Data analysis method, system and equipment of deep sea observation network system and medium

Abstract

The embodiment of the invention discloses a data analysis method, system and equipment of a deep sea observation network system and a medium, and relates to the technical field of data processing. The method comprises the following steps: an edge node receives original data sent by a sensor, converts the original data into initial feature data, and transmits the initial feature data to a cloud platform; the cloud platform performs time alignment on the initial feature data transmitted by all the edge nodes to obtain aligned data; the cloud platform performs feature fusion on the alignment data of all the edge nodes to obtain fused feature data; and the cloud platform carries out seabed disaster early warning based on the fused feature data. According to the method, data redundancy is reduced through edge node feature conversion, time sequence deviation is eliminated through cloud time alignment, complementary information is mined through multi-source feature fusion, misinformation is inhibited by using a cross validation mechanism, and signal sensitivity is improved through a cooperative enhancement effect, so that disaster early warning accuracy is systematically improved, and reliable technical support is provided for ocean disaster prevention.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a data analysis method, system, equipment and medium for a deep-sea observation network system. Background Technology

[0002] The ocean is rich in biological and mineral resources, holding irreplaceable strategic significance for human survival and development. However, the deep-sea environment is complex and full of unknowns. Sudden disasters such as submarine earthquakes not only seriously threaten the stability of marine ecosystems and the operational safety of marine engineering projects, but may also trigger secondary disasters (such as tsunamis), causing enormous damage to coastal areas. Therefore, achieving early and accurate warnings of deep-sea disasters is a core challenge and urgent need in the field of marine environmental monitoring and protection.

[0003] Effective disaster early warning mechanisms rely heavily on comprehensive, timely, and accurate perception and analysis of the marine environment. Thanks to advancements in modern marine observation technology, various sensors deployed on the seabed can continuously collect diverse physical, chemical, and biological parameters, forming a multi-source heterogeneous data stream. This data primarily includes temperature data reflecting the thermodynamic properties of water, salinity data indicating changes in seawater chemical composition, acoustic data capturing the propagation characteristics of underwater sound waves, and image data providing intuitive images of seabed topography and activity. Each data source reveals the local state of the marine environment from a specific dimension, theoretically containing crucial information related to the formation and occurrence of disasters.

[0004] However, existing deep-sea disaster early warning technology systems face significant bottlenecks. The core issue lies in the lack of efficient and robust capabilities for deep fusion and collaborative analysis of multi-source heterogeneous data. Specifically: First, data from different sources (such as physical quantities, acoustic signals, and visual information) differ greatly in format, scale, accuracy, and physical meaning, making effective correlation and calibration within a unified framework difficult. Second, seabed observation networks are constrained by complex physical environments, and data acquisition and transmission often involve delays and uncertainties, making it difficult to meet the stringent real-time requirements of early warning. More importantly, current mainstream methods tend to focus on the analysis of single or a few types of data, failing to fully explore and utilize the potential complementarity and synergistic effects between multiple data sources. For example, temperature anomalies may be related to hydrothermal activity, acoustic signals can capture vibrations from crustal ruptures, and image changes can visually reflect seabed topographical shifts, but these independent clues have not been systematically integrated and verified.

[0005] This lack of multi-source information fusion capability directly leads to insufficient sensitivity and a high false alarm rate in identifying disaster precursor signals. Early warning models struggle to accurately extract weak but crucial disaster indication signals from complex, noisy, massive amounts of data, and are unable to make high-confidence comprehensive judgments on the time, location, and intensity of disasters. Therefore, the reliability of existing early warning systems is insufficient to meet practical application needs, and there is an urgent need to overcome key technological bottlenecks such as deep fusion, real-time processing, and intelligent analysis of multi-source heterogeneous marine data to significantly improve the accuracy and timeliness of deep-sea disaster early warning. Summary of the Invention

[0006] This invention provides a data analysis method, system, equipment, and medium for a deep-sea observation network system, aiming to solve the problems of poor accuracy and timeliness in existing seabed disaster early warning systems.

