power consumption after introducing intelligent algorithms by dividing the hardware architecture and coordinating the control of the main control module and the intelligent computing module. Specifically, the system utilizes the intelligent computing module to perform online inference based on currently collected marine environmental data combined with local historical data. This enables it to accurately perceive the changing trends of the marine environment, thereby autonomously making decisions and dynamically adjusting the next sampling interval. This breaks the limitations of fixed-period sampling and ensures that encrypted observation can be obtained to acquire complete data when the environment changes drastically. At the same time, the main control module adopts an on-demand power supply strategy to control the intelligent computing module. It only wakes up the intelligent computing module within a short window when inference and decision-making are needed after data acquisition, and cuts off its power after receiving feedback instructions. This intermittent operation mechanism avoids the power waste caused by long-term standby of high-performance computing units, thus balancing the contradiction between the high energy consumption brought by intelligent computing and the low power consumption requirements of long-term operation of marine buoys.

[0036] Based on the above embodiments, the intelligent computing module 120 includes: Data storage unit 121 is used to convert the received marine environmental data into a preset format file and then store it locally; The inference calculation unit 122 is used to read the most recent preset number of historical sampling data stored in the data storage unit 121, and input the historical sampling data and the currently received marine environment data into a preset artificial neural network model for online inference calculation to obtain the current marine environment change probability value output by the artificial neural network model; The decision feedback unit 123 is used to compare the current probability value of marine environmental change with a preset probability threshold, determine the next sampling interval based on the comparison result, and feed the next sampling interval back to the main control module 110 through a communication interface.

[0037] Specifically, the intelligent computing module logically includes a data storage unit, an inference computing unit, and a decision feedback unit. Figure 2 This is a schematic diagram of the process by which the intelligent computing module provided by the present invention performs online inference calculations, as shown below. Figure 2 As shown, once the intelligent computing module is powered on, it begins receiving marine environmental data sent by the active module, such as raw binary or ASCII code streams containing information like depth, temperature, and salinity. Upon receiving the marine environmental data, the data storage unit takes over first.

[0038] The data storage unit converts the received marine environmental data into a preset format file for local storage. In this embodiment of the invention, the preset format file is preferably a CSV (Comma-Separated Values) file because this format is highly versatile and facilitates subsequent algorithm reading and analysis. The data storage unit parses the raw sensor data and organizes it by column (e.g., depth, temperature, salinity) and saves it to an onboard storage medium (e.g., an industrial-grade SD card or eMMC chip). To facilitate time series analysis, each file typically corresponds to a complete profile observation and is named with a timestamp. This step is not only for current calculations but also for establishing long-term local data as a source of historical sampling data.

[0039] Next, the inference computing unit reads a preset number of recent historical sampled data stored in the data storage unit, and inputs this historical sampled data and the currently received marine environmental data into a preset artificial neural network model for online inference computing to obtain the probability value of current marine environmental changes. Here, the inference computing unit is the core of intelligent computing. To capture the dynamic trends of marine environmental changes, data at the current moment is often insufficient. Therefore, the inference computing unit uses a sliding window approach to load the N most recent historical files (i.e., a preset number, such as N=16 or N=32) from the storage medium, which, together with the currently received and converted data, form a time series input matrix.

[0040] The input matrix is ​​fed into a pre-trained artificial neural network model deployed locally. This model, trained on extensive polar and marine data, is capable of identifying specific physical processes (characterizing whether the marine environment has changed). The model's output is not a simple yes or no, but a value between 0 and 1, representing the probability of a change in the current marine environment. A higher value indicates that the model believes there is a greater likelihood of a significant change in the underwater environment or the presence of a physical phenomenon of interest.

[0041] Finally, the decision feedback unit compares the current probability value of changes in the marine environment with a preset probability threshold, determines the next sampling interval based on the comparison result, and feeds back the next sampling interval to the main control module through the communication interface. Specifically, this unit is responsible for converting the AI's inference results into specific control commands. The system has a preset probability threshold (e.g., TH=0.8). If the probability value output by the model is greater than this threshold, the decision feedback unit judges that the environment has changed significantly and more intensive data needs to be obtained, so it sets the next sampling interval to a shorter time (e.g., 6 hours); conversely, if the probability value is less than or equal to the threshold, it considers the environment to be relatively stable, and in order to save energy, the next sampling interval is maintained at a longer, normal time (e.g., 12 hours). After determining the time parameters, the unit encapsulates them into a command frame of a specific protocol through communication interfaces such as RS232 and sends it back to the main control module.

