Intelligent monitoring and dynamic management method and system for fish gathering process

Abstract

The present application relates to the technical field of fish gathering dynamic management, and specifically discloses a fish gathering process intelligent monitoring and dynamic management method and system. The present application respectively perceives and collects water body environment data, fish school distribution data, water surface situation data, fish gathering equipment data, meteorological and hydrological data, and fish catch data. The construction of this three-dimensional network eliminates the perception blind area, covers all key elements from air monitoring to underwater detection, from environmental parameters to equipment status, and realizes full-element digitization. Based on AI algorithm, seven types of fish school behavior patterns are identified in real time, and the post-event statistics of fish school behavior are changed into real-time perception. Once the fish school appears abnormal behaviors such as dispersion escape, abnormal floating or diving, etc., the response can be immediately made to minimize the loss. Six types of fusion analysis tags and comprehensive situation scores are monitored in real time to determine whether to trigger comprehensive warning or special warning and determine the warning level, so as to effectively improve the warning response level in the scene of multiple abnormalities.

Description

Technical Field

[0001] This invention relates to the field of fish collection dynamic management technology, and in particular to intelligent monitoring and dynamic management methods and systems for the fish collection process. Background Technology

[0002] Fish aggregating operations are a crucial step in aquatic fishing and fish fry cultivation. Their core objective is to efficiently gather target fish schools to facilitate subsequent fishing or management.

[0003] A dynamic management method for fish collection includes the following steps: A fixed fish collection facility is installed at the tailrace outlet of a hydropower station, introducing the tailrace water into a fish collection pond. The water flows out from a specially arranged double-row, multi-layered fish inlet, forming a three-dimensional water flow covering the entire surface, middle, and bottom water layers, with full vertical coverage and multiple horizontal gradients. Fish are attracted into the fish collection pond by the water flow, collected in fish collection boxes, and then lifted off the water surface and transported to the upstream waters using hoisting equipment to complete the dam crossing transport. Parameters such as the type, quantity, weight, and body length of the collected fish are measured and recorded, acquiring basic data manually or semi-automatically. Based on the collected data, combined with the power station's scheduled power generation flow and downstream water level changes, the key factors affecting the fish collection effect are analyzed, and the optimal fish collection operation scheme under multi-factor conditions is proposed. The collected fish are sorted by a grading grid, temporarily held, and then transferred to the upstream waters to help the fish migrate upstream across the dam.

[0004] Existing methods possess advantages such as mature and reliable technology, proven effectiveness in engineering practice, significantly reduced investment and operating costs, and high equipment reliability. However, they also have some drawbacks: First, the perception dimension is singular, with obvious blind spots in information coverage. Existing methods mainly rely on flow rate and water level as the primary control basis, with limited monitoring data sources. They lack real-time perception capabilities for information such as fish distribution density, fish health status, changes in fish swimming direction, and fish activity levels. The collection and fusion of multi-source data, including water environment data, abnormal water surface conditions, meteorological and hydrological data, and equipment operating parameters, are insufficient, resulting in blind spots in the perception of the fish gathering process and difficulty in timely detection of abnormal fish behavior or sudden changes in water quality. Second, the ability to recognize fish behavior patterns is insufficient. The dynamic management of existing methods is essentially still based on post-event statistical analysis and human experience judgment, lacking real-time identification of complex fish behaviors, leading to decision-making lags. Third, the decision-making capability is weak. The dynamic management of existing methods remains at the level of passive optimization of operating parameters, with a weak ability to autonomously generate multi-dimensional control decision schemes. Summary of the Invention

[0005] This invention addresses the problems in the prior art, such as a single perception dimension, obvious blind spots in information coverage, insufficient ability to recognize fish behavior patterns, and weak decision-making capabilities. It proposes an intelligent monitoring and dynamic management method and system for the fish gathering process.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent monitoring and dynamic management method for fish collection process, comprising the following steps:

[0007] Construct a three-dimensional sensing network encompassing air, water, and bottom to collect data on aquatic environment, fish distribution, water surface conditions, fish collection equipment, meteorological and hydrological data, and fish catch.

[0008] Preprocess the data collected by the sensors;

[0009] Based on AI algorithms, fish behavior patterns are identified, fish gathering efficiency is evaluated, and fish gathering situation is assessed in sequence, and then a comprehensive fish gathering situation assessment report is output.

[0010] Real-time monitoring of six types of integrated analysis tags and comprehensive situation scores determines whether a comprehensive early warning or a special early warning is triggered and determines the early warning level. When multiple conditions are triggered simultaneously, the level is upgraded, and an early warning event containing early warning content and handling suggestions is generated and pushed to the corresponding personnel according to the level.

[0011] Based on the comprehensive analysis report of early warning events and fish gathering situation, the system integrates and analyzes tags and comprehensive situation scores to generate multi-dimensional control decision-making plans and execute them automatically.

[0012] To achieve the above objectives, the present invention provides the following technical solution: an intelligent monitoring and dynamic management system for fish collection process, comprising a three-dimensional perception acquisition module, a preprocessing module, an AI analysis and judgment module, an intelligent early warning management module, an intelligent control module, and an automatic execution module;

[0013] The three-dimensional sensing and acquisition module constructs an air-water-bottom three-dimensional sensing network to sense and acquire water environment data, fish distribution data, water surface situation data, fish collection equipment data, meteorological and hydrological data, and fish catch data, respectively.

[0014] The preprocessing module preprocesses the data collected by the sensor.

[0015] The AI ​​analysis and judgment module is based on AI algorithms and sequentially performs fish school behavior pattern recognition, fish gathering efficiency assessment and fish gathering situation judgment, and then outputs a comprehensive fish gathering situation judgment report.

[0016] The intelligent early warning management module monitors six types of integrated analysis tags and comprehensive situation scores in real time, determines whether a comprehensive early warning or a special early warning is triggered and determines the early warning level. When multiple conditions are triggered simultaneously, the level is upgraded, and an early warning event containing early warning content and handling suggestions is generated and pushed to the corresponding personnel according to the level.

[0017] The intelligent control module generates a multi-dimensional control decision-making scheme based on the comprehensive judgment report of early warning events and fish gathering situation, and by associating and integrating the analysis of tags and comprehensive situation scores.

[0018] The automatic execution module is used to automatically execute multi-dimensional control and decision-making schemes.

[0019] Compared with the prior art, the present invention has at least one of the following beneficial technical effects:

[0020] First, in response to the shortcomings of existing technologies, such as a single sensing dimension and obvious blind spots in information coverage, this invention constructs an air-water-bottom three-dimensional sensing network to sense and collect water environment data, fish distribution data, water surface status data, fish collection equipment data, meteorological and hydrological data, and fish catch data. This three-dimensional sensing network eliminates sensing blind spots and covers all key elements from aerial monitoring to underwater detection, from environmental parameters to equipment status, realizing full-element digitization of the fish collection process.

[0021] Secondly, in response to the shortcomings of existing technologies in recognizing fish behavior patterns, this invention uses AI algorithms to identify seven types of fish behavior patterns in real time: directional swimming, circling in place, dense gathering, dispersing and escaping, deep diving, surface surfacing, and static suspension. This transforms post-event statistics of fish behavior into real-time perception. Once fish exhibit abnormal behaviors such as dispersing and escaping, abnormal surfacing, or diving, the system can respond immediately to minimize losses.

[0022] Third, addressing the issue of weak decision-making capabilities in existing technologies, this invention monitors six types of integrated analysis tags and comprehensive situation scores in real time to determine whether a comprehensive or specific early warning has been triggered and to establish the warning level. When multiple conditions are triggered simultaneously, the warning level is upgraded, generating an early warning event containing warning content and handling suggestions. This event is then pushed to the corresponding personnel based on the different levels. Based on the early warning event and the comprehensive assessment report, equipment control plans, vessel control plans, and fish harvesting decision plans are automatically generated and executed. This combines a dual-track early warning system that integrates comprehensive and specific early warnings with a linkage mechanism that automatically upgrades the warning level when multiple conditions are triggered simultaneously. This achieves adaptive improvement in the level of early warning response in scenarios with multiple anomalies and concurrency. At the same time, multi-dimensional solutions are automatically generated and executed, saving time and effort and achieving a high level of automation. Attached Figure Description

[0023] Figure 1 A step-by-step diagram of intelligent monitoring and dynamic management methods for the fish collection process;

[0024] Figure 2 This is a module connection diagram for the intelligent monitoring and dynamic management system for the fish collection process. Detailed Implementation

[0025] 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 embodiments of the present invention, and not all embodiments. 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.

