Intelligent digital aquaculture management system and method

Abstract

The application discloses an intelligent digital aquaculture management system and method, and particularly relates to the technical field of digital agriculture management, and is used for solving the problem that the long-term coordination of local dynamic adjustment and overall production plan is insufficient in the existing breeding management, which causes the production process to gradually deviate from the expected target; the breeding production plan, production state data and management operation data are acquired and continuously monitored; the response logic between the management operation and the production state is analyzed to determine the deviation trend; when the deviation occurs, the historical and current causal relationship networks are constructed and compared to identify the change of the key causal relationship; meanwhile, the uniformity change mode of the individual size in the breeding population is analyzed; the specific management adjustment instruction is generated and executed according to the above analysis results; finally, the breeding production plan is dynamically updated according to the instruction execution effect; the closed-loop adaptive management of the breeding process is realized, and the predictability and stability of the production result are ensured.

Description

Technical Field

[0001] This invention relates to the field of digital agricultural management technology, and in particular to an intelligent digital aquaculture management system and method. Background Technology

[0002] In the existing digital management of aquaculture, in order to achieve the predetermined production goals, a complete aquaculture production plan is usually formulated before the start of aquaculture, based on the species and conditions. This aquaculture production plan includes phased operation arrangements and the expected production status. In the actual execution process, the system monitors environmental and aquaculture information through sensors, and makes frequent local adjustments and optimizations to operations such as feeding and environmental control based on such real-time data in order to cope with short-term fluctuations and strive to maintain the stability of the current production status.

[0003] However, the existing methods have shortcomings: there is a lack of systematic coordination between the preset fixed plan and the continuous dynamic optimization scheduling. Although frequent local adjustments are intended to optimize in real time, their long-term cumulative effect will cause the actual state changes of the breeding process to gradually deviate from the expected evolution path in the plan. Since the existing technology mainly focuses on comparing data with the plan at a single point or in the short term, it lacks an effective identification and evaluation mechanism for such long-term, trend-based overall deviations. This leads to the initial plan gradually becoming ineffective during execution, and the basis for subsequent optimization decisions is less correlated with the overall production goals, making it difficult to guarantee the stability and predictability of production results. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing an intelligent digital aquaculture management system and method.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] A smart digital aquaculture management method includes:

[0007] S1. Obtain the preset breeding production plan and continuously obtain production status data and management operation data during the actual breeding process;

[0008] S2. Analyze the response logic between the actual management operation data and the corresponding production status data. When the response logic is inconsistent with the preset causal logic relationship, it is determined that a deviation trend has occurred.

[0009] S3. When a deviation trend is detected, a causal relationship network is constructed and compared based on historical and current production status data using a causal relationship analysis algorithm to identify causal relationships that have changed significantly.

[0010] S4. Analyze the pattern of uniformity of individual size within the breeding population when deviations from the trend occur;

[0011] S5. Based on the significant changes in causal relationships and uniformity change patterns, generate and execute management adjustment instructions for subsequent stages;

[0012] S6. Update the aquaculture production plan based on the execution results of the management adjustment instructions.

[0013] Furthermore, S1 includes:

[0014] Obtain an aquaculture production plan that includes a phased sequence of expected production states;

[0015] Continuous monitoring of aquaculture environment parameters and biological parameters of the aquaculture population through sensors to generate production status data;

[0016] Continuously record the types, times, and intensity of feeding, aeration, and water exchange operations performed during the aquaculture process to form management operation data.

[0017] Furthermore, production status data is correlated with the expected production status sequence in the aquaculture production plan, and management operation data is correlated with production status data over time.

[0018] Furthermore, S2 includes:

[0019] Based on the temporal relationship between management operation data and production status data, a mapping relationship between management operation events and subsequent production status change events is established.

