Multi-parameter in-situ detection data analysis system for marine ranching

Abstract

The invention discloses a multi-parameter in-situ detection data analysis system for a marine ranch, and relates to the technical field of data analysis, the multi-parameter in-situ detection data analysis system comprises a multi-parameter sensor group and a self-adaptive multi-dimensional trigger type sampling control module, the multi-parameter sensor group comprises an environmental parameter sensor and a biological behavior sensor, the environmental parameter sensor is used for measuring real-time environmental parameters of the marine ranch, the biological behavior sensor is used for monitoring biological behavior parameters of the marine ranch in real time, and the self-adaptive multi-dimensional trigger type sampling control module comprises a data preprocessing unit and a dynamic sampling strategy unit. The self-adaptive multi-dimensional trigger type sampling control module is designed, so that sampling modes are dynamically switched according to real-time dual-mode data, environment sudden changes are effectively captured, data redundancy or key change sampling leakage caused by traditional fixed sampling frequency is avoided, monitoring precision is improved, energy optimization is achieved, the service life of equipment is prolonged, and the system is suitable for large-scale popularization and application. And high quality and real-time performance of data are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a multi-parameter in-situ detection data analysis system for marine ranching. Background Art

[0002] With the continuous expansion of marine ranching and the diversification of aquaculture models, the complexity and dynamic changes of the marine environment have put forward higher requirements for aquaculture management.

[0003] Traditional marine ranch monitoring systems often rely on single parameter collection and offline data analysis, and are unable to adjust the monitoring sampling frequency in real time according to changing environmental and biological behavior data. This results in redundancy in data collected at a fixed sampling frequency and omission of key changes, making it difficult to detect and respond to sudden environmental anomalies in the first place, thus affecting the analysis of the water quality and ecological status of marine ranches.

[0004] Patent document CN118393098B discloses a method and system for remote monitoring of water quality in marine ranches. The above patent realizes comprehensive monitoring of substances at different depths in the marine ranch and their substance contents, which can not only greatly broaden the scope of monitoring, but also reduce monitoring deviations and effectively improve monitoring accuracy.

[0005] In summary, the above patent improves monitoring accuracy by comprehensively monitoring substances and contents at different depths, but does not consider dynamically adjusting the monitoring mode according to environmental changes during the monitoring process, and cannot avoid data redundancy and missed sampling of key changes caused by fixed sampling frequency; To this end, the present application proposes a multi-parameter in-situ detection data analysis system for marine ranching that is capable of dynamically switching sampling modes based on real-time dual-modal data. Summary of the invention

[0006] The object of the present invention is to provide a multi-parameter in-situ detection data analysis system for marine ranching, so as to solve the technical problems of data redundancy and missed sampling of key changes caused by fixed sampling frequency proposed in the above background technology.

[0007] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a multi-parameter in-situ detection data analysis system for marine ranches, comprising a multi-parameter sensor group and an adaptive multi-dimensional trigger sampling control module, wherein the multi-parameter sensor group comprises an environmental parameter sensor and a biological behavior sensor, wherein the environmental parameter sensor is used to measure the real-time environmental parameters of the marine ranch, and the biological behavior sensor is used to monitor the biological behavior parameters of the marine ranch in real time, wherein the adaptive multi-dimensional trigger sampling control module comprises a data preprocessing unit and a dynamic sampling strategy unit, wherein the data preprocessing unit performs denoising, outlier screening and time-space alignment processing on the data collected by the multi-parameter sensor group, and the dynamic sampling strategy unit analyzes the gradient changes of environmental parameters and biological behavior patterns, and adjusts the sampling frequency according to dynamic logic: When the dissolved oxygen change rate is greater than the preset threshold or an abnormal increase in the activity intensity of fish schools is detected, the high-frequency sampling mode is triggered; When environmental parameters and biological behavior characteristics are stable for a long time, switch to low-power intermittent sampling mode.

[0008] Preferably, the environmental parameter sensors are used to measure water temperature, salinity, dissolved oxygen, pH, chlorophyll a, turbidity and nutrient concentration; the biological behavior sensors include: dual-frequency underwater sonar for detecting fish density, group movement patterns, and individual behavior, high-definition cameras for analyzing fish health status, feeding behavior and population structure, and hydrophone arrays for passive detection of fish vocalization behavior and abnormal signals.

[0009] Preferably, the adaptive multi-dimensional triggered sampling control module combines a sliding window algorithm and an adaptive neural network to optimize dynamic logic to adjust the trigger threshold of the sampling frequency, thereby improving the sensitivity and reliability of environmental anomalies.