[0026] like Figure 1 As shown, this embodiment provides an intelligent monitoring and dynamic management method for the fish collection process, including the following steps:

[0027] Construct a three-dimensional sensing network encompassing air, water, and bottom to collect data on aquatic environment, fish distribution, water surface conditions, fish collection equipment, meteorological and hydrological data, and fish catch.

[0028] Preprocess the data collected by the sensors;

[0029] Based on AI algorithms, fish behavior patterns are identified, fish gathering efficiency is evaluated, and fish gathering situation is assessed in sequence, and then a comprehensive fish gathering situation assessment report is output.

[0030] Real-time monitoring of six types of integrated analysis tags and comprehensive situation scores determines whether a comprehensive early warning or a special early warning is triggered and determines the early warning level. When multiple conditions are triggered simultaneously, the level is upgraded, and an early warning event containing early warning content and handling suggestions is generated and pushed to the corresponding personnel according to the level.

[0031] Based on the comprehensive analysis report of early warning events and fish gathering situation, the system integrates and analyzes tags and comprehensive situation scores to generate multi-dimensional control decision-making plans and execute them automatically.

[0032] Furthermore, the aquatic environment data includes water temperature, dissolved oxygen, pH, ammonia nitrogen, nitrite, and turbidity; the fish distribution data includes location, density, swimming direction, geometric shape, activity level, estimated weight, and fish health status; the water surface situation data includes vessel location, operational status, and water surface anomalies; the fish collection equipment data includes start / stop status, operating parameters, operating time, and fault codes; the meteorological and hydrological data includes wind force, wind direction, tides, water flow, and siltation; and the fish catch data includes the number of fish caught, fish species, size, and quality grade.

[0033] Furthermore, the fish health status includes the percentage of healthy fish, the percentage of sub-healthy fish, the percentage of sick fish, and the percentage of dead fish; the abnormalities on the water surface include abnormal foam accumulation, abnormal floating objects, abnormal ripples or whirlpools, water discoloration, and abnormal boat lingering; the specifications include weight specifications and body length specifications, with weight specifications including small, medium, and large sizes, and body length specifications including small, medium, and large sizes; the quality grade refers to the comprehensive quality evaluation and grading of the catch in terms of freshness, body surface integrity, and muscle tissue condition, with specific indicators including eyeballs, gill color, body surface, and muscle.

[0034] Specifically, it should be noted that the equipment used for sensing and collecting aquatic environmental data is a multi-parameter water quality sensor; the equipment used for sensing and collecting fish distribution data is an underwater camera and sonar; the equipment used for sensing and collecting water surface situation data is a high-definition camera and AIS; the equipment used for sensing and collecting fish collection equipment data is an intelligent sensor; the equipment used for sensing and collecting meteorological and hydrological data is a weather station and tidal buoys; and the equipment used for sensing and collecting fish catch data is an electronic weighbridge, an RFID UHF handheld device, an AI visual sorting device, an AI visual quality rating camera, and a K-value rapid detector.

[0035] Furthermore, the preprocessing includes the following steps:

[0036] The data collected by the sensors are uniformly connected to the gateway through their respective communication links, and a UTC timestamp and a unique device identifier are uniformly added at the gateway to form the original data packets;

[0037] Data cleaning is performed on each raw data packet. The cleaned data is divided into valid data, repairable data, and discarded data. Repairable data is converted into valid data after repair.

[0038] Periodic sensor calibration and compensation were performed on the water environment data, and then all valid data were standardized in dimensions and normalized in format.

[0039] Perform time and spatial alignment on all valid data that have undergone dimensional standardization and format normalization.

[0040] By integrating spatiotemporally aligned effective aquatic environment data, fish distribution data, water surface status data, fish collection equipment data, meteorological and hydrological data, and fish catch data, a fused data record of the fish collection process is generated.

[0041] All cleaned, calibrated, aligned, and fused data are categorized and stored in the cloud platform database according to type and purpose, and a multidimensional index is created for all data.

[0042] Specifically, it should be noted that the fusion of spatiotemporally aligned effective aquatic environment data, fish distribution data, water surface condition data, fish collection equipment data, meteorological and hydrological data, and fish catch data includes the following steps:

[0043] By integrating aquatic environment data and fish distribution data, water quality parameters and fish behavior data within the same time window and spatial grid are correlated.

[0044] By integrating water surface situation data and fish collection equipment data, the operational status of vessels within the same time window is correlated with the operating parameters of the fish collection equipment;

[0045] By integrating fish distribution data and catch data, the re-estimation data of fish populations in the same catch batch is correlated with the actual catch quantity, size, and quality grade.

[0046] By integrating meteorological and hydrological data with aquatic environment data, wind force, tide, and water flow data are correlated with changes in water temperature, dissolved oxygen, and turbidity;

[0047] By integrating surface anomaly data and fish distribution data, abnormal surface events are correlated with fish behavior in the areas where the anomalies occur.

[0048] By integrating data from fish collection equipment and meteorological and hydrological data, the operating parameters of the equipment are correlated with wind force, water flow, and wind direction;

[0049] Once the fusion is complete, a record of the fish collection process fusion data will be generated.

[0050] Furthermore, the fish collection process fusion data record includes timestamps, spatial grid codes, key field summaries of various data types, and fusion analysis tags;

[0051] The fusion analysis tags include fish collection efficiency tags, fish school status tags, water quality risk tags, equipment operation tags, operation safety tags, and fish collection timing tags.

[0052] Specifically, the key field summaries for each type of data are as follows: For aquatic environment data, the key field summaries are the latest measured water temperature, dissolved oxygen, pH, ammonia nitrogen, nitrite, and turbidity values ​​within the current time window; for fish distribution data, the key field summaries are the latest status within the current time window, including location coordinates, density, swimming direction, geometric shape, activity level, estimated weight, and fish health status; for water surface situation data, the key field summaries are the latest status within the current time window, including vessel location, operational status, and currently active water surface. Anomaly type coding; for fish collection equipment data, the key field summary is the latest operating status within the current time window, including start / stop status, operating parameters, working duration, and fault codes; for meteorological and hydrological data, the key field summary is the latest measured values ​​within the current time window, including wind force, wind direction, tide level, water flow velocity, water flow direction, and siltation thickness; for fish catch data, the key field summary is a complete record of the most recent fish catch batch, including catch quantity, fish species code, weight specifications, body length specifications, and quality grade. If there is no fish catch operation at present, this field is marked as empty.

[0053] Specifically, it should be noted that for fish distribution data, the location coordinates are based on WGS84 latitude and longitude, and the density unit is tails / m. 3 The swimming direction refers to the azimuth angle, the geometric shape refers to the compactness, the activity level is [0,1], and the weight estimate is in kg. For water surface situation data, the vessel position is based on WGS84 latitude and longitude, the operating status includes in operation, standby, maintenance, and departure, and is marked as none if there are no abnormalities on the water surface. For fish collection equipment data, the start / stop status is represented by 1 and 0, with 1 indicating start and 0 indicating stop. The operating parameters (such as the current value of core parameters such as aerator power kW and feeder speed RPM) are specified, the working time is in h, and the fault code is marked as none if there is no fault. For meteorological and hydrological data, the wind force is in m / s, the wind direction refers to the azimuth angle, the tide level is in m, the water flow velocity is in m / s, the water flow direction refers to the azimuth angle, and the siltation thickness is in cm. For fish catch data, the catch amount is in kg, and the quality grade is excellent, good, medium, or poor.