[0042] In this embodiment of the invention, by decoupling the intelligent computing task into three independent units—storage, inference, and decision-making—standardized data flow processing is achieved. In particular, the introduction of a joint input mechanism for historical and current data enables the buoy to make judgments based on the evolutionary trend over time, rather than solely on threshold triggers from single-point data. This significantly improves the accuracy and robustness of environmental identification and reduces misjudgments caused by sensor noise.

[0043] Based on any of the above embodiments, the artificial neural network model is a time series mixture model, and the artificial neural network model is configured to support a format that supports quantization acceleration; The intelligent computing module 120 is also used to adjust the input and output dimensions of the artificial neural network model, the preset probability threshold, and the preset quantity through a configuration file.

[0044] Specifically, the time-series mixing model used in this embodiment of the invention can be a TSMixer (Time-Series Mixer) architecture. This is a full MLP (Multi-Layer Perceptron) architecture model. Compared with complex Transformer or RNN (Recurrent Neural Network) models, TSMixer maintains the ability to capture long-series time dependencies while having lower computational complexity and faster inference speed, making it very suitable for embedded edge deployment.

[0045] To further improve inference efficiency, the model is not run directly in its original floating-point format, but rather undergoes model quantization. For example, the trained model is first exported to the standard ONNX (Open Neural Network Exchange) format, and then further converted to a dedicated format (such as NBG format) that supports NPU (Neural Processing Unit) hardware acceleration using a conversion tool. This quantization-accelerated format (such as Int8 quantization) can significantly reduce the memory space occupied by the model and greatly reduce power consumption during inference, achieving inference response times in the order of seconds or even milliseconds.

[0046] Furthermore, to adapt to the needs of different sea areas and scientific missions, a configuration file (such as YAML or JSON format) is pre-installed in the intelligent computing module's file system. Users can define key parameters by modifying this configuration file without modifying the source code or recompiling the program. For example, regarding input / output dimensions, if the sensor is changed, the dimension parameter can be modified to adapt to the new input data structure. Regarding preset probability thresholds, if the mission focuses more on capturing extreme events, the threshold can be lowered; if more emphasis is placed on endurance, the threshold can be raised. In addition, the preset quantity, namely the number of historical sampling data N mentioned above, can be adjusted by the user according to the rate of sea state changes. The intelligent computing module's software automatically reads this configuration file upon startup and dynamically loads the corresponding model parameters and operating strategies.

[0047] This invention, through the use of a lightweight time-series mixture model combined with quantization acceleration technology, successfully enables the real-time execution of complex AI algorithms on low-power embedded hardware, solving the edge computing bottleneck problem. Simultaneously, the introduction of configuration files endows the system with extremely high flexibility and scalability, allowing the same control system to adapt to the observation needs of different sea areas and different sensors, greatly reducing equipment maintenance and upgrade costs.

[0048] Based on any of the above embodiments, the decision feedback unit 123 is further configured to: If the probability value of the current marine environment change is greater than the preset probability threshold, it is determined that the current underwater environment has changed, and the next sampling interval is set to the first preset duration corresponding to the encrypted observation mode; If the current probability value of marine environmental change is less than or equal to the preset probability threshold, it is determined that the current underwater environment has not changed, and the next sampling interval is set to the second preset duration corresponding to the normal observation mode; Wherein, the first preset duration is less than the second preset duration.

[0049] Specifically, in actual ocean observation missions, balancing data density and equipment power consumption is the core challenge. Therefore, this system pre-sets two basic operating modes: encrypted observation mode and normal observation mode.

[0050] In the specific operational logic of the decision feedback unit, when the probability value of the current marine environment change output by the inference calculation unit is greater than a preset probability threshold (e.g., TH=0.8), the system determines that the underwater environment where the buoy is currently located is undergoing a significant physical process (such as frontal passage, eddy activity, etc.). In order to fully capture the evolution details of this process, the decision feedback unit determines that the current underwater environment has changed and sets the next sampling interval for feedback to the main control module to a first preset duration. For example, this first preset duration can preferably be 6 hours. This means that the buoy will enter a high-frequency, intensive observation state.

[0051] Conversely, if the probability value of current marine environmental change is less than or equal to a preset probability threshold, the system determines that the current underwater environment is relatively stable or does not change significantly. In this case, there is no need to waste valuable electrical energy on high-frequency sampling. The decision feedback unit determines that the current underwater environment has not changed and sets the next sampling interval to a second preset duration. This second preset duration is preferably 12 hours, meaning the buoy returns to a standard low-power cruise observation state.