[0007] In a first aspect, embodiments of the present invention provide a data analysis method for a deep-sea observation network system, the deep-sea observation network system including sensors, edge nodes, and a cloud platform, the method comprising:

[0008] The edge node receives raw data sent by the sensor, converts the raw data into initial feature data, and transmits the initial feature data to the cloud platform;

[0009] The cloud platform performs time alignment on the initial feature data transmitted by all edge nodes to obtain aligned data;

[0010] The cloud platform performs feature fusion on the alignment data of all edge nodes to obtain fused feature data;

[0011] The cloud platform uses the fused feature data to provide early warning of seabed disasters.

[0012] A further technical solution is that converting the original data into initial feature data includes:

[0013] The original data is preprocessed to obtain preprocessed data;

[0014] The initial feature data is obtained by downsampling the preprocessed data.

[0015] A further technical solution is that the preprocessing of the original data to obtain preprocessed data includes:

[0016] The original data is subjected to wavelet transform denoising and outlier removal to obtain denoised data;

[0017] Feature extraction is performed on the denoised data to obtain the preprocessed data.

[0018] A further technical solution is that transmitting the initial feature data to the cloud platform includes:

[0019] Determine whether the initial feature data belongs to disaster data;

[0020] If the initial feature data is disaster data, the initial feature data is added to a preset high-priority queue.

[0021] A further technical solution involves fusing the alignment data of all edge nodes to obtain fused feature data, including:

[0022] The alignment data is subjected to feature extraction by a preset feature extraction module to obtain alignment feature data.

[0023] Based on a preset feature fusion model, the alignment feature data of all edge nodes are fused to obtain the fused feature data.

[0024] A further technical solution is that the seabed disaster early warning based on the fused feature data includes:

[0025] The fused feature data is used to perform isolated forest anomaly detection to obtain anomaly detection results;

[0026] The fused feature data is subjected to AutoEncoder reconstruction error analysis to obtain the error analysis results;

[0027] The fused feature data is trend-predicted using a pre-defined LSTM-ATT model to obtain the trend prediction results.

[0028] By using a pre-trained early warning rule engine, early warning of seabed disasters is generated based on the anomaly detection results, the error analysis results, and the trend prediction results.

[0029] A further technical solution is that the method further includes:

[0030] The warning result is sent to a preset recipient.

[0031] Secondly, embodiments of the present invention also provide a deep-sea observation network system, including sensors, edge nodes, and a cloud platform, wherein the deep-sea observation network system is used to perform the method described in the first aspect.

[0032] Thirdly, embodiments of the present invention also provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0033] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the above-described method.

[0034] This invention provides a data analysis method, system, device, and medium for a deep-sea observation network system. The deep-sea observation network system includes sensors, edge nodes, and a cloud platform. The method includes: edge nodes receiving raw data from sensors, converting the raw data into initial feature data, and transmitting the initial feature data to the cloud platform; the cloud platform performing time alignment on the initial feature data transmitted by all edge nodes to obtain aligned data; the cloud platform fusing the aligned data from all edge nodes to obtain fused feature data; and the cloud platform using the fused feature data for seabed disaster early warning. This invention achieves efficient integration and intelligent analysis of multi-source heterogeneous deep-sea data by constructing a collaborative processing architecture of "sensor-edge node-cloud platform." Specifically, the feature conversion operation performed by edge nodes on the raw data significantly reduces the redundancy and noise interference of the raw monitoring data (e.g., converting high-frequency acoustic waveforms into spectral features), greatly compressing the data volume while retaining disaster-sensitive information, thereby effectively alleviating the core bottleneck of limited deep-sea communication bandwidth resources. The time alignment mechanism implemented on the cloud platform establishes a unified time reference for cross-node data by correcting timing deviations caused by clock drift or transmission delays in distributed sensors (such as the time misalignment between salinity abrupt events and acoustic vibration signals), laying a foundation for time-series consistency in multi-source information correlation analysis. Based on this, the feature fusion process deeply mines the complementary characteristics of temperature, salinity, acoustic, and image data (e.g., temperature anomalies indicate hydrothermal activity, acoustic signals capture crustal stress release characteristics, and image data reflects seafloor topographic displacement), constructing a multi-dimensional disaster feature expression system and overcoming the feature loss problem caused by the limitations of a single data source perspective. Finally, the disaster early warning system implemented based on fused features significantly suppresses the risk of false alarms caused by equipment failure or environmental interference through a multi-source data cross-validation mechanism (e.g., acoustic pulse anomalies must be accompanied by simultaneous topographic displacement characteristics to trigger an early warning); simultaneously, it utilizes the synergistic enhancement effect of multimodal information (e.g., weak temperature changes combined with acoustic signal distortion can identify early earthquake precursors), effectively improving the detection sensitivity of weak disaster signals and significantly reducing the probability of missed alarms. This technical solution fundamentally breaks through the reliance of traditional deep-sea early warning systems on single-source data. Through a hierarchical data processing architecture and a multi-dimensional feature fusion mechanism, it achieves a systematic improvement in the accuracy of disaster identification while ensuring the timeliness of early warning, providing reliable technical support for marine disaster prevention and control. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 A flowchart illustrating the data analysis method for the deep-sea observation network system provided in this embodiment of the invention;