[0054] Specifically, the following labels should be included: for fish gathering efficiency, there are four categories, including but not limited to high fish gathering efficiency, medium fish gathering efficiency, low fish gathering efficiency, and fish gathering failure; for fish school status, there are five categories, including but not limited to active swarming, static waiting to gather, stress-induced swarming, surfacing and oxygen deficiency, and abnormal mortality; for water quality risk, there are five categories, including but not limited to normal water quality, low dissolved oxygen, ammonia nitrogen warning, water quality deterioration, and water quality critical; for equipment operation, there are four categories, including but not limited to all normal, partial efficiency reduction, key equipment failure, and equipment overload; for operation safety, there are four categories, including but not limited to operation safety, minor risk, moderate risk, and high danger; and for fish harvesting timing, there are four categories, including but not limited to recommending fish harvesting, temporarily not harvesting fish, mandatory fish harvesting, and prohibiting fish harvesting.

[0055] Furthermore, the fish school behavior patterns include directional swimming, circling in place, dense gathering, dispersed escape, deep diving, surface surfacing, and static levitation;

[0056] The fish collection efficiency assessment includes the following steps:

[0057] Based on the correlation modeling of the fish-attracting equipment's operating parameters and the density, swimming direction, and activity of the fish population at the same time, the attraction efficiency of the equipment's operation on the fish population is calculated. The attraction efficiency is calculated based on the rate of change in fish density before and after the equipment's operation, the proportion of the swimming direction that is biased towards the equipment area, and the change in activity.

[0058] The fish collection efficiency label is corrected based on the consistency between the fish collection volume of the most recent batch and the estimated fish population density within the time window before collection.

[0059] Based on the quality grade and its specific indicators in the fish catch data, the water environment data at the time of fish harvesting and the health status of the fish population are correlated to establish a quality degradation attribution model and conduct a fish catch quality-environment correlation analysis. The attribution results are written into the equipment operation label and water quality risk label in the fusion analysis label.

[0060] The fish collection efficiency score is calculated based on the fish behavior pattern, equipment attraction efficiency, consistency verification results of fish collection volume, and fish quality grade.

[0061] The analysis of fish aggregation trends includes the following steps:

[0062] Based on the time series variation trends of key field summaries of water environment data and key field summaries of meteorological and hydrological data, combined with the initial determination of water quality risk labels, dynamic trend extrapolation is performed.

[0063] Based on the key field summary of fish collection equipment data and the associated meteorological and hydrological data, perform equipment operation status diagnosis;

[0064] Based on the summary of key fields of water surface situation data, cross-risk assessment is conducted with meteorological and hydrological data and fish collection equipment data to output the operational safety risk level and description of the main risk sources.

[0065] Decisions on fish harvesting timing are made based on multiple constraints, including fish school status tags, fish catch data, quality grades, water quality risk tags, equipment operation tags, and operational safety tags.

[0066] Specifically, it should be noted that fish behavior pattern recognition includes the following steps:

[0067] Based on WGS84 latitude and longitude coordinates and timestamps, spatiotemporal clustering is performed on the movement trajectory of fish schools within a continuous time window to identify the current behavioral pattern category of the fish school. The behavioral pattern categories include, but are not limited to, seven categories: directional swimming, circling in place, dense gathering, dispersing and escaping, deep diving, surface surfacing, and static suspension. Among them, directional swimming refers to fish schools moving continuously in a single direction with a direction variance less than a set threshold; circling in place refers to fish schools repeatedly circling within a limited space with a position coordinate change radius less than a set threshold; dense gathering refers to fish schools with a density exceeding a set threshold and a geometric compactness higher than a set threshold; dispersing and escaping refers to fish schools with a sudden drop in density and a sharp increase in activity; deep diving refers to fish schools with a continuously increasing water depth value at their position coordinates; surface surfacing refers to fish schools with a continuously decreasing water depth value at their position coordinates; and static suspension refers to fish schools with an activity level below a set threshold and a position coordinate change close to zero.

[0068] Based on the health status fields (healthy percentage, sub-healthy percentage, diseased percentage, and mortality percentage) in the fish distribution data and the associated aquatic environment data (water temperature, dissolved oxygen, pH, ammonia nitrogen, nitrite, and turbidity), a multi-factor health assessment model is used to calculate the comprehensive fish health index. This comprehensive fish health index is a weighted sum of the fish activity score, the fish geometric compactness score, the comprehensive aquatic environment score, and the fish aggregation density score. The fish activity score is obtained by calculating the original activity level using a fish activity response model and then performing Min-Max normalization. The fish geometric compactness score is the geometric compactness... The original compactness of the index calculation is as follows: the comprehensive score of the water environment is obtained by weighted summation of the ratio of the evaluation standard of the corresponding category of water use to the water environment data, and then converted by the percentage grading method; the fish density score is obtained by Min-Max normalization of the ratio of the number of fish in the area to the area of ​​the area; the comprehensive health index ranges from [0,100], and the higher the index, the better the overall health of the fish population; when the comprehensive health index is lower than 60, the ammonia nitrogen warning or water quality deterioration correlation judgment in the water quality risk label is triggered; when the comprehensive health index is lower than 40, the water quality critical label is triggered and the fish status label is linked to determine that the fish collection has failed.

[0069] The fish activity level is coupled with associated aquatic environmental parameters and meteorological and hydrological parameters to establish a fish activity response model. This model is constructed based on the velocity-position update rule of the artificial fish swarm algorithm, and its scalar expression is as follows:

[0070] In the formula, A(t), N, v ix v iy The data are as follows: fish activity level, number of fish in the fish school, velocity component of the i-th fish in the x-direction, and velocity component of the i-th fish in the y-direction. The dominant environmental factors that cause abnormal increases or decreases in fish activity level are identified. When the activity level increases sharply within a short time window and is associated with abnormal ripples or eddies or abnormal foam aggregation in the abnormal water surface data, the fish school is determined to be in a state of disturbance and escape. When the activity level continues to decrease and is associated with low dissolved oxygen or abnormal water temperature, the fish school is determined to be in a state of stress inhibition.

[0071] Based on the geometric compactness index, the expression for the geometric compactness index is as follows: In the formula, C, π, A, and P represent the geometric compactness, pi, area of ​​the fish distribution area, and perimeter of the fish distribution area, respectively. The formula is combined with the fish behavior pattern category for cross-judgment. When the fish are in a dense gathering mode and the compactness is higher than the set threshold, it is judged as an effective fish gathering and can be used as a fish gathering target. When the fish are in a dispersed escape mode and the compactness is lower than the set threshold, it is judged as a dispersed fish group, and the fish gathering efficiency label should be downgraded to low fish gathering efficiency or fish gathering failure. When the fish geometry shows irregular stretching and is associated with abnormal ripples or eddies in the abnormal water surface data, it is judged as a fish group affected by external disturbances, and the disturbance source needs to be further analyzed in conjunction with the water surface situation data.

[0072] Specifically, it should be noted that when the equipment is in the start / stop state of 1 (started) but the fish density does not increase significantly and the swimming direction does not deviate towards the equipment area, the equipment is judged to have failed to attract fish, and the fish gathering efficiency label is downgraded to low fish gathering efficiency; when the equipment is in the start / stop state of 0 (stopped) but the fish still remain densely gathered and their activity level is normal, the fish are judged to have gathered naturally, the fish gathering efficiency label is maintained, but the equipment contribution needs to be marked as zero.

[0073] Specifically, when verifying the consistency between the fish harvested in the most recent batch and the estimated fish weight and density within the time window prior to harvesting, if the harvested fish volume is significantly higher than the theoretical harvestable volume calculated based on the estimated fish density and weight, it is determined that there is a data anomaly or a replenishment of fish. In this case, it is necessary to trace the source by combining the vessel position and operational status in the water surface situation data. If the harvested fish volume is significantly lower than the theoretical harvestable volume, it is determined that there is a loss in the fish collection process. In this case, it is necessary to analyze whether the increased rejection rate is due to poor fish health by combining the quality grade data.

[0074] Specifically, when conducting fish catch quality-environment correlation analysis, if the quality grade is excellent or good, it is necessary to determine whether the corresponding aquatic environmental parameters are within the optimal range (e.g., dissolved oxygen is higher than the set threshold, ammonia nitrogen is lower than the set threshold, and water temperature is within the suitable range for fish species). If so, it is marked as environmentally friendly high-quality fish catch. If the quality grade is medium or poor, the dominant factors leading to the decline in quality are identified, such as sunken eyes associated with a high mortality rate, abnormal gill color associated with low dissolved oxygen, and surface damage associated with abnormal operating parameters of fish collection equipment or mechanical damage caused by loose fish geometry.