[0052] Clearly, the first preset duration is shorter than the second preset duration. Through this binary switching mechanism based on AI inference results, this embodiment of the invention breaks the traditional fixed-time sampling rules of buoys. This not only ensures the acquisition of high-temporal-resolution scientific data when special ocean phenomena occur, avoiding the omission of key physical processes, but also significantly saves power by extending the dormancy time during most stable periods, thereby extending the overall operating life of the buoy.

[0053] Based on any of the above embodiments, in polar environments, the drift speed of sea ice varies greatly. If the buoy moves rapidly with the ice, even if the hydrological environment does not change much over time, the buoy will traverse a large geographical distance in a short period of time. To ensure the spatial resolution of the observation data, more frequent sampling is required. Conversely, if the sea ice stops drifting (e.g., due to ice jams), the buoy will remain almost stationary. In this case, frequent sampling will only yield duplicate data, resulting in waste.

[0054] To this end, the main control module is also configured to simultaneously send its current geographic location coordinates when transmitting marine environmental data to the intelligent computing module. This coordinate information is typically obtained by the GPS positioning module integrated into the main control module when the buoy transmits data.

[0055] After receiving the coordinate information, the decision feedback unit executes the following logic: First, it calculates the buoy's current drift speed based on the geographical coordinates received multiple times. Specifically, this can be obtained by dividing the spherical distance between two adjacent sampling points by the time interval. Next, the calculated speed is compared with a preset threshold. If the current drift speed exceeds the preset threshold, it indicates a rapid drift state. In this case, the system believes that not only should time changes be considered, but also spatial sampling density should be maintained. Therefore, regardless of whether the AI ​​model determines that the environment has changed, the decision feedback unit will adjust the next sampling interval to the first preset duration (i.e., denser observation) to prioritize spatial sampling resolution and prevent the omission of spatial hydrological features along the route due to sparse sampling.

[0056] If the current drift velocity is below the preset static threshold, it indicates that the system is in a near-static state, and the system considers it pointless to continue sampling at the original frequency. Therefore, the decision feedback unit will maintain the next sampling interval at the second preset duration or adjust it to a longer sleep duration (e.g., extended to 24 hours) to avoid invalid repeated observations.

[0057] The system provided in this invention extends intelligence from the time dimension to the spatial dimension. By introducing drift velocity as a correction factor, it solves the problem of traditional control methods ignoring the buoy's motion state, ensuring the uniformity of the distribution of observation data in geographic space, and effectively eliminating redundant data in a stationary state, further improving observation efficiency.

[0058] Based on any of the above embodiments, during the polar night, photovoltaic power supply is insufficient, and battery power is crucial in determining the buoy's operating time. If the system only considers the value of scientific observation while ignoring the cost of electricity, the buoy may run out of power early in the mission. Therefore, this embodiment of the invention constructs a dynamic game mechanism based on the observation value and observation cost.

[0059] Specifically, the main control module collects the current remaining battery power data and sends it to the intelligent computing module. Upon receiving the battery power data, the decision feedback unit performs the following calculations and judgments: First, the current observation cost coefficient is calculated based on the remaining battery power data. A cost function is defined here, with the logic that the lower the remaining battery power, the higher the observation cost coefficient. For example, the cost coefficient C = k / current_voltage, where k is a constant and current_voltage represents the current remaining battery power. This means that when the battery power is sufficient, the cost of making an observation is very small; however, when the battery power is critically low, the cost of making an observation becomes extremely high.

[0060] Next, an observational value coefficient is generated based on the current probability value of changes in the marine environment. A value function is also defined here, typically using the probability value output by the AI ​​model directly as the scientific value V of the observation. A higher probability value indicates more drastic environmental changes and greater scientific value of the observation.

[0061] Subsequently, the system calculates the ratio of the observation value coefficient to the observation cost coefficient (i.e., Ratio = V / C). Only when the ratio is greater than the preset start threshold is the current encrypted observation considered worthwhile, and the next sampling interval is determined as the first preset duration corresponding to the encrypted observation; otherwise, even if the environment changes to some extent, if the power is insufficient and the change is not extreme enough (i.e., the cost-effectiveness is not high), the system still chooses to maintain the second preset duration (normal observation or low power mode).

[0062] In this embodiment of the invention, when the power supply is sufficient, the system responds sensitively to even minor environmental changes, at a low cost and with thresholds easily met. Conversely, when the power supply is low, the system only performs intensive observations on environmental changes of significant scientific value. This strategy ensures the acquisition of the most critical scientific data while minimizing the risk of the buoy running out of power prematurely due to oversampling, achieving a balance between scientific objectives and equipment operation.