[0037] Figure 2 A schematic block diagram of a deep-sea observation network system provided in an embodiment of the present invention;

[0038] Figure 3 A flowchart of a data scheduling strategy provided in an embodiment of the present invention;

[0039] Figure 4 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present 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 the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0042] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0043] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0044] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrases "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0045] Please see Figure 1 This invention provides a data analysis method for a deep-sea observation network system. The deep-sea observation network system includes sensors, edge nodes, and a cloud platform. The number of sensors and edge nodes can be multiple. The method includes the following steps:

[0046] S1, the edge node receives the raw data sent by the sensor, converts the raw data into initial feature data, and transmits the initial feature data to the cloud platform.

[0047] In practice, the raw data includes temperature data reflecting the thermodynamic properties of water, salinity data indicating changes in the chemical composition of seawater, acoustic data capturing the propagation characteristics of underwater sound waves, and image data providing intuitive images of seabed topography and activity.

[0048] Edge nodes receive raw data from sensors, convert the raw data (temperature, salinity, acoustics, images, etc.) into initial feature data, and then transmit it to the cloud platform. Essentially, this completes data lightweighting and feature purification (such as compressing 1GB of raw images per second into key texture features), significantly reducing the transmission bandwidth pressure.

[0049] In some preferred embodiments, the above step "converting the original data into initial feature data" specifically includes the following steps: preprocessing the original data to obtain preprocessed data; and downsampling the preprocessed data to obtain the initial feature data.

[0050] In specific implementation, the original data is preprocessed to obtain preprocessed data; the preprocessed data is then downsampled to obtain the initial feature data. This invention achieves a balance between data quality and transmission efficiency through a two-stage processing approach of "preprocessing + downsampling." Preprocessing removes invalid information (such as periodic drift of salinity sensors due to biological attachment) to avoid noise contamination of the cloud model; downsampling intelligently compresses high-resolution data (such as 4K video streams) (e.g., extracting one frame of key action every 10 frames), reducing data transmission volume to 5%-10% of the original data while preserving disaster-sensitive features (such as the sliding trajectory of seabed sediments). For example, when downsampling acoustic data, the 500Hz-2kHz frequency band (characteristic frequency band of seabed earthquakes) is retained, while irrelevant high-frequency noise is discarded, reducing the transmission latency from edge nodes to the cloud from minutes to seconds.

[0051] In some preferred embodiments, the above step "preprocessing the original data to obtain preprocessed data" specifically includes the following steps: performing wavelet transform denoising and outlier removal on the original data to obtain denoised data; and performing feature extraction on the denoised data to obtain the preprocessed data.

[0052] In specific implementation, the original data undergoes wavelet transform denoising and outlier removal to obtain denoised data; feature extraction is then performed on the denoised data to obtain the preprocessed data. In this invention, wavelet transform denoising and 3σ outlier removal in the preprocessing form a complementary purification mechanism: wavelet transform separates the low-frequency trend (geothermal activity) and high-frequency noise (sensor circuit interference) in temperature data, while the 3σ principle dynamically identifies outliers in salinity data that exceed the normal fluctuation range (such as sudden chemical leaks). Subsequently, feature extraction transforms the original signal into quantifiable indicators (e.g., extracting the pixel expansion rate of seabed cracks from image data, or the resonant frequency offset from acoustic data), providing a unified mathematical basis for subsequent fusion processes.