[0075] Specifically, the fish collection efficiency score is a weighted sum of the fish school behavior pattern, equipment attraction efficiency, consistency verification results of fish catch, and indicators corresponding to the fish catch quality grade. The typical weights are 0.25, 0.30, 0.20, and 0.25, respectively. For the fish school behavior pattern, the corresponding indicator is the aggregation degree score, which is the ratio of the average occurrence rate of the fish school to the maximum occurrence rate. For the equipment attraction efficiency, the corresponding indicator is the equipment attraction efficiency score, which is the normalized score of the equipment attraction efficiency. For the consistency verification results of fish catch, the corresponding indicator is the consistency score, which is the value of 1 minus the difference between the current coefficient of variation and the maximum coefficient of variation. For the fish catch quality grade, the scoring method of the Yangtze River Aquatic Organism Integrity Index is adopted, using a 0-5 point system, which is then converted to a percentage to obtain the fish catch quality grade score.

[0076] Specifically, the implementation of dynamic trend extrapolation includes: when dissolved oxygen shows a continuous downward trend and increased wind force exacerbates water disturbance, the water quality risk label is upgraded from normal to low dissolved oxygen or ammonia nitrogen warning; when ammonia nitrogen and nitrite both exceed the set threshold and turbidity rises sharply, the water quality risk label is upgraded to deteriorating water quality; when dissolved oxygen is below the critical value and pH deviates significantly from the normal range, the water quality risk label is upgraded to critical water quality, and the fish health status reassessment and fish harvesting timing label prohibition determination are automatically triggered.

[0077] Specifically, the equipment operation status diagnosis includes: when the operating parameters deviate from the rated range and the working time exceeds the set threshold, the equipment is determined to be in a fatigue operation state, and the equipment operation label is downgraded from all normal to partially degraded; when the fault code is not empty and is associated with wind force exceeding the equipment's wind resistance level or water flow velocity exceeding the equipment's design flow rate, the equipment is determined to be in an environmental overload state, and the equipment operation label is downgraded to critical equipment failure or equipment overload; at the same time, based on the historical pattern matching of fault codes and operating parameters, the remaining available time of the equipment is predicted, and if the remaining available time is lower than the set threshold, an early warning is triggered.

[0078] Specifically, the cross-risk assessment includes: when the surface anomaly type is coded as abnormal foam accumulation, abnormal ripples, or whirlpools and is associated with abnormal vessel stay, the operation safety label is upgraded from safe to mild or moderate risk; when the surface anomaly type is coded as water discoloration and is associated with a water quality hazard label and wind exceeding a set threshold, the operation safety label is upgraded to high risk; when the vessel's operation status is maintenance or departure and the fish collection equipment's start / stop status is 1, it is determined that there is a risk of unattended equipment operation, and the operation safety label is upgraded to moderate risk; based on the above determinations, the operation safety risk level and description of the main risk sources are output.

[0079] Specifically, the decision-making process for multi-constraint fish harvesting timing includes: when the fish population status label indicates high fish gathering efficiency, the fish population is in a densely clustered mode, the expected quality level is excellent or good, the water quality risk label indicates normal water quality, the equipment operation label indicates all equipment is normal, and the operation safety label indicates safe operation, the fish harvesting timing label is determined to be "recommended to harvest fish," and the optimal fish harvesting time window is output; when the fish population status label indicates medium fish gathering efficiency, or the fish population is in a trend of dispersing and escaping, or the expected quality level is medium, or the equipment operation label indicates partial deterioration, the fish harvesting timing label is determined to be "temporarily not harvest fish," and the suggested waiting conditions and time are output; when the fish population status label indicates low fish gathering efficiency but the fish population still has a certain density, the quality level is acceptable, and there is no safety risk, the fish harvesting timing label is determined to be "mandatory fish harvesting," and the minimum acceptable fish harvesting threshold is output; when the water quality risk label indicates critical water quality, the operation safety label indicates high danger, or the mortality rate in the fish population's health status exceeds a set threshold, the fish harvesting timing label is determined to be "prohibited to harvest fish," and the specific reasons for prohibiting fish harvesting and the conditions for resuming fish harvesting are output.

[0080] Furthermore, the comprehensive fish gathering situation assessment report includes a situation overview, key alarms, and operational optimization suggestions. The situation overview is a comprehensive situation score of the current fish gathering process, which is calculated by weighting four dimensions: fish gathering efficiency score, water quality risk score, equipment operation score, and operational safety score. The fish gathering efficiency score is directly obtained from the assessment steps. The water quality risk score is calculated based on measured data of water temperature, dissolved oxygen, pH, ammonia nitrogen, nitrite, and turbidity. The equipment operation score is calculated based on a comprehensive calculation of equipment failure rate, sound and light equipment working rate, net integrity rate, energy consumption efficiency, and sensor online rate. The operational safety score is calculated based on a comprehensive calculation of wind and wave level, visibility, personnel operation compliance rate, emergency equipment integrity rate, and channel conflict risk. The key alarms are alarm information automatically generated when any fusion analysis tag reaches a high-risk level. The alarm information includes alarm type, alarm level, alarm content description, related data summary, and suggested handling measures. The operational optimization suggestions are operational optimization suggestions generated based on the results of fish school behavior pattern recognition and fish gathering efficiency assessment.

[0081] Specifically, the comprehensive situation score ranges from [0, 100]. The current overall situation level is indicated by four colors: red, orange, yellow, and green. Green corresponds to a comprehensive score ≥ 80 with no high-risk labels; yellow corresponds to a comprehensive score ≥ 60 with at least one medium-risk label; orange corresponds to a comprehensive score ≥ 40 with at least one high-risk label; and red corresponds to a comprehensive score < 40 or the presence of a highly dangerous label. Alarm types include water quality alarms, equipment alarms, safety alarms, and fish harvesting alarms; alarm levels include urgent, important, and general. Operation optimization suggestions include: when fish are densely clustered but fish collection efficiency is low, it is recommended to adjust the operating parameters of the fish collection equipment (such as increasing the power of the aerator, adjusting the speed of the feeder, and changing the placement of the equipment); when fish are disturbed and escaping, it is recommended to suspend the operation of the fish collection equipment and reduce the intensity of boat operations; when the water quality risk label is raised, it is recommended to adjust the operating area or reduce the fish collection frequency; when the equipment operation label is lowered, it is recommended to start the backup equipment or arrange maintenance; all suggestions are output in priority order, and the priority is determined based on the expected improvement in the fish collection efficiency score.

[0082] Specifically, based on AI algorithms, after sequentially identifying fish school behavior patterns, assessing fish collection efficiency, and judging fish collection trends, all intermediate results and final decision outputs of this AI analysis, along with pre-processed fish collection process fusion data records, are stored in the cloud platform database according to timestamps and spatial grid encoding as historical training samples. The fish school behavior recognition model, fish collection efficiency assessment model, water quality risk prediction model, equipment fault diagnosis model, and fish collection timing decision model are periodically iterated and trained and their parameters optimized based on the accumulated historical samples, continuously improving the accuracy and timeliness of AI analysis.

[0083] Furthermore, the comprehensive early warning levels, from severe to mild, are classified as Level 1, Level 2, Level 3, and Level 4. The specific early warnings include water quality-specific early warnings, equipment-specific early warnings, safety-specific early warnings, fish harvesting-specific early warnings, and fish school-specific early warnings. The water quality-specific early warning level is determined based on the water quality risk label level; the equipment-specific early warning level is determined based on the equipment operation label level; the safety-specific early warning level is determined based on the operational safety label level; and the fish harvesting-specific early warning level is determined based on the fish harvesting timing label level. The early warning content includes the early warning number, early warning type, early warning level, early warning content description, and associated data summary.

[0084] Specifically, it should be noted that when determining whether an early warning is triggered, the fish collection efficiency label, fish school status label, water quality risk label, equipment operation label, operation safety label, fish collection timing label, and comprehensive situation score are tracked and monitored in real time. Simultaneously, the results of fish school behavior pattern recognition, fish school comprehensive health index, activity score, geometric morphology compactness score, water environment comprehensive score, density score, equipment attraction efficiency, fish collection consistency verification results, quality degradation attribution results, equipment remaining usable time prediction, operation safety risk level and main risk source description, and multi-constraint fish collection timing decision results are also monitored.