[0063] Based on any of the above embodiments, the main control module 110 is specifically used for: After the intelligent computing module 120 is powered on, a preset system startup time is waited for the system to start up in order to detect whether the intelligent computing module 120 is ready to start up. When the system ready signal of the intelligent computing module 120 is detected, a positioning data frame and a sensor measurement data frame are sent to the intelligent computing module 120 in sequence, and a measurement end frame is sent to the intelligent computing module 120 after the data frames are sent, so as to trigger the intelligent computing module 120 to start executing online inference calculation.

[0064] It should be noted that in the intelligent ice-based towed ocean profiling buoy control system, the main control module (based on an MCU) and the intelligent computing module (based on a high-performance processor running Linux) are heterogeneous systems, with significant differences in their startup speeds. Therefore, precise timing control is a prerequisite for stable system operation.

[0065] Specifically, after the main control module closes the power supply circuit, the intelligent computing module begins loading the bootloader and operating system kernel. Since this process takes some time, the main control module cannot immediately send data. The main control module starts a timer to wait; this preset system startup time is set according to the hardware characteristics of the intelligent computing module, for example, approximately 12 seconds. During this period, the main control module continuously monitors status feedback signals from the intelligent computing module (e.g., by monitoring changes in the level of specific GPIO pins, or waiting for the system status indicator LED on the intelligent computing module to light up) to confirm that the operating system has been fully loaded and the application has been initialized.

[0066] Upon detecting a system ready signal from the intelligent computing module (e.g., detecting a system status indicator LED on the intelligent computing board illuminates), the data transmission phase begins. To ensure the logical clarity of the data packets, the main control module strictly follows the data transmission sequence: first, it sends a positioning data frame (containing GPS latitude and longitude, timestamp, etc.) to provide a spatiotemporal label for this observation; then, it sends a sensor measurement data frame (containing CTD profile data such as depth, temperature, and salinity), which is the core input for AI inference; after all data has been sent, the main control module sends a specific measurement end frame. This measurement end frame acts as a trigger. Before receiving this frame, the intelligent computing module only receives and buffers the data; once it receives this frame, it signifies the end of the transmission, and the intelligent computing module immediately invokes its internal algorithm to begin online inference calculations on the buffered data.

[0067] This invention addresses the communication packet loss problem caused by asynchronous startup between heterogeneous processors by using a preset startup waiting and ready detection mechanism. At the same time, the control logic of triggering inference with the end frame clarifies the time boundaries of data transmission and data processing, enabling efficient collaboration between the two boards and avoiding the error of the intelligent computing module preemptively calculating before the data is fully received.

[0068] Based on any of the above embodiments, the main control module 110 is specifically used for: Upon receiving the next sampling interval, a response frame is sent to the intelligent computing module 120, and a delay waiting operation is performed; After a preset delay period, the intelligent computing module 120 is powered off. The data transmission between the main control module 110 and the intelligent computing module 120 adopts a message format including a start symbol, a data segment, and a checksum.

[0069] Specifically, after receiving an instruction from the intelligent computing module (such as setting the next sampling interval to 6 hours), the main control module does not immediately cut off the power. Instead, it first replies with a response frame (ACK) to inform the intelligent computing module that the instruction has been received. Subsequently, the main control module enters a delayed waiting state, with the preset time set, for example, to 3 seconds.

[0070] The purpose of this delay is to give the intelligent computing module sufficient time to perform file system synchronization operations, ensuring that log files and data files stored on the SD card have been safely written to the physical media, preventing file system corruption due to sudden power outages. After the delay ends, the main control module physically cuts off the power to the intelligent computing module, completing this work cycle.

[0071] Furthermore, to ensure the accuracy of data transmission in harsh marine environments, the data transmission between the main control module and the intelligent computing module adopts a message format including a start character, data segments, and a checksum. In specific implementations, the physical layer interface can adopt the 3-wire RS-232C standard. The protocol layer specifies that: for the start character, all command and response frames begin with a specific character (such as $) as a frame synchronization flag; the data segment refers to the specific instructions or values, transmitted using ASCII code, with fields separated by delimiters (such as half-width English commas ",") to enhance readability; the checksum is located at the end of the frame, for example, using a two-character checksum. Its calculation method can be to calculate the sum of the ASCII codes of all characters between the start character and the checksum, and then convert the low byte to hexadecimal representation.

[0072] The receiver considers the data valid only after verification; otherwise, it will request a retransmission or discard the data, thus avoiding control errors caused by line interference.

[0073] The embodiments of this invention construct a complete closed-loop control flow. In particular, the introduced response, delay, and power-off timing effectively protect the file system security of the intelligent computing module and solve the pain point that embedded Linux systems are prone to damage to storage devices under unexpected power failures. The text protocol design based on checksums improves the communication reliability of the system in environments with strong electromagnetic interference or poor contact.