[0010] Preferably, the analysis system further comprises an anomaly detection algorithm based on biological behavior feedback, the anomaly detection algorithm comprising: a time series prediction model and an anomaly alarm mechanism; The time series prediction model builds an LSTM model based on the fish movement trajectory, feeding behavior and habitat status to calculate the probability of normal behavior; The abnormal alarm mechanism is that when the probability of normal behavior is lower than the set threshold, the early warning mechanism is automatically triggered, the time and space coordinates of the abnormal event are marked, and the abnormal report is transmitted to the remote monitoring center through underwater acoustic communication.

[0011] Preferably, the analysis system further includes a multimodal data fusion method, which includes: a multi-source data alignment algorithm based on deep learning and a spatiotemporal attention mechanism; A multi-source data alignment algorithm based on deep learning is used to synchronize the time and space of sonar point clouds, video streams, and environmental parameters; The spatiotemporal attention mechanism analyzes the correlation between multi-source data, calculates and outputs the marine ranch health index, which is used to evaluate the ecological status of the ranch.

[0012] Preferably, the analysis system further comprises a deployed edge-cloud collaborative computing architecture, the edge-cloud collaborative computing architecture comprising: an edge computing unit and a cloud computing center; The edge computing unit performs data preprocessing at the monitoring node end, including anomaly screening, feature extraction, and preliminary analysis; Based on high-dimensional environmental models and AI prediction algorithms, the cloud computing center deduces the nutrient cycle of marine ranches and predicts future trends in water quality changes.

[0013] Preferably, the analysis system adopts an adaptive energy management strategy, which includes: an energy scheduling algorithm and a low power consumption mode; The energy scheduling algorithm dynamically switches the power supply mode according to system load and environmental changes, giving priority to solar energy and assisting wave energy and tidal energy in power supply; The low power consumption mode automatically reduces the operating frequency of the multi-parameter sensor group to reduce energy consumption when the monitoring data is stable for a long time.

[0014] Preferably, the multi-parameter sensor group packaging structure includes an anti-fouling self-cleaning component for reducing biological attachment pollution on the sensor surface, and the anti-fouling self-cleaning component includes: a micro-mechanical scraper arm and an ultrasonic generator; The micro-mechanical scraper arm periodically cleans the sensor surface attachments to ensure measurement accuracy; The ultrasonic generator uses high-frequency ultrasonic oscillations to inhibit the attachment of microorganisms and prevent errors in collected data caused by long-term immersion in seawater.

[0015] Preferably, the analysis system further includes a three-dimensional visualization engine and a blockchain data storage mechanism. The three-dimensional visualization engine is used to render the marine ranch environment status in real time. The three-dimensional visualization engine includes: a real-time environmental data rendering module and a virtual reality interactive support. The real-time environmental data rendering module is used to generate a temperature field, a dissolved oxygen concentration gradient field, and a fish movement vector field. The virtual reality interactive support allows users to interactively view the distribution of marine ranch hot zones through VR devices and perform real-time data analysis. The blockchain data notarization mechanism includes: hash value generation algorithm and private chain notarization system. The hash value generation algorithm calculates a unique hash value for each sampling batch to indicate data integrity. The private chain notarization system writes the hash value, timestamp and spatial coordinates into the blockchain node to provide traceable data source capabilities.