[0085] Specifically, it should be noted that the determination of triggering a special early warning includes the following steps:

[0086] A special water quality warning will be triggered when dissolved oxygen is below the critical value, or when ammonia nitrogen and nitrite exceed the set threshold at the same time and turbidity rises sharply, or when pH deviates significantly from the normal range.

[0087] When the operating parameters deviate from the rated range and the working time exceeds the set threshold, or the fault code is not empty and is associated with wind force exceeding the equipment's wind resistance level or water flow speed exceeding the equipment's design flow rate, or the remaining available time is less than the set threshold, a special warning for the equipment will be triggered.

[0088] When the water surface anomaly type is coded as abnormal foam accumulation, abnormal ripples, or whirlpools and is associated with abnormal vessel stay, or when the water surface anomaly type is coded as water surface discoloration and is associated with water quality hazard tag and wind exceeding the set threshold, or when the vessel's operation status is maintenance or departure and the fish collection equipment's start / stop status is 1, a special safety warning is triggered.

[0089] When the amount of fish collected is significantly higher than the theoretical amount that can be collected based on the estimated values ​​of fish density and weight, or when the amount of fish collected is significantly lower than the theoretical amount that can be collected and the quality grade continues to decline, or when the fish are in a state of being disturbed and escaping, a special warning for fish collection will be triggered.

[0090] When the overall health index of the fish population is below 60, or the fish behavior pattern is determined to be disturbed and escaped with a sharp increase in activity, or the fish population is in a dispersed escape mode and the density is below the set threshold, a special warning for the fish population will be triggered.

[0091] Specifically, it should be noted that the comprehensive early warning trigger determination includes the following steps:

[0092] If the overall situation score is less than the first threshold of the overall situation, the label is "high danger", the water quality risk label is "critical water quality", the operation safety label is "high danger", the mortality rate of fish in the health status exceeds the set threshold, or the equipment operation label is "critical equipment failure" or "equipment overload" and the remaining available time is less than the set threshold, the overall warning level is level one.

[0093] If the comprehensive situation score is greater than or equal to the first threshold of the comprehensive situation and less than the second threshold of the comprehensive situation, and there is at least one high-risk label, the water quality risk label is water quality deterioration, the operation safety label is moderate risk, the equipment operation label is partial efficiency reduction and the fault code is not empty, and the fish school status label is fish collection failure, then the comprehensive warning level is level two.

[0094] If the comprehensive situation score is greater than or equal to the second threshold of the comprehensive situation and less than the third threshold of the comprehensive situation, and there is at least one medium risk label, the water quality risk label is low dissolved oxygen or ammonia nitrogen warning, the operation safety label is mild risk, and the fish harvesting time label is temporarily not to harvest fish and the duration exceeds the set threshold, the comprehensive warning level is level three.

[0095] If the overall situation score is greater than or equal to the third threshold of the overall situation but any label shows a downward trend, it is considered a trend warning, i.e., the overall warning level is level four.

[0096] Specifically, it should be noted that when multiple conditions are triggered simultaneously, the warning level will be upgraded as follows: when any one condition triggers a Level 1 warning, the overall warning level will be directly set to Level 1; when multiple Level 2 warnings are triggered simultaneously, the overall warning level will be upgraded to Level 1; when Level 2 and Level 3 warnings are triggered simultaneously, the overall warning level will be set to Level 2; and when Level 3 and Level 4 warnings are triggered simultaneously, the overall warning level will be set to Level 3.

[0097] Specifically, the differentiated push notifications to relevant personnel based on risk level include: For Level 1 warnings, immediate and simultaneous notifications are sent to the operation supervisor, safety management personnel, and platform monitoring center through all channels, including audible and visual alarms, SMS, and APP push notifications. If it is a water quality emergency warning, a fish harvesting timing tag prohibiting harvesting is automatically triggered, suggesting adjustments to the operation area and upgrading the water quality risk tag to water quality emergency. If it is a critical equipment failure warning, it automatically suggests starting backup equipment and arranging maintenance, downgrading the equipment operation tag to critical equipment failure or equipment overload. If it is a high-risk safety warning, it automatically suggests suspending all operations and evacuating personnel, upgrading the operation safety tag to high-risk. If it is a fish mortality rate exceeding a threshold warning, a fish collection failure tag is automatically triggered, linking to quality level attribution analysis. For Level 2 warnings, notifications are sent to the operation supervisor and safety management personnel via APP push notifications and platform monitoring center pop-ups. For Level 3 warnings, notifications are sent to on-duty personnel via platform monitoring center prompts and APP messages. For Level 4 warnings, notifications are only recorded in the platform monitoring center logs and are not proactively pushed, for management personnel to review periodically and analyze trends.

[0098] Specifically, it should be noted that dynamic adaptive adjustment of warning rules is supported, including: threshold reduction, that is, when a certain type of warning is triggered continuously within a set time window and is not effectively alleviated after handling, the trigger threshold of the warning type is automatically reduced and the warning level is increased to improve response sensitivity; trend warning, that is, when a certain type of warning has not been triggered for a long time but the related indicators show a slow deterioration trend (such as dissolved oxygen continuously and slowly decreasing but not yet reaching the low dissolved oxygen threshold), a trend warning (level four) is automatically generated and early intervention is recommended; threshold revert, that is, when the situation returns to normal after handling and remains stable for more than a set time window after the warning is triggered, the threshold of the warning type is automatically reverted to the baseline value to avoid oversensitivity.

[0099] Specifically, it should be noted that each early warning event undergoes a closed-loop management process throughout its entire lifecycle, including: full-process recording, which records the early warning trigger time, warning content, target audience, response actions, handling results, and final situation recovery; early warning cancellation, which automatically generates an early warning cancellation notification and pushes it to the original recipients when the early warning conditions are lifted, the fusion analysis label that triggered the early warning returns to the normal level, and the comprehensive situation score rises back to the corresponding color range; sample accumulation, which stores the triggering conditions, handling process, and final effect of each early warning event as samples in the cloud platform database, forming a training dataset together with historical AI analysis samples; and iterative optimization, which periodically iterates and optimizes the early warning threshold model, level determination rules, linkage response strategies, and trend early warning algorithms based on accumulated early warning samples, continuously reducing false alarm and false negative rates, and improving the accuracy and timeliness of early warnings.

[0100] Furthermore, the multi-dimensional control decision-making scheme includes equipment control scheme, vessel control scheme, and fish harvesting decision-making scheme;

[0101] The equipment control scheme is generated based on equipment operation tags, equipment operation tags in the fusion analysis tags, equipment-specific early warnings in early warning events, and operation optimization suggestions in the comprehensive fish gathering situation analysis report. It includes attraction efficiency optimization control, fatigue operation intervention control, fault emergency handling control, disturbance escape response control, and quality protection control.

[0102] The vessel control scheme is generated based on operational safety tags, key field summaries of surface situation data, safety-specific early warnings in early warning events, and operational optimization suggestions in the comprehensive fish gathering situation analysis report. It includes operational intensity control, operational status switching control, operational area adjustment control, and emergency avoidance control.

[0103] The fish harvesting decision-making scheme is generated based on fish harvesting timing tags, key field summaries of fish catch data, quality grades, fish school status tags, and fish harvesting-specific early warnings in early warning events. It includes suggested fish harvesting decisions, decisions to temporarily refrain from fish harvesting, mandatory fish harvesting decisions, and decisions to prohibit fish harvesting.

[0104] Specifically, it should be noted that the attraction efficiency optimization and control includes: when the fish are in a dense gathering mode but the fish gathering efficiency is low, an equipment parameter optimization scheme is generated based on the fish behavior pattern recognition results and the equipment attraction efficiency score. The equipment parameter optimization scheme is selected from at least one of increasing the power of the aerator, adjusting the speed of the feeder, and changing the placement position of the equipment. The adjustment range of each parameter is determined proportionally based on the difference between the equipment attraction efficiency score and the fish gathering efficiency score. The upper limit of the adjustment range shall not exceed 20% of the rated parameters of the equipment, so as to avoid drastic changes in parameters causing secondary disturbance to the fish.