[0074] Based on any of the above embodiments, the intelligent computing module 120 further includes a hardware configuration unit, the hardware configuration unit being used for: In debug mode, the power supply circuits of non-core peripherals in the intelligent computing module are connected via physical jumper interfaces; Alternatively, in deployment mode, the power supply circuit of the non-core peripheral can be disconnected via the physical jumper interface; The non-core peripherals include at least one of an Ethernet controller and a universal serial bus interface.

[0075] It should be noted that in the circuit design of intelligent computing modules, in addition to the core processor, memory, and serial port circuit for communication with the main controller, a large number of peripheral interfaces are usually integrated for the convenience of developers. However, the control chips associated with these interfaces (such as PHY chips and Hub chips) consume a considerable amount of quiescent current even in idle state, which is unacceptable for ice-based buoys where energy is extremely precious.

[0076] Therefore, the intelligent computing module also includes a hardware configuration unit. In practice, this hardware configuration unit is not a complex logic circuit, but rather uses a simple and reliable physical connection structure, such as onboard pin headers and jumper caps, or DIP switches.

[0077] When researchers are developing algorithms, porting systems, or exporting data from recovered buoys in a laboratory environment, the equipment is in debug mode. At this time, the operator inserts a jumper cap into the physical jumper interface, thus closing the power supply circuit. The power management chip then begins supplying power to non-core peripherals.

[0078] Here, non-core peripherals may include Ethernet controllers, Universal Serial Bus interfaces, etc. For example, after the Ethernet PHY chip is powered on, R&D personnel can remotely log in to the Linux system of the intelligent computing module via SSH (Secure Shell) through a standard RJ45 network port to perform efficient code uploads, log viewing, and model updates; or they can connect a keyboard and mouse via USB for local debugging.

[0079] When the buoy is about to be deployed to the polar ice surface and enter its unattended operational state (i.e., deployment mode), these debugging interfaces will no longer be used. At this point, the operator can manually remove the jumper caps. This operation completely cuts off the power path to non-core peripherals such as the Ethernet PHY chip and USB hub chip at the physical level.

[0080] This invention employs physical isolation for power management. Compared to simply disabling peripheral drivers via software, physical power-off is more thorough, eliminating static leakage current in the chip. In extremely low-temperature environments, this low-power design conserves valuable power, ensuring that every ounce of energy is used for core AI inference and data acquisition, thereby significantly enhancing the buoy system's survivability in the wild.

[0081] Based on any of the above embodiments Figure 3 This is a schematic diagram of the structure of the intelligent computing module provided by the present invention, as shown below. Figure 3As shown, the intelligent computing module (also known as the intelligent computing board) adopts a highly integrated embedded circuit board design. Its core components and interface layout are as follows: A high-performance main processor is located at the center of the board; in this embodiment, the STM32MP257 chip is preferred. This processor integrates an ARM Cortex-A7 application processor core, sufficient to run the Linux operating system and a lightweight neural network inference framework. A metal shield is mounted on the surface of the main processor to shield against electromagnetic interference, ensuring operational stability in the complex marine electromagnetic environment.

[0082] The left side of the board features industrial-grade terminal blocks, namely the buoy main control board interface, including 12V_IN, TX, RX, and GND. 12V_IN and GND are used to receive controlled power from the buoy main control module. Voltage input to this interface only occurs when the main control module closes its power supply relay, at which point the intelligent computing module starts. TX and RX are RS232 level communication lines used to receive marine environmental data (CTD data, GPS data) sent by the main control module and to provide feedback on the next sampling interval command. A test serial port is also located at the lower left of the board for low-level debugging during the development phase.

[0083] A system SD card is located at the bottom of the board. The SD card is logically divided into multiple partitions, used to store the Linux operating system image and bootloader, pre-installed artificial neural network model files and configuration files, historical sampling data CSV files generated during operation, and system logs. The board also features a boot mode switch, allowing users to select between booting from the SD card, USB, or Flash memory for easier system maintenance.

[0084] To facilitate intuitive assessment of the system's operating status, the board is equipped with multiple sets of LED indicators. For example, the system status LED can be set to green, indicating that the Linux system kernel has finished loading and entered a ready state; the inference status LED can be set to blue or red, flashing or remaining constantly lit when the main control module sends a measurement completion frame to trigger inference, until the inference ends and power is cut off. This provides debugging personnel with intuitive visual feedback.