[0016] Preferably, the analysis system further deploys mobile detection nodes and multi-pasture comparative analysis modules. The mobile detection nodes are used to expand the monitoring range, including: an underwater autonomous robot and a hydroacoustic communication network. The underwater autonomous robot is equipped with a multi-parameter sensor group and cruises along a preset spiral path to cover the monitoring area. The hydroacoustic communication network is used for data synchronization between the autonomous robot and the fixed nodes to achieve multi-node collaborative perception and fill the monitoring blind spots. The multi-pasture comparative analysis module is used to optimize fishery management decisions, including a spatiotemporal data feature extraction unit and a transfer learning optimization strategy. The spatiotemporal data feature extraction unit is used to analyze the spatiotemporal distribution patterns of environmental parameters of multiple pastures. The transfer learning optimization strategy trains models based on data sets of different marine pastures and outputs recommendations on the best feeding and harvesting times to improve pasture production efficiency.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention is designed with an adaptive multi-dimensional triggered sampling control module to achieve dynamic switching of sampling modes according to real-time dual-modal data, effectively capture environmental mutations, avoid data redundancy or missed sampling of key changes caused by traditional fixed sampling frequencies, improve monitoring accuracy, achieve energy optimization, extend equipment life, and ensure high data quality and real-time performance; 2. The present invention realizes the multi-node collaborative sensing function by designing a mobile detection node, overcomes the problem of limited coverage of traditional fixed monitoring equipment, provides more comprehensive and dynamic marine environment monitoring data, and improves the accuracy of management decisions; 3. The present invention is designed with a three-dimensional visualization engine to realize VR immersive data analysis and display functions, which makes up for the shortcomings of traditional data presentation methods that are abstract and difficult to intuitively understand, helps managers quickly grasp the dynamics of marine ranches, and provides intuitive support for real-time regulation and decision-making; 4. The present invention realizes the traceability of collected data by designing a blockchain data storage mechanism, prevents data from being tampered with during transmission and storage, enhances data security, and provides a reliable data source guarantee for marine ranch monitoring data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the workflow of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] See also Figure 1 , an embodiment provided by the present invention: a multi-parameter in-situ detection data analysis system for marine ranching, comprising a multi-parameter sensor group and an adaptive multi-dimensional trigger sampling control module, the multi-parameter sensor group comprising an environmental parameter sensor and a biological behavior sensor, the environmental parameter sensor is used to measure the real-time environmental parameters of the marine ranch, the biological behavior sensor is used to monitor the biological behavior parameters of the marine ranch in real time, the adaptive multi-dimensional trigger sampling control module comprises a data preprocessing unit and a dynamic sampling strategy unit, the data preprocessing unit performs denoising, outlier screening and time-space alignment processing on the data collected by the multi-parameter sensor group, the dynamic sampling strategy unit analyzes the gradient changes of environmental parameters and biological behavior patterns, and adjusts the sampling frequency according to dynamic logic: When the dissolved oxygen change rate is greater than the preset threshold or an abnormal increase in the activity intensity of fish schools is detected, the high-frequency sampling mode is triggered; When environmental parameters and biological behavior characteristics are stable for a long time, switch to low-power intermittent sampling mode; The environmental parameter sensors are used to measure water temperature, salinity, dissolved oxygen, pH, chlorophyll a, turbidity and nutrient concentration; the biological behavior sensors include: dual-frequency underwater sonar for detecting fish density, group movement patterns, and individual behavior; high-definition cameras for analyzing fish health status, feeding behavior and population structure; and hydrophone arrays for passive detection of fish vocalization behavior and abnormal signals; The adaptive multi-dimensional triggered sampling control module combines the sliding window algorithm and the adaptive neural network optimization dynamic logic to adjust the trigger threshold of the sampling frequency, thereby improving the sensitivity and reliability of environmental anomalies; Furthermore, water temperature, salinity, and pH sensors are based on electrode or MEMS sensing technology. They collect water surface temperature through buoys and monitor environmental changes in different water layers in combination with CTD. Dissolved oxygen sensors use fluorescence quenching to collect dissolved oxygen in water. Chlorophyll a sensors use fluorescence detection technology to estimate primary productivity of water bodies and judge phytoplankton concentration in combination with turbidity data. Nutrient sensors include nitrate, nitrite, and phosphate sensors. Based on ion-selective electrodes or spectral absorption principles, they analyze the degree of eutrophication of water bodies in real time. Dual-frequency underwater sonar uses a low-frequency mode of 100-300kHz to detect fish density and distribution range, and a high-frequency mode of more than 500kHz to provide individual-level fish morphology and movement status information. High-definition cameras are mounted on underwater buoys or underwater robots and use image processing technology to assess fish health, including abnormal swimming and feeding rates. Hydrophone arrays are used to capture fish vocalizations and use machine learning to analyze changes in fish behavior and infer environmental stress factors such as hypoxia and pollution. By combining environmental sensors with biological behavior sensors, physical-chemical-biological multimodal monitoring is achieved to improve the monitoring accuracy of abnormal events. The raw data collected by the multi-parameter sensor group is denoised by wavelet transform in the data preprocessing unit, and the sliding window outlier detection algorithm is used to screen out drift data, and the data of multiple sensors are aligned in time and space to ensure that different data sources can be analyzed at the same time point to avoid misjudgment due to acquisition time difference; if the system detects that the dissolved oxygen change rate is greater than the preset threshold of 0.3mg / L / h, it may indicate the risk of water hypoxia. The system immediately increases the monitoring frequency of dissolved oxygen, chlorophyll a and fish behavior to determine whether it is an algal bloom or pollution event. If the underwater sonar or hydrophone detects an abnormal increase in the intensity of fish activity, it may indicate that the fish are under environmental stress. The system increases the water quality sampling frequency to find out the cause. If the environmental parameters and biological behavior are stable for 24 hours, such as the dissolved oxygen change rate is lower than 0.05mg / L / h, the system automatically reduces the operating frequency of the sensor to extend the battery life of the equipment, and uses the sliding window algorithm to continuously update the preset threshold; through the dynamic sampling strategy, the system can automatically adjust the monitoring accuracy, which can capture emergencies and reduce power consumption; Through the joint monitoring of environmental parameters and biological behavior, the accuracy of anomaly detection is improved, and the sampling frequency is adjusted according to actual conditions, so that energy consumption can be reduced by 30%-50%, while the response speed to emergencies is improved.