[0053] Specifically, feature extraction is performed using pre-configured feature extraction methods for different types of data. For example, for acoustic data, Short Time Fourier Transform (STFT, window length 512, overlap rate 75%) can be used for feature extraction. For image data, a lightweight ResNet-18 model is used for feature extraction.

[0054] In some preferred embodiments, the above step "transmitting the initial feature data to the cloud platform" specifically includes the following steps: determining whether the initial feature data belongs to disaster data; if the initial feature data is disaster data, adding the initial feature data to a preset high-priority queue.

[0055] In specific implementation, based on preset judgment rules, it is determined whether the initial feature data belongs to disaster data; if the initial feature data is disaster data, the initial feature data is added to a preset high-priority queue to achieve priority transmission of disaster data. Optionally, underwater acoustic link is used for priority transmission.

[0056] In this invention, a priority transmission mechanism for disaster data establishes an emergency response channel. When an edge node detects initial characteristic data that matches a disaster pattern (e.g., a sudden temperature rise of 3°C + a salinity drop of 20% indicates a precursor to a submarine eruption), it immediately places the data in a high-priority queue to preempt bandwidth resources for transmission.

[0057] S2, the cloud platform performs time alignment on the initial feature data transmitted by all edge nodes to obtain aligned data.

[0058] In practical implementation, the cloud uses time alignment to resolve the data timing disorder caused by transmission delays in deep-sea sensors (e.g., aligning temperature data delayed by 2 seconds from node A with real-time salinity data from node B), ensuring the comparability of multi-dimensional data. Time alignment can be based on timestamps in the initial feature data; this invention does not specifically limit this.

[0059] S3, the cloud platform performs feature fusion on the alignment data of all edge nodes to obtain fused feature data.

[0060] In practice, the cloud platform performs feature fusion on the aligned data of all edge nodes to obtain fused feature data. Based on feature fusion, the limitations of single data are overcome—for example, when abnormal temperature rise (thermodynamic feature) is fused with high-frequency noise (acoustic feature) in acoustic data, composite signs of submarine volcanic activity can be identified, whereas relying solely on temperature data may lead to misjudgment as instrument error.

[0061] In some preferred embodiments, the above step "to fuse the alignment data of all edge nodes to obtain fused feature data" specifically includes the following steps: extracting features from the alignment data using a preset feature extraction module to obtain alignment feature data; and fusing the alignment feature data of all edge nodes based on a preset feature fusion model to obtain the fused feature data.

[0062] In practice, a pre-defined feature extraction module is used to specifically optimize the representation capabilities of different modalities of data (converting the original image into a landform displacement vector and the acoustic data into a spectrogram). For example, temporal features are extracted using LSTM, acoustic features are extracted using CNN, and image features are extracted using ResNet. The feature fusion model can be specifically a CNN-LSTM fusion model, which fuses multimodal features through fully connected layers to obtain the fused feature data.

[0063] S4, the cloud platform conducts early warning of seabed disasters based on the fused feature data.

[0064] In practice, early warning based on fusion features can reduce the false alarm rate compared to single-source data analysis (for example, relying solely on temperature data may overlook tectonic earthquakes without thermal anomalies), while suppressing false alarms through multi-source cross-validation (such as eliminating single data anomalies caused by instrument malfunction).

[0065] For example, in some preferred embodiments, the above step "conducting seabed disaster early warning based on the fused feature data" specifically includes the following steps: performing isolated forest anomaly detection on the fused feature data to obtain anomaly detection results; performing AutoEncoder reconstruction error analysis on the fused feature data to obtain error analysis results; performing trend prediction on the fused feature data using a preset LSTM-ATT model to obtain trend prediction results; and conducting seabed disaster early warning based on the anomaly detection results, the error analysis results, and the trend prediction results using a pre-trained early warning rule engine to obtain early warning results.