[0020] Extract the actual response characteristic parameters between management operation data and production status data from the mapping relationship;

[0021] The actual response feature parameters are compared with the expected response feature parameters in the preset causal logic relationship;

[0022] When the difference between the actual response characteristic parameters and the expected response characteristic parameters exceeds a preset difference threshold, the response logic is determined to be inconsistent with the preset causal logic relationship.

[0023] Furthermore, the pre-defined causal logical relationship is established in the following way: analyze the management operation data and production status data during the historical normal period, extract the stable temporal correlation and quantitative response characteristics between management operation events and subsequent production status change events, and establish the pre-defined causal logical relationship based on the extracted temporal correlation and quantitative response characteristics.

[0024] Furthermore, S3 includes:

[0025] Based on the time point for determining the deviation trend from the production status data, historical normal period production status data and current deviation period production status data are divided.

[0026] Based on the causal relationship analysis algorithm, the production status data of the historical normal period and the production status data of the current deviation period are processed respectively to obtain the historical causal relationship network reflecting the causal influence relationship between production status variables in the historical period, and the current causal relationship network reflecting the causal influence relationship between production status variables in the current period.

[0027] The current causal relationship network is compared with the historical causal relationship network. The specific comparison includes: identifying causal relationships that exist in both causal relationship networks but whose strength changes of the causal edges connecting the same pair of production state variables exceed a preset strength threshold; and identifying causal relationships that exist only in one of the causal relationship networks.

[0028] Causal relationships that meet the comparison criteria are identified as causal relationships that have undergone significant changes.

[0029] Furthermore, S4 includes:

[0030] Obtain sampled data on individual size within the aquaculture population at different time points before and after the occurrence of deviation from the trend;

[0031] Based on the sampled data of individual size within the breeding population, a uniformity index sequence characterizing the dispersion of individual size within the population is calculated and formed.

[0032] Analyze the changing trends of the evenness index sequence before and after the occurrence of deviation trends;

[0033] Based on the changing trends of the evenness index sequence, evenness change patterns that characterize the intensification of population differentiation, the maintenance of stability, or the tendency towards evenness can be identified.

[0034] Furthermore, S5 includes:

[0035] Based on the production state variables involved in the significant changes in causal relationships, identify potential key control points;

[0036] Based on the population differentiation status indicated by the uniformity change pattern, management adjustment instructions including operation type, adjustment timing and adjustment intensity are formulated for key control links;

[0037] The management adjustment instructions are output to the aquaculture management system to drive the execution mechanism to perform the corresponding feeding, oxygenation or water change operations in the subsequent stages.

[0038] Furthermore, S6 includes:

[0039] After management adjustment instructions are executed in subsequent stages, production status data for those stages is obtained to evaluate the effectiveness of the instructions.

[0040] The analysis examines the degree to which the execution of instructions corrects deviations from the trend and the improvement in uniformity variation patterns.

[0041] Based on the comprehensive analysis of the degree of correction and the improvement, adaptive adjustments are made to the expected production status sequence of the subsequent stages in the aquaculture production plan;

[0042] The adjusted expected production status sequence is updated to the aquaculture production plan, forming an updated aquaculture production plan to guide subsequent aquaculture processes.

[0043] On the other hand, the present invention provides an intelligent digital aquaculture management system, comprising:

[0044] The data acquisition module is used to acquire the preset breeding production plan and continuously acquire production status data and management operation data executed in the actual breeding process;

[0045] The trend determination module is used to analyze the response logic between the actual management operation data and the corresponding production status data. When the response logic is inconsistent with the preset causal logic relationship, it is determined that a deviation from the trend has occurred.

[0046] The causal identification module is used to identify significantly changed causal relationships when a deviation trend is detected, based on a causal relationship analysis algorithm that constructs and compares a causal relationship network using historical and current production status data.

[0047] The pattern analysis module is used to analyze the pattern of uniformity changes in individual size within a breeding population when deviations from the trend occur.