[0021] See also Figure 1 , an embodiment provided by the present invention: a multi-parameter in-situ detection data analysis system for marine ranching, the analysis system also includes an anomaly detection algorithm based on biological behavior feedback, the anomaly detection algorithm includes: a time series prediction model and an anomaly alarm mechanism; The time series prediction model builds an LSTM model based on the fish movement trajectory, feeding behavior and habitat status to calculate the probability of normal behavior; The abnormal alarm mechanism is that when the probability of normal behavior is lower than the set threshold, the early warning mechanism is automatically triggered, and the time and space coordinates of the abnormal event are marked, and the abnormal report is transmitted to the remote monitoring center through underwater acoustic communication; The analysis system also includes a multimodal data fusion method, which includes: a multi-source data alignment algorithm based on deep learning and a spatiotemporal attention mechanism; A multi-source data alignment algorithm based on deep learning is used to synchronize the time and space of sonar point clouds, video streams, and environmental parameters; The spatiotemporal attention mechanism analyzes the correlation between multi-source data, calculates and outputs the marine ranch health index, which is used to evaluate the ecological status of the ranch; Furthermore, data input: fish movement trajectory data collected by underwater sonar, feeding behavior data collected by high-definition cameras, and habitat status data collected by hydrophone arrays. The DBSCAN clustering algorithm is used to eliminate abnormal points, such as mutation data caused by noise interference, and standardized processing is performed to ensure that the scales of different data sources are consistent to improve the stability of model training; the LSTM model is trained with historical data from the past 1-2 weeks to learn the behavior patterns of fish schools, set a normal behavior probability threshold of 85%-95%, and predict the behavior status of fish schools in the next 10-30 minutes through a sliding window. The probability that the current behavior of fish schools deviates from the normal mode is calculated. When the normal behavior probability calculated by the LSTM model is lower than 80%, the system determines that abnormal behavior has occurred. The system marks the abnormal time through time and space coordinates, analyzes possible causes in combination with environmental parameters, triggers an early warning mechanism, and sends an abnormal report to the remote monitoring center through underwater acoustic communication or wireless communication; Compared with relying solely on water quality parameters, it can detect abnormal behavior in advance before the abnormal water body affects the health of fish schools. LSTM can avoid false alarms caused by short-term fluctuations by learning long-term behavior patterns; The multi-source data alignment algorithm based on deep learning and the spatiotemporal attention mechanism realize data fusion and intelligent analysis, and use the dynamic time warping algorithm to synchronize the time of data of different frequencies. For example, water quality parameters are sampled every 5 minutes, while the video has one frame every 30 seconds. The water quality data needs to be interpolated to the time point that matches the video. The point cloud matching based on the ICP algorithm is used to make the spatial coordinates of the sonar point cloud data consistent with the camera video data; correlation analysis in the spatiotemporal attention mechanism: calculate the temporal correlation of multiple data, such as: if the chlorophyll concentration suddenly increases, and the camera detects an increase in water turbidity, it may be an algal bloom. If the dissolved oxygen decreases and the fish feed decreases, it may be water pollution or hypoxia; the Transformer spatiotemporal attention model is used to calculate the impact weight of each data on the environmental health status and generate the marine ranch health index. The index range example is: 90-100: Healthy state, stable environment, normal fish activities; 70-90: Mild abnormality, such as reduced fish feeding but no obvious water quality abnormality; 50-70: Moderate risk, such as decreased dissolved oxygen, fish aggregation or rapid swimming; 0-50: severe warning, such as fish escape, feeding stops completely, dissolved oxygen falls below the survival threshold; Marine ranch managers can adjust their farming strategies based on the health index, such as adding oxygenation equipment or adjusting the amount of feed. By aligning spatiotemporal data and integrating information from different sensors, they can provide a complete ecological monitoring perspective, quantify the health of the environment, and facilitate decision-making and management.