[0085] The board has multiple low-power physical jumpers distributed near the Ethernet port, USB interface, and debug interface. The debug port (RJ45), system programming interface, and debug interface are high-power non-core peripherals. Even without data transmission, the Ethernet PHY chip consumes tens to hundreds of milliamps of current when operating. During laboratory debugging (debugging mode), technicians short-circuit jumper caps to the jumper pins, connecting the 3.3V / 5V power supply circuits of the PHY chip and USB controller. SSH login or file transfer can then be performed via Ethernet cable. Before field deployment (deployment mode), technicians can remove these jumper caps. After physically disconnecting the power supply circuits, the Ethernet and USB circuits consume no power, thus minimizing the board's standby power consumption and achieving ultimate energy saving at the physical level.

[0086] In addition, the board is equipped with a reset switch for manually resetting the system in case of a debugging crash. Through the above hardware layout, this intelligent computing module achieves a combination of high-performance computing capabilities and extremely low power consumption requirements at the physical level, ensuring the effective implementation of the control method on actual physical devices.

[0087] The intelligent ice-based towed ocean profiling buoy control method provided by the present invention will be described below. The intelligent ice-based towed ocean profiling buoy control method described below can be referred to in correspondence with the intelligent ice-based towed ocean profiling buoy control system described above.

[0088] Based on any of the above embodiments Figure 4 This is one of the flowcharts illustrating the intelligent ice-based towed ocean profiling buoy control method provided by the present invention, such as... Figure 4 As shown, this method is applied to the main control module, and the method includes: Step 410: Collect data according to the currently set sampling interval to obtain marine environmental data; Step 420: After data acquisition is completed, control the intelligent computing module to power on, so as to trigger the intelligent computing module to start. Step 430: Send the collected marine environment data to the intelligent computing module to trigger the intelligent computing module to perform online inference calculation based on the received marine environment data and locally stored historical sampling data to obtain the next sampling interval; Step 440: After receiving the next sampling interval from the intelligent computing module, control the intelligent computing module to power off and wait for the next data acquisition according to the next sampling interval.

[0089] The method provided in this invention effectively solves the problems of poor flexibility in traditional fixed sampling modes of buoys and excessive power consumption after introducing intelligent algorithms by dividing the hardware architecture and coordinating the control of the main control module and the intelligent computing module. Specifically, the system utilizes the intelligent computing module to perform online inference based on currently collected marine environmental data combined with local historical data. This enables it to accurately perceive the changing trends of the marine environment, thereby autonomously making decisions and dynamically adjusting the next sampling interval. This breaks the limitations of fixed-period sampling and ensures that encrypted observation can be obtained to acquire complete data when the environment changes drastically. At the same time, the main control module adopts an on-demand power supply strategy to control the intelligent computing module. It only wakes up the intelligent computing module within a short window when inference and decision-making are needed after data acquisition, and immediately cuts off its power supply after receiving feedback instructions. This intermittent operation mechanism avoids the power waste caused by long-term standby of high-performance computing units and balances the contradiction between the high energy consumption brought by intelligent computing and the low power consumption requirements for the long-term survival of marine buoys.

[0090] Based on any of the above embodiments Figure 5 This is the second flowchart illustrating the intelligent ice-based towed ocean profiling buoy control method provided by this invention, as shown below. Figure 5 As shown, this method is applied to an intelligent computing module, and the method includes: Step 510: In response to the power-on control of the main control module, a startup operation is performed. The main control module controls the intelligent computing module to power on after completing the marine environmental data collection according to the currently set sampling interval. Step 520: Receive marine environmental data sent by the main control module, and perform online inference calculations based on the marine environmental data and locally stored historical sampling data to determine the next sampling interval; Step 530: The next sampling interval is fed back to the main control module, so that after receiving the next sampling interval, the main control module controls the intelligent computing module to power off and waits to execute the next data acquisition according to the next sampling interval.

[0091] The method provided in this invention effectively solves the problems of poor flexibility in traditional fixed sampling modes of buoys and excessive power consumption after introducing intelligent algorithms by dividing the hardware architecture and coordinating the control of the main control module and the intelligent computing module. Specifically, the system utilizes the intelligent computing module to perform online inference based on currently collected marine environmental data combined with local historical data. This enables it to accurately perceive the changing trends of the marine environment, thereby autonomously making decisions and dynamically adjusting the next sampling interval. This breaks the limitations of fixed-period sampling and ensures that encrypted observation can be obtained to acquire complete data when the environment changes drastically. At the same time, the main control module adopts an on-demand power supply strategy to control the intelligent computing module. It only wakes up the intelligent computing module within a short window when inference and decision-making are needed after data acquisition, and immediately cuts off its power supply after receiving feedback instructions. This intermittent operation mechanism avoids the power waste caused by long-term standby of high-performance computing units and balances the contradiction between the high energy consumption brought by intelligent computing and the low power consumption requirements for the long-term survival of marine buoys.