[0105] Specifically, fatigue operation intervention and control include: when the equipment operation tag is partially degraded and the operating parameters deviate from the rated range and the working time exceeds the set threshold, a frequency reduction operation plan or a rotation operation plan is generated based on the predicted value of the remaining available time of the equipment. The frequency reduction operation plan lowers the equipment operating parameters to a conservative value within the rated range. The rotation operation plan switches the current equipment start-stop status from 1 to 0 and starts the backup equipment. The startup of the backup equipment is based on the list of all normal backup equipment in the equipment operation tag, selected in order of priority.

[0106] Specifically, it should be noted that the emergency response and control of the fault includes: when the equipment operation tag is a critical equipment fault or equipment overload and the remaining available time is less than a set threshold, based on the historical pattern matching results of the fault code and the operating parameters, a fault isolation scheme and a backup equipment takeover scheme are generated. The fault isolation scheme switches the start-stop status of the faulty equipment from 1 to 0 and marks the fault code. The backup equipment takeover scheme automatically starts the backup equipment with the highest priority and adjusts its operating parameters to the default value that is consistent with the most recent normal operating parameters of the faulty equipment.

[0107] Specifically, it should be noted that the disturbance and escape response control includes: when the fish are in a state of disturbance and escape and their activity level increases sharply, a device silencing scheme is generated based on the analysis results of the fish activity response model. The device silencing scheme switches the start / stop status of the fish gathering equipment from 1 to 0, and reduces the power of the aerator and the speed of the feeder in the operating parameters to the minimum maintenance value, so as to reduce the continuous disturbance of the fish to the fish. The equipment can be gradually resumed only after the activity level of the fish returns to the normal range and the fish behavior pattern changes to dense gathering or directional swimming.

[0108] Specifically, it should be noted that the quality protection and control includes: when the quality grade is medium or poor and the quality degradation attribution result is associated with abnormal operating parameters of the fish collection equipment, a device parameter correction scheme is generated. The device parameter correction scheme is based on the dominant factors identified by the quality degradation attribution model and adjusts the operating parameters of the equipment that cause mechanical damage in a targeted manner, such as reducing the net release and take-up speed to reduce damage to the fish body surface and adjusting the power of the aerator to avoid abnormal gill color caused by drastic fluctuations in dissolved oxygen.

[0109] Specifically, for equipment control schemes with warning levels of 1 and 2, the system automatically generates control commands and sends them to the control unit of the corresponding fish-gathering equipment via the communication link for execution. For equipment control schemes with warning levels of 3 and 4, the system generates control suggestions and pushes them to the operation supervisor, who then confirms them before execution via the control link. After equipment control is executed, the system collects feedback data on the equipment's start / stop status and operating parameters in real time, compares it with the control target value, and automatically performs secondary fine-tuning if the deviation exceeds the set tolerance until the equipment operating parameters converge to the target range.

[0110] Specifically, it should be noted that the operation intensity control includes: when the fish are in a disturbed and escaped state or the operation safety label is mild risk or above, the system generates a plan to reduce the operation intensity. The plan includes one or more of the following: reducing the speed of the boat, increasing the safe distance between the boat and the fish collection equipment area, and reducing the frequency of boat operations. The adjustment range of the safe distance is determined based on a comprehensive calculation of the water surface anomaly type code and the wind force value.

[0111] Specifically, the operation status switching control includes: when the vessel's operation status is maintenance or departure but the fish collection equipment's start / stop status is 1, the system determines that there is a risk of unattended equipment operation and generates a vessel return to port plan or a remote takeover plan. The vessel return to port plan sends a return command to the vessel and switches the operation status from maintenance or departure to operation. The remote takeover plan switches the fish collection equipment's start / stop status from 1 to 0 and switches to remote monitoring mode when the vessel cannot return to port in time.

[0112] Specifically, the adjustment and control of the operating area includes: when the water quality risk label is upgraded to low dissolved oxygen or above, or the water quality risk label is critical water quality, the system generates an operating area relocation plan based on the spatial distribution of key field summaries of water environment data and the water flow direction in meteorological and hydrological data. The plan recommends that the vessel be moved to an adjacent spatial grid area with a water quality risk label of normal water quality. The moving distance is determined based on a comprehensive calculation of the spatial gradient of the water quality risk label and the water flow velocity, ensuring that the vessel reaches the area with excellent water quality within an acceptable time window.

[0113] Specifically, it should be noted that the emergency evacuation and control includes: when the operation safety label is highly dangerous or the water surface anomaly type is coded as water surface discoloration and is associated with a water quality hazard label and wind force exceeding a set threshold, the system generates an emergency evacuation plan. The plan includes immediately stopping all operations, evacuating vessels to a safe anchorage, evacuating personnel to a safe area, and automatically sending an emergency notification to the platform monitoring center and relevant rescue agencies.

[0114] Specifically, for vessel control plans with a warning level of 1, the system automatically generates navigation control commands and issues them for execution via the shipboard communication module, while simultaneously sending a vessel status change notification to the platform monitoring center. For vessel control plans with a warning level of 2, the system automatically issues navigation suggestions and simultaneously sends a confirmation request to the operation supervisor; if no confirmation is received within the specified time, the system executes automatically. For vessel control plans with warning levels of 3 and 4, the system generates navigation suggestions and pushes them to the operation supervisor for manual decision-making and execution. After the vessel control is executed, the system tracks the vessel's position and operational status in real time based on surface situation data to confirm the effective implementation of the control measures.

[0115] Specifically, it should be noted that the recommended fish harvesting decision includes: when the fish harvesting time tag is recommended to harvest fish, the fish school status tag is high fish gathering efficiency and the fish school is in a dense gathering mode and the expected quality level is excellent or good, the water quality risk tag is normal water quality, the equipment operation tag is all normal and the operation safety tag is safe operation, the system generates the optimal fish harvesting plan. The plan includes the optimal fish harvesting time window, the recommended number of fish harvesting batches and the expected fish harvesting volume range. The optimal fish harvesting time window is determined based on the intersection of the peak activity in the fish school behavior pattern and the optimal interval of the water environment data.

[0116] Specifically, it should be noted that the decision to temporarily not collect fish includes: when the fish collection timing tag is "temporarily not collecting fish" and the fish status tag is "fish collection efficiency is medium" or "fish are in a trend of dispersing and escaping" or "quality level is expected to be medium" or "equipment operation tag is partially degraded", the system generates a waiting fish collection plan. The plan includes the suggested waiting conditions (such as fish density recovering to above the set threshold, quality level expected to improve to "good" or above, and equipment operation tag recovering to "all normal") and the suggested maximum waiting time. If the waiting conditions are not met after the maximum time, the fish collection timing tag will be automatically downgraded to "forced fish collection".

[0117] Specifically, it should be noted that the forced fish harvesting decision includes: when the fish harvesting time tag is forced fish harvesting and the fish group status tag is low fish collection efficiency but the fish group still has a certain density and the quality level is acceptable and there is no safety risk, the system generates a minimum fish harvesting plan. The plan includes a minimum acceptable fish harvesting quantity threshold and a recommended fish harvesting batch quantity. The minimum acceptable fish harvesting quantity threshold is determined by multiplying the theoretical harvestable fish quantity calculated based on the current fish group repopulation estimate and density by a set reduction factor.

[0118] Specifically, the decision to prohibit fish harvesting includes: when the fish harvesting timing label is "prohibited", the water quality risk label is "critical", the operation safety label is "highly dangerous", or the mortality rate of the fish population exceeds a set threshold, the system generates a prohibition plan. The plan includes the specific reasons for prohibiting fish harvesting (one or more of the following: critical water quality, safety hazard, or excessive mortality rate of the fish population) and the conditions for resuming fish harvesting (such as dissolved oxygen rising above the critical value, wind speed dropping below the set threshold, or mortality rate dropping below the set threshold). When all the conditions for resuming fish harvesting are met and remain stable for more than a set time window, the system automatically adjusts the fish harvesting timing label from "prohibited" to "temporarily not harvesting" or "suggesting fish harvesting".