[0066] In practical implementation, the isolated forest algorithm detects sudden anomalies (such as instantaneous jumps in salinity data), the AutoEncoder reconstruction error identifies complex pattern anomalies (such as overall distribution shifts in geomagnetic data), and the LSTM-ATT model captures temporal evolution patterns (such as a sustained, gradual temperature rise indicating hydrothermal eruption), thus covering a full range of early warning scenarios from "instantaneous anomalies to pattern anomalies to trend anomalies" (a single method might miss slowly developing tectonic stress accumulation). Furthermore, the early warning rule engine integrates these three types of results (e.g., a red alert is triggered if both isolated forest and LSTM alarms simultaneously), outputting early warning results based on preset judgment rules, thereby improving the generalization ability for new types of disasters. It should be noted that the core principle of the early warning rule engine lies in transforming expert experience, disaster mechanism models, and machine learning decision-making logic into executable computer reasoning rules. Through weighted fusion and conflict resolution of multi-dimensional anomaly evidence, it achieves high-confidence disaster assessment.

[0067] In some preferred embodiments, the method further includes sending the warning result to a preset recipient.

[0068] In practice, the recipients include research institution terminals, monitoring platforms, and emergency command centers. Early warning results are simultaneously pushed to research institutions, monitoring platforms, and emergency command centers to achieve a closed-loop disaster response with multi-terminal collaboration. This ensures that early warning information is transformed into actual disaster reduction actions and avoids response delays caused by information silos (such as only notifying research institutions without triggering an emergency response).

[0069] This invention proposes a data analysis method for a deep-sea observation network system, which includes sensors, edge nodes, and a cloud platform. The method includes: edge nodes receiving raw data from sensors, converting the raw data into initial feature data, and transmitting the initial feature data to the cloud platform; the cloud platform performing time alignment on the initial feature data transmitted by all edge nodes to obtain aligned data; the cloud platform performing feature fusion on the aligned data of all edge nodes to obtain fused feature data; and the cloud platform performing seabed disaster early warning based on the fused feature data. This invention achieves efficient integration and intelligent analysis of multi-source heterogeneous deep-sea data by constructing a collaborative processing architecture of "sensor-edge node-cloud platform". Specifically, the feature conversion operation performed by the edge nodes on the raw data significantly reduces the redundancy and noise interference of the raw monitoring data (e.g., converting high-frequency acoustic waveforms into spectral features), greatly compressing the data volume while retaining disaster-sensitive information, thereby effectively alleviating the core bottleneck of limited deep-sea communication bandwidth resources. The time alignment mechanism implemented on the cloud platform establishes a unified time reference for cross-node data by correcting timing deviations caused by clock drift or transmission delays in distributed sensors (such as the time misalignment between salinity abrupt events and acoustic vibration signals), laying a foundation for time-series consistency in multi-source information correlation analysis. Based on this, the feature fusion process deeply mines the complementary characteristics of temperature, salinity, acoustic, and image data (e.g., temperature anomalies indicate hydrothermal activity, acoustic signals capture crustal stress release characteristics, and image data reflects seafloor topographic displacement), constructing a multi-dimensional disaster feature expression system and overcoming the feature loss problem caused by the limitations of a single data source perspective. Finally, the disaster early warning system implemented based on fused features significantly suppresses the risk of false alarms caused by equipment failure or environmental interference through a multi-source data cross-validation mechanism (e.g., acoustic pulse anomalies must be accompanied by simultaneous topographic displacement characteristics to trigger an early warning); simultaneously, it utilizes the synergistic enhancement effect of multimodal information (e.g., weak temperature changes combined with acoustic signal distortion can identify early earthquake precursors), effectively improving the detection sensitivity of weak disaster signals and significantly reducing the probability of missed alarms. This technical solution fundamentally breaks through the reliance of traditional deep-sea early warning systems on single-source data. Through a hierarchical data processing architecture and a multi-dimensional feature fusion mechanism, it achieves a systematic improvement in the accuracy of disaster identification while ensuring the timeliness of early warning, providing reliable technical support for marine disaster prevention and control.

[0070] See Figure 2 , Figure 2 This is a schematic block diagram of a deep-sea observation network system provided in an embodiment of the present invention. The deep-sea observation network system includes sensors 10, edge nodes 20, and a cloud platform 30. The deep-sea observation network system is used to execute the data analysis method of the deep-sea observation network system provided in any of the above method embodiments.