[0022] See also Figure 1, an embodiment provided by the present invention: a multi-parameter in-situ detection data analysis system for marine ranching, the analysis system further comprises a deployed edge-cloud collaborative computing architecture, the edge-cloud collaborative computing architecture comprises: an edge computing unit and a cloud computing center; The edge computing unit performs data preprocessing at the monitoring node end, including anomaly screening, feature extraction, and preliminary analysis; The cloud computing center uses high-dimensional environmental models and AI prediction algorithms to deduce the nutrient cycle of marine ranches and predict future water quality trends; Furthermore, at each monitoring node of the marine ranch, the edge computing unit is directly connected to the multi-parameter sensor group to collect water temperature, salinity, dissolved oxygen, pH, chlorophyll a, turbidity, nutrient concentration and biological behavior data in real time. After collecting the data, the edge computing unit first removes noise from the raw data using an embedded processor. For example, a Kalman filter or wavelet denoising algorithm is used to filter out high-frequency noise caused by waves, sensor drift or environmental interference; the collected data is compared in real time through set thresholds and statistical methods such as anomaly detection algorithms based on sliding windows to remove abnormal values ​​that exceed a reasonable range. The edge node can also use simple machine learning models such as decision trees or support vector machines to preliminarily classify the data, mark potential abnormal states, and extract key features from the preprocessed data, such as the rate of change of dissolved oxygen, temperature gradient, fish density and movement speed, and fish behavior characteristics captured by the camera. After compression and format conversion, these characteristic data form structured information, which is convenient for subsequent transmission and advanced analysis. The edge computing unit can perform simple statistical analysis, such as trend judgment of recent data. If local anomalies are detected, such as a sudden drop in dissolved oxygen or a sharp change in fish density, an early warning signal is immediately generated locally and key data is uploaded first. In addition, the edge can also perform data tagging, classifying data according to timestamps, spatial coordinates and anomaly tags, providing a basis for subsequent cloud-based fusion analysis. Data from multiple edge nodes are collected to the cloud computing center through wireless or underwater acoustic communication. The cloud platform first aggregates the structured data from different nodes, and uses time synchronization and space matching algorithms such as dynamic time warping (DTW) and ICP algorithms to ensure the temporal and spatial consistency of multi-source data. Based on long-term historical data and real-time data, the cloud platform builds a high-dimensional environmental model. The model comprehensively considers water temperature, salinity, dissolved oxygen, nutrient concentration and biological parameters, and establishes a mathematical model to describe the nutrient cycle in the marine ranch. Using this model, the absorption, transformation and sedimentation process of nutrients such as nitrogen and phosphorus in the marine ranch can be deduced, and then the algae growth, primary productivity and ecological carrying capacity can be evaluated; advanced AI prediction algorithms such as deep neural networks and LSTM time series models are deployed on the cloud to train and predict the integrated high-dimensional data. By learning about the nutrient cycle and changes in environmental parameters, the cloud platform can predict the trend of water quality changes in the next few hours to days, such as further decline in dissolved oxygen, pH fluctuations or abnormal increases in nutrient concentrations. In addition, the prediction results will be presented in the form of graphics, trend curves and health indexes to form a detailed environmental assessment report; The edge-cloud collaborative architecture realizes the full-process closed loop of data collection, preprocessing, comprehensive analysis, prediction and early warning, and feedback and control, which greatly improves the intelligence level and application efficiency of the monitoring system, and helps to achieve sustainable and stable management of the marine ranch environment.