[0092] It should be noted that the various method embodiments of the present invention applied to the main control module or intelligent computing module can refer to the above system embodiments, and will not be repeated here.

[0093] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 610, a communication interface 620, a memory 630, and a communication bus 640. The processor 610, communication interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a smart ice-based towed ocean profiling buoy control method. This method is applied to the main control module and includes: acquiring ocean environmental data according to a currently set sampling interval; powering on the smart computing module after data acquisition to trigger its startup; sending the acquired ocean environmental data to the smart computing module to trigger it to perform online inference calculations based on the received ocean environmental data and locally stored historical sampling data to obtain the next sampling interval; and powering off the smart computing module after receiving the next sampling interval from the smart computing module and waiting for the next data acquisition to proceed according to the next sampling interval.

[0094] The processor 610 can also call logic instructions in the memory 630 to execute a smart ice-based towed ocean profiling buoy control method. This method is applied to the intelligent computing module and includes: responding to the power-on control of the main control module to perform a startup operation, wherein the main control module controls the intelligent computing module to power on after completing the acquisition of ocean environmental data according to the currently set sampling interval; receiving ocean environmental data sent by the main control module and performing online inference calculations based on the ocean environmental data and locally stored historical sampling data to determine the next sampling interval; and feeding back the next sampling interval to the main control module so that the main control module controls the intelligent computing module to power off after receiving the next sampling interval and waits to execute the next data acquisition according to the next sampling interval.

[0095] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to related technologies, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the intelligent ice-based towed ocean profiling buoy control method provided by the above methods. This method is applied to the main control module and includes: acquiring ocean environmental data according to a currently set sampling interval; controlling the intelligent computing module to power on after the data acquisition is completed to trigger the intelligent computing module to start; sending the acquired ocean environmental data to the intelligent computing module to trigger the intelligent computing module to perform online inference calculation based on the received ocean environmental data and locally stored historical sampling data to obtain the next sampling interval; and controlling the intelligent computing module to power off after receiving the next sampling interval from the intelligent computing module, and waiting to execute the next data acquisition according to the next sampling interval.

[0097] When the computer program is executed by the processor, the computer can also execute the intelligent ice-based towed ocean profiling buoy control method provided by the above methods. This method is applied to the intelligent computing module and includes: responding to the power-on control of the main control module to perform a startup operation, wherein the main control module controls the intelligent computing module to power on after completing the acquisition of ocean environmental data according to the currently set sampling interval; receiving the ocean environmental data sent by the main control module, and performing online inference calculation based on the ocean environmental data and locally stored historical sampling data to determine the next sampling interval; feeding back the next sampling interval to the main control module, so that after receiving the next sampling interval, the main control module controls the intelligent computing module to power off and waits to execute the next data acquisition according to the next sampling interval.

[0098] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the intelligent ice-based towed ocean profiling buoy control method provided by the methods described above. This method is applied to a main control module and includes: acquiring ocean environmental data according to a currently set sampling interval; powering on an intelligent computing module after data acquisition to trigger the intelligent computing module to start; sending the acquired ocean environmental data to the intelligent computing module to trigger the intelligent computing module to perform online inference calculations based on the received ocean environmental data and locally stored historical sampling data to obtain the next sampling interval; and, upon receiving the next sampling interval from the intelligent computing module, powering off the intelligent computing module and waiting to execute the next data acquisition according to the next sampling interval.

[0099] When executed by a processor, the computer program implements the intelligent ice-based towed ocean profiling buoy control method provided by the methods described above. This method is applied to an intelligent computing module and includes: responding to a power-on control of a main control module by performing a startup operation, wherein the main control module powers on the intelligent computing module after completing ocean environmental data acquisition according to a currently set sampling interval; receiving ocean environmental data sent by the main control module and performing online inference calculations based on the ocean environmental data and locally stored historical sampling data to determine the next sampling interval; and feeding back the next sampling interval to the main control module, so that upon receiving the next sampling interval, the main control module controls the intelligent computing module to power off and waits to execute the next data acquisition according to the next sampling interval.