[0048] The instruction generation module is used to synthesize the causal relationships and uniformity change patterns that have undergone significant changes, generate and execute management and adjustment instructions for subsequent stages;

[0049] The plan update module is used to update the aquaculture production plan based on the execution results of management adjustment instructions.

[0050] The beneficial effects of this invention are:

[0051] 1. By establishing a complete closed-loop process from plan acquisition, trend deviation identification, causal analysis, population evenness diagnosis to instruction generation and plan update, effective control over long-term and trend-based deviations in the aquaculture process is achieved. The system systematically combines real-time local adjustments based on sensor data with in-depth state diagnosis based on causal networks and population structure analysis. It not only focuses on the immediate optimization of single-point states, but also identifies the long-term changing trends of the response logic between actual management operations and production states, as well as the structural changes in causal relationships within the aquaculture system. Thus, on top of frequent local operations, a macro-management perspective that can gain insight into the overall evolution direction of the process and changes in internal driving factors is constructed.

[0052] 2. The system transforms the preset static production plan into an adaptive guide that can dynamically adapt to the actual breeding process. By comprehensively analyzing the deviation of response logic, changes in causal relationships, and population evenness patterns, the adjustment instructions generated by the system have a stronger overall goal orientation. Based on the feedback of the instruction execution effect, the expected value of the subsequent production plan is adaptively revised, so that the optimization decisions in the breeding process are always strongly correlated with the final production goal. This effectively avoids the accumulation of overall deviation caused by local optimization, significantly improves the predictability and stability of breeding production results, and realizes the qualitative change of digital breeding management from passive response to proactive guidance. Attached Figure Description

[0053] Figure 1 This is a flowchart of an intelligent digital aquaculture management method according to the present invention;

[0054] Figure 2 This is a schematic diagram of the structure of an intelligent digital aquaculture management system according to the present invention. Detailed Implementation

[0055] 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.

[0056] Example 1

[0057] Figure 1 This invention provides an intelligent digital aquaculture management method, comprising:

[0058] S1. Obtain the preset breeding production plan and continuously obtain production status data and management operation data during the actual breeding process;

[0059] S2. Analyze the response logic between the actual management operation data and the corresponding production status data. When the response logic is inconsistent with the preset causal logic relationship, it is determined that a deviation trend has occurred.

[0060] S3. When a deviation trend is detected, a causal relationship network is constructed and compared based on historical and current production status data using a causal relationship analysis algorithm to identify causal relationships that have changed significantly.

[0061] S4. Analyze the pattern of uniformity of individual size within the breeding population when deviations from the trend occur;

[0062] S5. Based on the significant changes in causal relationships and uniformity change patterns, generate and execute management adjustment instructions for subsequent stages;

[0063] S6. Update the aquaculture production plan based on the execution results of the management adjustment instructions.

[0064] S1. Obtain the preset aquaculture production plan and continuously acquire production status data and management operation data during the actual aquaculture process. Specifically, this is implemented as follows:

[0065] To obtain the aquaculture production plan, a pre-set aquaculture production plan text or database file is first read from the storage device. This aquaculture production plan is a detailed planning document divided into stages based on a timeline, containing multiple time nodes and a corresponding expected production state sequence for each time node. The expected production state sequence is a set of quantitative indicators of the aquaculture state that are pre-defined in the aquaculture plan and expected to be achieved at that time node. For example, for an 80-day Litopenaeus vannamei aquaculture cycle, the aquaculture production plan may include multiple key stages such as day 10, day 30, day 50, and harvest day. Each stage corresponds to an expected production state sequence, which typically includes biological parameters such as expected average body weight (e.g., 8 to 10 grams per shrimp), expected survival rate (e.g., 90% to 95%), and expected feed conversion ratio (e.g., 1.2 to 1.5), as well as environmental parameters such as expected water temperature range (e.g., 28 to 30 degrees Celsius), expected dissolved oxygen concentration range (e.g., 5 to 6 mg / L), and expected ammonia nitrogen concentration range (e.g., below 0.2 mg / L). The range or target values ​​of these indicators are pre-defined based on the conventional growth model of the aquaculture species and historical aquaculture experience in specific ponds, and are explicitly written into the planning document. Obtaining this plan means loading these structured stage and target data from persistent storage media and parsing them into internal data tables that can be called by subsequent steps, such as a mapping table with stage days as keys and a dictionary of status indicators and target value ranges as values.