[0023] See also Figure 1 , an embodiment provided by the present invention: a multi-parameter in-situ detection data analysis system for marine ranching, the analysis system adopts an adaptive energy management strategy, the adaptive energy management strategy includes: an energy scheduling algorithm and a low power consumption mode; The energy scheduling algorithm dynamically switches the power supply mode according to system load and environmental changes, giving priority to solar energy and assisting wave energy and tidal energy in power supply; Low power consumption mode means automatically reducing the operating frequency of the multi-parameter sensor group to reduce energy consumption when the monitoring data is stable for a long time; The multi-parameter sensor group packaging structure includes an anti-fouling self-cleaning component for reducing biological attachment pollution on the sensor surface, and the anti-fouling self-cleaning component includes: a micro-mechanical scraper arm and an ultrasonic generator; The micro-mechanical scraper arm periodically cleans the sensor surface attachments to ensure measurement accuracy; The ultrasonic generator uses high-frequency ultrasonic oscillation to inhibit the attachment of microorganisms and prevent errors in collected data caused by long-term immersion in seawater; Furthermore, light sensors, wave sensors and tidal current sensors are integrated into the system to detect the real-time status of solar energy, wave energy and tidal current energy respectively. At the same time, the embedded controller regularly collects energy sensor data to monitor the current load demand and battery energy storage status of the system. The embedded controller determines the current environmental conditions and system load according to the preset algorithm: when the sunlight is sufficient, solar panels are used for power supply first; when the solar energy output decreases or the load demand is high, the wave energy and tidal current energy conversion device is activated to convert mechanical energy into electrical energy through the integrated energy converter to supplement the power supply; The housing of each multi-parameter sensor group is designed with a built-in micro-mechanical scraper arm, which is driven by a micro-motor or a stepper motor. The system automatically starts the cleaning program according to the preset time interval or when the sensor detects abnormal data. The scraper arm reciprocates along the predetermined path on the sensor surface to physically remove algae, shells and other marine biological sediments attached to the sensor surface, ensuring that the sensor always remains clean, thereby ensuring measurement accuracy. The sensor package is embedded with a micro-ultrasonic generator, which can generate high-frequency ultrasonic waves, such as tens to hundreds of kilohertz. Ultrasonic oscillations cause the formation and rupture of tiny bubbles on the sensor surface and in the surrounding medium, namely cavitation. This continuous high-frequency vibration can interfere with the process of microorganisms and other organisms attaching to the sensor surface, thereby preventing the accumulation of dirt caused by long-term immersion in seawater; Dynamic adjustment of power supply mode and sampling frequency effectively improves the system energy efficiency and endurance. The automatic cleaning mechanism ensures that the sensor is always in the best working condition and guarantees data accuracy. The overall system is highly intelligent and has autonomous maintenance capabilities, providing stable, low-energy, and high-precision technical support for marine ranch environmental monitoring.

[0024] See also Figure 1 , an embodiment provided by the present invention: a multi-parameter in-situ detection data analysis system for marine ranching, the analysis system further comprising a three-dimensional visualization engine and a blockchain data storage mechanism, the three-dimensional visualization engine is used to render the marine ranch environment status in real time, the three-dimensional visualization engine comprises: a real-time environmental data rendering module and a virtual reality interactive support, the real-time environmental data rendering module is used to generate a temperature field, a dissolved oxygen concentration gradient field and a fish school motion vector field, the virtual reality interactive support allows a user to interactively view the marine ranch hot zone distribution through a VR device, and perform real-time data analysis; The blockchain data notarization mechanism includes: hash value generation algorithm and private chain notarization system. The hash value generation algorithm calculates a unique hash value for each sampling batch to indicate data integrity. The private chain notarization system writes the hash value, timestamp and spatial coordinates into the blockchain node to provide the ability to trace data sources. Furthermore, the preprocessed data is converted into numerical information in three-dimensional space through a mapping algorithm. Using GPU acceleration and a high-performance rendering engine, the system renders the temperature field, dissolved oxygen concentration gradient field, and fish movement vector field in three-dimensional scenes in real time, ensuring that each parameter field can be displayed in a dynamic and continuous manner. Users can enter the virtual scene through a VR headset for real-time viewing. For each batch of sampling data, the system uses a preset cryptographic hash function such as SHA-256 to calculate the data summary, that is, to generate a unique hash value. The hash value can uniquely identify the batch of data and ensure that even if the data is slightly modified, the hash value will change significantly, thereby reflecting the integrity of the data. While generating the hash value, the system will package the batch data with key information such as the sampling time and the spatial coordinates of the sampling location, and encrypt it to ensure the security of data transmission and storage. The data summary, timestamp and spatial coordinates after hash calculation will be written to the blockchain node as block content through the private chain network. Each block contains the hash value of the previous block to achieve chain storage, ensuring that the data cannot be tampered with once recorded, and the evidence storage system establishes a complete traceability record of the data. Any questions about the integrity of the data can be verified by querying the hash value, timestamp and spatial coordinates of the corresponding block to verify whether the original sampling data has been modified, thereby providing a highly reliable data source; Real-time rendering of the ocean ranch's temperature field, dissolved oxygen gradient, and fish movement vector helps managers intuitively understand the state of the environment. Hash values ​​are used to ensure the integrity and uniqueness of each batch of sampling data to prevent data from being tampered with during transmission or storage. The private chain evidence system records detailed time and space information, providing a reliable basis for data traceability and auditing, and enhancing the credibility and legal effect of the data.