[0071] Specifically, the sensors include:

[0072] ① Temperature, salinity and depth sensor (CTD): The Sea-Bird SBE 41CP model is used, with a temperature measurement accuracy of ±0.001℃, a salinity accuracy of ±0.003, a depth range of 0-10000 meters, and supports a sampling frequency of 1Hz.

[0073] ② Acoustic sensors: including a wideband hydrophone (frequency band 20Hz-20kHz, sensitivity -170dB re1V / μPa) and an acoustic Doppler current meter (ADCP, measurement range 0.01-5m / s, accuracy ±1%).

[0074] ③ Image sensor: Deep-sea camera (4K resolution, 30fps frame rate, equipped with 500 lumens LED fill light, withstand pressure of 110MPa).

[0075] Sensor deployment method: fixed deployment of submarine base stations + mobile deployment of AUVs to form a three-dimensional observation network.

[0076] 2. Edge nodes

[0077] Hardware configuration:

[0078] ① Main control chip: NVIDIA Jetson AGX Orin (6-core ARM Cortex-A78AE, computing power 200 TOPS);

[0079] ② Storage: 16GB LPDDR5 memory + 512GB eMMC, supporting real-time data caching;

[0080] ③ Interfaces: 2 Gigabit Ethernet ports (for connecting sensors), 1 underwater acoustic communication interface (RS-485).

[0081] Core functions:

[0082] ① Data preprocessing: Wavelet transform denoising (for temperature data, decomposed into 5 layers, soft thresholding function) + outlier removal (3σ principle);

[0083] ②Feature extraction: Short-time Fourier transform (STFT, window length 512, overlap rate 75%) is used to extract time-frequency features from the acoustic signal.

[0084] Edge nodes are configured with a communication transport layer, including hybrid communication links, including underwater links and surface links.

[0085] The underwater link includes:

[0086] ① Underwater acoustic communication module: uses Kraken M3-Modem, QPSK modulation, 5kbps transmission rate (1000m depth), and forward error correction coding (FEC, code rate 1 / 2) to reduce bit error rate;

[0087] ②Optical communication alternative: Blue-green light communication module (wavelength 532nm, transmission rate 100kbps, suitable for water depth <200 meters).

[0088] Surface link: 4G / 5G wireless communication (bandwidth 10-100Mbps), supporting TCP / IP protocol stack.

[0089] Data scheduling strategy:

[0090] See Figure 3 Figure 3 is a flowchart of the data scheduling strategy. This flowchart describes the scheduling logic of data transmission in the deep-sea observation network system. First, the data is classified, and then it is determined whether it belongs to disaster data. If it is determined to be disaster data, it enters the high-priority queue and is transmitted preferentially through the underwater acoustic link. If it is not disaster data, it enters the regular queue and is transmitted in a time-slice polling manner. This ensures that disaster data can be transmitted quickly and improves the response efficiency of deep-sea disaster monitoring.

[0091] 4. Cloud platform

[0092] Infrastructure:

[0093] ① Server cluster: 10 Dell PowerEdge R750 servers (2×Intel Xeon Platinum 8368, 512GB memory, 8×A100 GPU);

[0094] ② Storage system: Ceph distributed file system (total capacity 10PB, redundancy strategy with 3 replicas).

[0095] Computational framework:

[0096] ① Real-time computing: Apache Flink (processing rate 100,000 records / second);

[0097] ② Batch computing: Apache Spark (supports parallel processing of TB-level data).

[0098] The cloud platform deploys an intelligent analytics engine, including algorithm components, which include:

[0099] Anomaly detection module:

[0100] ①Isolation Forest: 100 trees, 256 samples, used to identify isolated points such as sudden temperature changes;

[0101] ② Autoencoder: Input dimension 100 (100 time steps of data), hidden layer dimension 10, reconstruction error threshold 0.05.

[0102] Trend prediction module:

[0103] ①LSTM network: 2 hidden layers (128 neurons each), activation function tanh, input features include temperature, salinity, depth, pressure, and flow rate, predict data for the next 6 hours, loss function MSE;

[0104] ②CNN-LSTM fusion model: CNN extracts image features (ResNet-18 architecture), LSTM processes temporal features, and is used for seabed topography change prediction.