[0119] Specifically, for decisions where the fish harvesting timing is tagged as "recommended," the system automatically generates a fish harvesting operation instruction and pushes it to the operation manager, while simultaneously initiating real-time statistics on the harvested quantity. For decisions where the fish harvesting timing is tagged as "mandatory," the system automatically generates a fish harvesting operation instruction and simultaneously initiates a rapid quality inspection process. For decisions where the fish harvesting timing is tagged as "prohibited," the system automatically locks fish harvesting operation permissions and sends a prohibition notice to all relevant personnel. For decisions where the fish harvesting timing is tagged as "temporarily not harvesting," the system continuously monitors the fulfillment of waiting conditions and automatically upgrades the fish harvesting timing tag when the conditions are met. During the fish harvesting process, the system evaluates the harvesting effect in real time based on the consistency verification results of the harvested quantity. If the harvested quantity is significantly lower than the theoretically achievable harvested quantity, it automatically triggers quality level attribution analysis and generates a loss report.

[0120] Specifically, it should be noted that when there are constraint conflicts between the equipment control plan, the vessel control plan, and the fish harvesting decision plan, the system performs multi-constraint collaborative control and conflict resolution, including: the safety priority principle, that is, when the operational safety label is moderate risk or above, regardless of other label statuses, the emergency avoidance measures in the vessel control plan and the prohibition of fish harvesting in the fish harvesting decision plan are executed first, and the equipment silencing measures in the equipment control plan are executed simultaneously; the quality priority principle, that is, when the quality level is excellent or good and the correlation analysis determines that it is environmentally friendly high-quality fish harvesting, under the premise that the safety risk is controllable, the recommended fish harvesting decision in the fish harvesting decision plan takes precedence over the decision not to harvest fish temporarily, and the equipment control plan is based on maintaining the current optimal operating parameters; and the efficiency priority principle. In other words, when the fish school status tag indicates high fish gathering efficiency and the fish harvesting timing tag indicates recommended fish harvesting with no safety risks, the attraction efficiency optimization measures in the equipment control plan and the optimal fish harvesting plan in the fish harvesting decision plan are executed in tandem, while the vessel control plan is based on coordinating with the fish harvesting operation; the conflict resolution mechanism is as follows: when the equipment control plan requires increasing the aerator power but the water quality risk tag indicates critical water quality and dissolved oxygen is already in the supersaturated risk range, the system automatically reduces the aerator power adjustment range to zero and triggers a review of the water quality risk tag; when the vessel control plan requires moving to an area with good water quality but the fish harvesting timing tag indicates recommended fish harvesting and the fish school is about to leave the current area, the system prioritizes the fish harvesting decision based on the fish school behavior pattern prediction results and then performs area adjustment after the fish harvesting is completed.

[0121] Specifically, it should be noted that the execution results of each control command are fed back and evaluated in real time, including the following steps: real-time collection of feedback data such as equipment start-up and shutdown status, operating parameters, vessel location, operation status, and fish catch, which are compared with the control target value to generate an execution deviation report. If the deviation exceeds the set tolerance, secondary control is triggered or the warning level is upgraded.

[0122] Based on the changes in the integrated analysis labels and the changes in the comprehensive situation score after the implementation of the control measures, the control effect is evaluated. The effect evaluation includes five indicators: changes in equipment attraction efficiency, changes in fish school aggregation degree, changes in fish catch consistency verification results, changes in quality grade, and changes in the comprehensive situation score. If the improvement in the comprehensive situation score after the control measures reaches the set threshold, the control plan is marked as an effective plan and stored in the plan library. If the comprehensive situation score does not improve or decreases after the control measures, it is marked as an invalid plan and a review of the control plan is triggered.

[0123] When the early warning conditions are lifted after the control measures are implemented, that is, when the fusion analysis label that triggered the early warning returns to the normal level and the comprehensive situation score rises back to the corresponding color range, the system automatically generates a situation recovery notification and pushes it to the original recipient. At the same time, the fish harvesting timing label is changed from prohibiting fish harvesting or temporarily not harvesting fish to recommending fish harvesting or resuming fish harvesting operations.

[0124] The entire process of each control command, including its trigger time, control content, target, execution result, effect evaluation, and situation recovery, is recorded and stored in the cloud platform database using timestamps and spatial grid codes.

[0125] It should be noted that the undefined preset values, weights, and thresholds are selected based on actual needs and are not subject to specific value restrictions here.

[0126] like Figure 2 As shown, this embodiment provides an intelligent monitoring and dynamic management system for the fish collection process, including a three-dimensional perception acquisition module, a preprocessing module, an AI analysis and judgment module, an intelligent early warning management module, an intelligent control module, and an automatic execution module;

[0127] The three-dimensional sensing and acquisition module constructs an air-water-bottom three-dimensional sensing network to sense and acquire water environment data, fish distribution data, water surface situation data, fish collection equipment data, meteorological and hydrological data, and fish catch data, respectively.

[0128] The preprocessing module preprocesses the data collected by the sensor.

[0129] The AI ​​analysis and judgment module is based on AI algorithms and sequentially performs fish school behavior pattern recognition, fish gathering efficiency assessment and fish gathering situation judgment, and then outputs a comprehensive fish gathering situation judgment report.

[0130] The intelligent early warning management module monitors six types of integrated analysis tags and comprehensive situation scores in real time, determines whether a comprehensive early warning or a special early warning is triggered and determines the early warning level. When multiple conditions are triggered simultaneously, the level is upgraded, and an early warning event containing early warning content and handling suggestions is generated and pushed to the corresponding personnel according to the level.

[0131] The intelligent control module generates a multi-dimensional control decision-making scheme based on the comprehensive judgment report of early warning events and fish gathering situation, and by associating and integrating the analysis of tags and comprehensive situation scores.

[0132] The automatic execution module is used to automatically execute multi-dimensional control and decision-making schemes.

[0133] Specifically, it should be noted that, in addition to the modules mentioned above, the intelligent monitoring and dynamic management system for the fish collection process also includes a database, which stores data information from all modules in the system. The three-dimensional perception acquisition module, preprocessing module, AI analysis and judgment module, intelligent early warning management module, intelligent control module, automatic execution module, and three-dimensional perception acquisition module are sequentially connected in a closed loop. The preprocessing module is connected to the intelligent early warning management module, the AI ​​analysis and judgment module is connected to the intelligent control module, and the automatic execution module is connected to the AI ​​analysis and judgment module, intelligent early warning management module, and intelligent control module. All modules in the system are connected to the database.

[0134] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring and dynamic management of the fish collection process, characterized in that, Includes the following steps: Construct a three-dimensional sensing network encompassing air, water, and bottom to collect data on aquatic environment, fish distribution, water surface conditions, fish collection equipment, meteorological and hydrological data, and fish catch. Preprocess the data collected by the sensors; Based on AI algorithms, fish behavior patterns are identified, fish gathering efficiency is evaluated, and fish gathering situation is assessed in sequence, and then a comprehensive fish gathering situation assessment report is output. Real-time monitoring of six types of integrated analysis tags and comprehensive situation scores determines whether a comprehensive early warning or a special early warning is triggered and determines the early warning level. When multiple conditions are triggered simultaneously, the level is upgraded, and an early warning event containing early warning content and handling suggestions is generated and pushed to the corresponding personnel according to the level. Based on the comprehensive analysis report of early warning events and fish gathering situation, the system integrates and analyzes tags and comprehensive situation scores to generate multi-dimensional control decision-making plans and execute them automatically.

2. The intelligent monitoring and dynamic management method for the fish collection process according to claim 1, characterized in that, The aquatic environment data includes water temperature, dissolved oxygen, pH, ammonia nitrogen, nitrite, and turbidity; the fish distribution data includes location, density, swimming direction, geometric shape, activity level, estimated weight, and fish health status; the water surface situation data includes vessel location, operational status, and water surface anomalies; the fish collection equipment data includes start / stop status, operating parameters, operating time, and fault codes; the meteorological and hydrological data includes wind force, wind direction, tides, water flow, and siltation; and the fish catch data includes the number of fish caught, fish species, size, and quality grade.