[0025] See also Figure 1 , an embodiment provided by the present invention: a multi-parameter in-situ detection data analysis system for marine ranches, the analysis system further deploys mobile detection nodes and multi-ranch comparison analysis modules, the mobile detection nodes are used to expand the monitoring range, including: an underwater autonomous robot and a hydroacoustic communication network, the underwater autonomous robot is equipped with a multi-parameter sensor group, and cruises along a preset spiral path to cover the monitoring area, the hydroacoustic communication network is used for data synchronization between the autonomous robot and the fixed node, to achieve multi-node collaborative perception, and fill the monitoring blind spot; The multi-ranch comparative analysis module is used to optimize fishery management decisions, including a spatiotemporal data feature extraction unit and a transfer learning optimization strategy. The spatiotemporal data feature extraction unit is used to analyze the spatiotemporal distribution patterns of environmental parameters of multiple ranches. The transfer learning optimization strategy trains models based on data sets from different marine ranches and outputs recommendations on the best feeding and harvesting times to improve ranch production efficiency. Furthermore, the underwater autonomous robot is equipped with a multi-parameter sensor group. The underwater autonomous robot cruises in the marine ranch according to a pre-planned spiral path to ensure coverage of all monitoring areas, thereby making up for the blind spots of fixed monitoring nodes. During the cruise, the robot synchronizes data with fixed monitoring nodes in real time through an underwater acoustic communication network. This communication mechanism enables data to be shared between nodes, forming a collaborative perception network, further improving the overall monitoring coverage and data integrity. Environmental parameter data such as water quality, temperature, and fish distribution are collected from different pastures. The distribution patterns of each pasture at different time and space scales are extracted using spatiotemporal analysis algorithms. The spatiotemporal feature data of each pasture are integrated to provide a unified feature representation for subsequent model training. Based on different pasture data sets, the prediction model is trained using the transfer learning method, so that the knowledge learned from one pasture can be transferred to other pastures to compensate for the impact of data differences. After training, the model outputs recommendations on the best feeding and harvesting times, thereby helping fishery managers optimize breeding strategies and improve production efficiency.

[0026] Working principle: A multi-parameter sensor group that integrates environmental parameter sensors and biological behavior sensors is used to collect real-time environmental data such as water temperature, salinity, dissolved oxygen, pH, chlorophyll a, turbidity, nutrient concentration, and biological information such as fish density, movement patterns, and feeding behavior in the marine ranch. After the data is filtered out by noise, screened for outliers, and aligned in time and space by the pre-processing unit, it provides high-quality input for subsequent dynamic sampling and data fusion. The adaptive multi-dimensional triggered sampling control module dynamically adjusts the sampling frequency according to the real-time detected environmental parameter gradient changes and biological behavior patterns: when the dissolved oxygen change rate exceeds the threshold or the fish activity is abnormal, the system quickly switches to high-frequency sampling mode to capture sudden changes; when the data is stable for a long time, the sampling frequency is automatically reduced to enter low-power mode to optimize energy utilization; After edge computing and cloud data fusion processing, the system uses a three-dimensional visualization engine to render the temperature field, dissolved oxygen concentration gradient field and fish movement vector field in real time. At the same time, it uses the blockchain data storage mechanism to generate a unique hash value for each batch of data and record the time and space coordinates to ensure data integrity and traceability, providing precise support for intelligent decision-making in marine ranches.

[0027] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A multi-parameter in-situ detection data analysis system for marine ranching, comprising a multi-parameter sensor group and an adaptive multi-dimensional trigger sampling control module, characterized in that: The multi-parameter sensor group includes an environmental parameter sensor and a biological behavior sensor. The environmental parameter sensor is used to measure the real-time environmental parameters of the marine ranch. The biological behavior sensor is used to monitor the biological behavior parameters of the marine ranch in real time. The adaptive multi-dimensional triggered sampling control module includes a data preprocessing unit and a dynamic sampling strategy unit. The data preprocessing unit performs denoising, outlier screening and time-space alignment processing on the data collected by the multi-parameter sensor group. The dynamic sampling strategy unit analyzes the gradient changes of environmental parameters and biological behavior patterns, and adjusts the sampling frequency according to dynamic logic: When the dissolved oxygen change rate is greater than the preset threshold or an abnormal increase in the activity intensity of fish schools is detected, the high-frequency sampling mode is triggered; When environmental parameters and biological behavior characteristics are stable for a long time, switch to low-power intermittent sampling mode.

2. A multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The environmental parameter sensors are used to measure water temperature, salinity, dissolved oxygen, pH, chlorophyll a, turbidity and nutrient concentration; the biological behavior sensors include: dual-frequency underwater sonar for detecting fish density, group movement patterns, and individual behavior, high-definition cameras for analyzing fish health status, feeding behavior and population structure, and hydrophone arrays for passive detection of fish vocalization behavior and abnormal signals.

3. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The adaptive multi-dimensional triggered sampling control module combines a sliding window algorithm and an adaptive neural network to optimize dynamic logic to adjust the trigger threshold of the sampling frequency, thereby improving the sensitivity and reliability of environmental anomalies.

4. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The analysis system also includes an anomaly detection algorithm based on biological behavior feedback, the anomaly detection algorithm includes: a time series prediction model and an anomaly alarm mechanism; The time series prediction model builds an LSTM model based on the fish movement trajectory, feeding behavior and habitat status to calculate the probability of normal behavior; The abnormal alarm mechanism is that when the probability of normal behavior is lower than the set threshold, the early warning mechanism is automatically triggered, the time and space coordinates of the abnormal event are marked, and the abnormal report is transmitted to the remote monitoring center through underwater acoustic communication.

5. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The analysis system also includes a multimodal data fusion method, which includes: a multi-source data alignment algorithm based on deep learning and a spatiotemporal attention mechanism; A multi-source data alignment algorithm based on deep learning is used to synchronize the time and space of sonar point clouds, video streams, and environmental parameters; The spatiotemporal attention mechanism analyzes the correlation between multi-source data, calculates and outputs the marine ranch health index, which is used to evaluate the ecological status of the ranch.

6. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The analysis system further includes a deployed edge-cloud collaborative computing architecture, the edge-cloud collaborative computing architecture including: an edge computing unit and a cloud computing center; The edge computing unit performs data preprocessing at the monitoring node end, including anomaly screening, feature extraction, and preliminary analysis; Based on high-dimensional environmental models and AI prediction algorithms, the cloud computing center deduces the nutrient cycle of marine ranches and predicts future trends in water quality changes.

7. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The analysis system adopts an adaptive energy management strategy, which includes: an energy scheduling algorithm and a low power consumption mode; The energy scheduling algorithm dynamically switches the power supply mode according to system load and environmental changes, giving priority to solar energy and assisting wave energy and tidal energy in power supply; The low power consumption mode automatically reduces the operating frequency of the multi-parameter sensor group to reduce energy consumption when the monitoring data is stable for a long time.

8. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The multi-parameter sensor group packaging structure includes an anti-fouling self-cleaning component for reducing biological attachment pollution on the sensor surface, and the anti-fouling self-cleaning component includes: a micro-mechanical scraper arm and an ultrasonic generator; The micro-mechanical scraper arm periodically cleans the sensor surface attachments to ensure measurement accuracy; The ultrasonic generator uses high-frequency ultrasonic oscillations to inhibit the attachment of microorganisms and prevent errors in collected data caused by long-term immersion in seawater.

9. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The analysis system further includes a three-dimensional visualization engine and a blockchain data storage mechanism. The three-dimensional visualization engine is used to render the marine ranch environment status in real time. The three-dimensional visualization engine includes: a real-time environmental data rendering module and a virtual reality interactive support. The real-time environmental data rendering module is used to generate a temperature field, a dissolved oxygen concentration gradient field, and a fish school movement vector field. The virtual reality interactive support allows users to interactively view the distribution of marine ranch hot zones through VR devices and perform real-time data analysis. The blockchain data notarization mechanism includes: hash value generation algorithm and private chain notarization system. The hash value generation algorithm calculates a unique hash value for each sampling batch to indicate data integrity. The private chain notarization system writes the hash value, timestamp and spatial coordinates into the blockchain node to provide traceable data source capabilities.

10. The multi-parameter in-situ detection data analysis system for marine ranching according to claim 1, characterized in that: The analysis system further deploys mobile detection nodes and multi-pasture comparative analysis modules. The mobile detection nodes are used to expand the monitoring range, including: underwater autonomous robots and hydroacoustic communication networks. The underwater autonomous robots are equipped with a multi-parameter sensor group and cruise along a preset spiral path to cover the monitoring area. The hydroacoustic communication network is used for data synchronization between the autonomous robots and fixed nodes to achieve multi-node collaborative perception and fill in the monitoring blind spots. The multi-pasture comparative analysis module is used to optimize fishery management decisions, including a spatiotemporal data feature extraction unit and a transfer learning optimization strategy. The spatiotemporal data feature extraction unit is used to analyze the spatiotemporal distribution patterns of environmental parameters of multiple pastures. The transfer learning optimization strategy trains models based on data sets of different marine pastures and outputs recommendations on the best feeding and harvesting times to improve pasture production efficiency.

Images