[0100] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0101] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of software products. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart ice-based towed ocean profiling buoy control system, characterized in that, include: The main control module is used to collect data according to the currently set sampling interval, and after the collection is completed, it controls the intelligent computing module to power on and sends the collected marine environmental data to the intelligent computing module. The intelligent computing module is used to start in response to the power-on control of the main control module, receive the marine environment data sent by the main control module, perform online inference calculation based on the marine environment data and locally stored historical sampling data to determine the next sampling interval, and feed the next sampling interval back to the main control module; The main control module is also used to control the intelligent computing module to power off after receiving the next sampling interval, and wait to execute the next data acquisition according to the next sampling interval.

2. The intelligent ice-based towed ocean profiling buoy control system according to claim 1, characterized in that, The intelligent computing module includes: The data storage unit is used to convert the received marine environmental data into a preset format file and then store it locally. The inference calculation unit is used to read a preset number of historical sampling data stored in the data storage unit, and input the historical sampling data and the currently received marine environment data into a preset artificial neural network model for online inference calculation to obtain the current marine environment change probability value output by the artificial neural network model; The decision feedback unit is used to compare the current probability value of marine environmental change with a preset probability threshold, determine the next sampling interval based on the comparison result, and feed the next sampling interval back to the main control module through a communication interface.

3. The intelligent ice-based towed ocean profiling buoy control system according to claim 2, characterized in that, The decision feedback unit is also used for: If the probability value of the current marine environment change is greater than the preset probability threshold, it is determined that the current underwater environment has changed, and the next sampling interval is set to the first preset duration corresponding to the encrypted observation mode; If the current probability value of marine environmental change is less than or equal to the preset probability threshold, it is determined that the current underwater environment has not changed, and the next sampling interval is set to the second preset duration corresponding to the normal observation mode; Wherein, the first preset duration is less than the second preset duration.

4. The intelligent ice-based towed ocean profiling buoy control system according to claim 2, characterized in that, The artificial neural network model is a time series mixture model, and the artificial neural network model is configured to support quantization acceleration. The intelligent computing module is also used to adjust the input and output dimensions of the artificial neural network model, the preset probability threshold, and the preset quantity through a configuration file.

5. The intelligent ice-based towed ocean profiling buoy control system according to claim 1, characterized in that, The main control module is specifically used for: After the intelligent computing module is powered on, a preset system startup time is waited for the system to start up in order to detect whether the intelligent computing module is ready to start. When the system ready signal of the intelligent computing module is detected, a positioning data frame and a sensor measurement data frame are sent to the intelligent computing module in sequence. After sending the data frames, a measurement end frame is sent to the intelligent computing module to trigger the intelligent computing module to start executing online inference calculation.

6. The intelligent ice-based towed ocean profiling buoy control system according to claim 5, characterized in that, The main control module is specifically used for: Upon receiving the next sampling interval, a response frame is sent to the intelligent computing module, and a delay waiting operation is performed; After a preset delay, the intelligent computing module is powered off. The data transmission between the main control module and the intelligent computing module adopts a message format that includes a start symbol, a data segment, and a checksum.

7. The intelligent ice-based towed ocean profiling buoy control system according to any one of claims 1 to 6, characterized in that, The intelligent computing module further includes a hardware configuration unit, which is used for: In debug mode, the power supply circuits of non-core peripherals in the intelligent computing module are connected via physical jumper interfaces; Alternatively, in deployment mode, the power supply circuit of the non-core peripheral can be disconnected via the physical jumper interface; The non-core peripherals include at least one of an Ethernet controller and a universal serial bus interface.

8. A method for controlling an intelligent ice-based towed ocean profiling buoy, characterized in that, The method is applied to the main control module, and the method includes: Data is collected according to the currently set sampling interval to obtain marine environmental data; After data acquisition is completed, the intelligent computing module is powered on to trigger its startup. The collected marine environmental data is sent to the intelligent computing module to trigger the intelligent computing module to perform online inference calculations based on the received marine environmental data and locally stored historical sampling data to obtain the next sampling interval; After receiving the next sampling interval from the intelligent computing module, the system controls the intelligent computing module to power off and waits for the next data acquisition according to the next sampling interval.

9. A method for controlling an intelligent ice-based towed ocean profiling buoy, characterized in that, The method is applied to an intelligent computing module, and the method includes: In response to the power-on control of the main control module, a startup operation is executed. The main control module controls the intelligent computing module to power on after completing the collection of marine environmental data according to the currently set sampling interval. The system receives marine environmental data sent by the main control module and performs online inference calculations based on the marine environmental data and locally stored historical sampling data to determine the next sampling interval. The next sampling interval is fed back to the main control module, so that after receiving the next sampling interval, the main control module controls the intelligent computing module to power off and waits to execute the next data acquisition according to the next sampling interval.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the intelligent ice-based towed ocean profiling buoy control method as described in claim 8 or 9.

Images