3. The intelligent monitoring and dynamic management method for the fish collection process according to claim 2, characterized in that, The fish health status is defined as the percentage of healthy fish, the percentage of sub-healthy fish, the percentage of sick fish, and the percentage of dead fish. Water surface anomalies include abnormal foam accumulation, abnormal floating objects, abnormal ripples or whirlpools, water discoloration, and abnormal boat lingering. Sizes include weight and length; weight sizes are categorized as small, medium, and large, and length sizes are also categorized as small, medium, and large. Quality grade refers to the comprehensive quality evaluation and grading of the catch based on freshness, body surface integrity, and muscle tissue condition, with specific indicators including eye color, gill color, body surface, and muscle.

4. The intelligent monitoring and dynamic management method for the fish collection process according to claim 3, characterized in that, The preprocessing includes the following steps: The data collected by the sensors are uniformly connected to the gateway through their respective communication links, and a UTC timestamp and a unique device identifier are uniformly added at the gateway to form the original data packets; Data cleaning is performed on each raw data packet. The cleaned data is divided into valid data, repairable data, and discarded data. Repairable data is converted into valid data after repair. Periodic sensor calibration and compensation were performed on the water environment data, and then all valid data were standardized in dimensions and normalized in format. Perform time and spatial alignment on all valid data that have undergone dimensional standardization and format normalization. By integrating spatiotemporally aligned effective aquatic environment data, fish distribution data, water surface status data, fish collection equipment data, meteorological and hydrological data, and fish catch data, a fused data record of the fish collection process is generated. All cleaned, calibrated, aligned, and fused data are categorized and stored in the cloud platform database according to type and purpose, and a multidimensional index is created for all data.

5. The intelligent monitoring and dynamic management method for the fish collection process according to claim 4, characterized in that, The fish collection process fusion data record includes timestamps, spatial grid codes, key field summaries of various data types, and fusion analysis tags; The fusion analysis tags include fish collection efficiency tags, fish school status tags, water quality risk tags, equipment operation tags, operation safety tags, and fish collection timing tags.

6. The intelligent monitoring and dynamic management method for the fish collection process according to claim 5, characterized in that, The fish's behavioral patterns include directional swimming, circling in place, dense gathering, dispersed escape, deep diving, surface surfacing, and static suspending. The fish collection efficiency assessment includes the following steps: Based on the correlation modeling of the fish-attracting equipment's operating parameters and the density, swimming direction, and activity of the fish population at the same time, the attraction efficiency of the equipment's operation on the fish population is calculated. The attraction efficiency is calculated based on the rate of change in fish density before and after the equipment's operation, the proportion of the swimming direction that is biased towards the equipment area, and the change in activity. The fish collection efficiency label is corrected based on the consistency between the fish collection volume of the most recent batch and the estimated fish population density within the time window before collection. Based on the quality grade and its specific indicators in the fish catch data, the water environment data at the time of fish harvesting and the health status of the fish population are correlated to establish a quality degradation attribution model and conduct a fish catch quality-environment correlation analysis. The attribution results are written into the equipment operation label and water quality risk label in the fusion analysis label. The fish collection efficiency score is calculated based on the fish behavior pattern, equipment attraction efficiency, consistency verification results of fish collection volume, and fish quality grade. The analysis of fish aggregation trends includes the following steps: Based on the time series variation trends of key field summaries of water environment data and key field summaries of meteorological and hydrological data, combined with the initial determination of water quality risk labels, dynamic trend extrapolation is performed. Based on the key field summary of fish collection equipment data and the associated meteorological and hydrological data, perform equipment operation status diagnosis; Based on the summary of key fields of water surface situation data, cross-risk assessment is conducted with meteorological and hydrological data and fish collection equipment data to output the operational safety risk level and description of the main risk sources. Decisions on fish harvesting timing are made based on multiple constraints, including fish school status tags, fish catch data, quality grades, water quality risk tags, equipment operation tags, and operational safety tags.

7. The intelligent monitoring and dynamic management method for the fish collection process according to claim 6, characterized in that, The comprehensive fish gathering situation assessment report includes a situation overview, key alarms, and operational optimization suggestions. The situation overview is a comprehensive situation score of the current fish gathering process, calculated by weighting four dimensions: fish gathering efficiency score, water quality risk score, equipment operation score, and operational safety score. The fish gathering efficiency score is directly obtained from the assessment steps. The water quality risk score is calculated based on measured data of water temperature, dissolved oxygen, pH, ammonia nitrogen, nitrite, and turbidity. The equipment operation score is calculated based on a comprehensive assessment of equipment failure rate, sound and light equipment working rate, net integrity rate, energy consumption efficiency, and sensor online rate. The operational safety score is calculated based on a comprehensive assessment of wind and wave level, visibility, personnel operation compliance rate, emergency equipment availability rate, and channel conflict risk. Key alarms are automatically generated when any fusion analysis tag reaches a high-risk level. The alarm information includes alarm type, alarm level, alarm content description, related data summary, and suggested handling measures. The operational optimization suggestions are generated based on the results of fish school behavior pattern recognition and fish gathering efficiency assessment.

8. The intelligent monitoring and dynamic management method for the fish collection process according to claim 7, characterized in that, The comprehensive early warning levels, from most severe to least severe, are Level 1, Level 2, Level 3, and Level 4; the specific early warnings include water quality-specific early warnings, equipment-specific early warnings, safety-specific early warnings, fish harvesting-specific early warnings, and fish school-specific early warnings. The water quality-specific early warning level is determined based on the water quality risk label level, the equipment-specific early warning level is determined based on the equipment operation label level, the safety-specific early warning level is determined based on the operation safety label level, and the fish harvesting-specific early warning level is determined based on the fish harvesting timing label level; the early warning content includes the early warning number, early warning type, early warning level, early warning content description, and related data summary.

9. The intelligent monitoring and dynamic management method for the fish collection process according to claim 8, characterized in that, The multi-dimensional control and decision-making scheme includes equipment control scheme, vessel control scheme, and fish harvesting decision-making scheme. The equipment control scheme is generated based on equipment operation tags, equipment operation tags in the fusion analysis tags, equipment-specific early warnings in early warning events, and operation optimization suggestions in the comprehensive fish gathering situation analysis report. It includes attraction efficiency optimization control, fatigue operation intervention control, fault emergency handling control, disturbance escape response control, and quality protection control. The vessel control scheme is generated based on operational safety tags, key field summaries of surface situation data, safety-specific early warnings in early warning events, and operational optimization suggestions in the comprehensive fish gathering situation analysis report. It includes operational intensity control, operational status switching control, operational area adjustment control, and emergency avoidance control. The fish harvesting decision-making scheme is generated based on fish harvesting timing tags, key field summaries of fish catch data, quality grades, fish school status tags, and fish harvesting-specific early warnings in early warning events. It includes suggested fish harvesting decisions, decisions to temporarily refrain from fish harvesting, mandatory fish harvesting decisions, and decisions to prohibit fish harvesting.

10. An intelligent monitoring and dynamic management system for the fish collection process, employing the intelligent monitoring and dynamic management method for the fish collection process as described in any one of claims 1-9, characterized in that, It includes a stereo perception acquisition module, a preprocessing module, an AI analysis and judgment module, an intelligent early warning management module, an intelligent control module, and an automatic execution module; The three-dimensional sensing and acquisition module constructs an air-water-bottom three-dimensional sensing network to sense and acquire water environment data, fish distribution data, water surface situation data, fish collection equipment data, meteorological and hydrological data, and fish catch data, respectively. The preprocessing module preprocesses the data collected by the sensor. The AI ​​analysis and judgment module is based on AI algorithms and sequentially performs fish school behavior pattern recognition, fish gathering efficiency assessment and fish gathering situation judgment, and then outputs a comprehensive fish gathering situation judgment report. The intelligent early warning management module monitors six types of integrated analysis tags and comprehensive situation scores in real time, determines whether a comprehensive early warning or a special early warning is triggered and determines the early warning level. When multiple conditions are triggered simultaneously, the level is upgraded, and an early warning event containing early warning content and handling suggestions is generated and pushed to the corresponding personnel according to the level. The intelligent control module generates a multi-dimensional control decision-making scheme based on the comprehensive judgment report of early warning events and fish gathering situation, and by associating and integrating the analysis of tags and comprehensive situation scores. The automatic execution module is used to automatically execute multi-dimensional control and decision-making schemes.

Images