[0105] The data analysis device of the aforementioned deep-sea observation network system can be implemented as a computer program, which can perform operations such as... Figure 4 It runs on the computer device shown.

[0106] Please see Figure 4 , Figure 4 This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 500 can be a terminal or a server. The terminal can be an electronic device with communication functions, such as a smartphone, tablet, laptop, desktop computer, personal digital assistant, or wearable device. The server can be a standalone server or a server cluster composed of multiple servers.

[0107] The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504.

[0108] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, it enables the processor 502 to execute a data analysis method for a deep-sea observation network system.

[0109] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0110] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a data analysis method for a deep-sea observation network system.

[0111] The network interface 505 is used for network communication with other devices. Those skilled in the art will understand that the above structure is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 500 to which the present application is applied. A specific computer device 500 may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements.

[0112] The processor 502 is used to run a computer program 5032 stored in a memory to implement the steps of a data analysis method for a deep-sea observation network system provided in any of the above method embodiments.

[0113] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0114] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program may be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0115] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program. When executed by a processor, the computer program causes the processor to perform the steps of a data analysis method for a deep-sea observation network system provided in any of the above method embodiments.

[0116] The storage medium is a physical, non-transient storage medium, such as a USB flash drive, external hard drive, read-only memory (ROM), magnetic disk, or optical disk, or any other physical storage medium capable of storing program code. The computer-readable storage medium can be non-volatile or volatile.

[0117] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0118] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0119] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0120] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part 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, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0121] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0122] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Since these modifications and variations fall within the scope of the claims and their equivalents, this invention also intends to include these modifications and variations.

[0123] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A data analysis method for a deep-sea observation network system, characterized in that, The deep-sea observation network system includes sensors, edge nodes, and a cloud platform, and the method includes: The edge node receives raw data sent by the sensor, converts the raw data into initial feature data, and transmits the initial feature data to the cloud platform; The cloud platform performs time alignment on the initial feature data transmitted by all edge nodes to obtain aligned data; The cloud platform performs feature fusion on the alignment data of all edge nodes to obtain fused feature data; The cloud platform uses the fused feature data to provide early warning of seabed disasters.

2. The data analysis method for the deep-sea observation network system according to claim 1, characterized in that, The process of converting the original data into initial feature data includes: The original data is preprocessed to obtain preprocessed data; The initial feature data is obtained by downsampling the preprocessed data.

3. The data analysis method for the deep-sea observation network system according to claim 2, characterized in that, The preprocessing of the original data to obtain preprocessed data includes: The original data is subjected to wavelet transform denoising and outlier removal to obtain denoised data; Feature extraction is performed on the denoised data to obtain the preprocessed data.

4. The data analysis method for the deep-sea observation network system according to claim 1, characterized in that, The step of transmitting the initial feature data to the cloud platform includes: Determine whether the initial feature data belongs to disaster data; If the initial feature data is disaster data, the initial feature data is added to a preset high-priority queue.

5. The data analysis method for the deep-sea observation network system according to claim 1, characterized in that, The step involves fusing the alignment data of all edge nodes to obtain fused feature data, including: The alignment data is subjected to feature extraction by a preset feature extraction module to obtain alignment feature data; Based on a preset feature fusion model, the alignment feature data of all edge nodes are fused to obtain the fused feature data.

6. The data analysis method for the deep-sea observation network system according to claim 5, characterized in that, The method of providing early warning of seabed disasters based on the fused feature data includes: The fused feature data is used to perform isolated forest anomaly detection to obtain anomaly detection results; The fused feature data is subjected to AutoEncoder reconstruction error analysis to obtain the error analysis results; The fused feature data is trend-predicted using a pre-defined LSTM-ATT model to obtain the trend prediction results. By using a pre-trained early warning rule engine, early warning of seabed disasters is generated based on the anomaly detection results, the error analysis results, and the trend prediction results.

7. The data analysis method for the deep-sea observation network system according to claim 6, characterized in that, The method further includes: The warning result is sent to a preset recipient.

8. A deep-sea observation network system, characterized in that, The deep-sea observation network system includes sensors, edge nodes, and a cloud platform, and is used to perform the method as described in any one of claims 1-7.

9. A computer device, characterized in that, The computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, can implement the method as described in any one of claims 1-7.

Images