[0066] To generate production status data, continuous monitoring is required during the actual aquaculture process using sensor arrays deployed in the aquaculture ponds. Monitoring of aquaculture environmental parameters is achieved by immersing water quality sensors in the aquaculture water. These sensors include temperature sensors, dissolved oxygen sensors, pH sensors, and ammonia nitrogen sensors. The temperature sensor uses a platinum resistance element, collecting water temperature data at a fixed frequency of once every 5 minutes, recording the data in degrees Celsius, and transmitting it to a data logger via a wired communication line. The dissolved oxygen sensor uses a probe based on the fluorescence principle, also collecting data at a frequency of once every 5 minutes and recording the dissolved oxygen concentration in milligrams per liter. The pH and ammonia nitrogen sensors operate at similar frequencies, recording pH values ​​and non-toxic ammonia nitrogen concentrations, respectively. All sensors are calibrated with standard solutions before being put into use to ensure data accuracy. Acquisition of biological parameters of the aquaculture population is accomplished through periodic sampling and automated measurement equipment. For example, every 7 days, at least 50 randomly selected cultured individuals from the pond are weighed using an electronic scale with an accuracy of 0.1 grams. The sum of all weights is divided by the number of individuals to calculate the average weight, which is recorded in grams. Survival rate is assessed by observing the size of the feeding group during regular feeding periods, combined with periodic sampling counts, and estimated as a percentage. These raw readings, directly generated by sensors and measuring devices, are timestamped at a data logger according to a unified time base and arranged in chronological order to constitute production status data. This production status data is a time-series dataset where each data point includes a timestamp and a set of environmental and biological parameter values ​​collected at that moment.

[0067] To generate management operation data, continuous and standardized recording is required every time the aquaculture process is manually or automatically intervened, through an operation log system. For feeding operations, the record should include the operation type as "feeding," the execution time (year, month, day, hour, minute), e.g., October 26, 2025, 09:05, and the operation intensity as the specific weight of feed fed, e.g., 5.5 kg. For aeration operations, the record should include the operation type as "aeration," the execution time (year, month, day, hour, minute), and the operation intensity as a combination of the power and duration of the aeration equipment used in that operation, e.g., running two 1.5 kW aerators continuously for 120 minutes. For water change operations, the record should include the operation type as "water change," the execution time, and the operation intensity as the percentage of the pond's total water volume changed, e.g., changing 20%. These records are manually entered and confirmed by operators on handheld terminals equipped with input interfaces, or automatically generated by automated control equipment when executing preset commands. All record entries are automatically appended with a timestamp synchronized with the production status data when they are generated, and are sorted according to the execution timestamp to form a management operation data sequence.

[0068] To establish data correlation, production status data is linked to the expected production status sequence in the aquaculture production plan during the data processing stage. Specifically, each data point in the production status data sequence is assigned to a specific time period defined in the aquaculture production plan based on the number of days of aquaculture corresponding to its timestamp. For example, if the timestamp of a collected production status data point corresponds to the 32nd day after the start of aquaculture, this data point is associated with the planned period from day 30 to day 50. The actual monitored value of this data point is then aligned and preliminarily compared with the corresponding indicator range in the expected production status sequence set for that period in the plan. Simultaneously, management operation data is correlated with production status data in time through a unified time base and comparative analysis. All sensor data records and management operation log entries are synchronized using the same clock source. The specific correlation logic involves, for a management operation data entry, using its execution timestamp as the starting point, searching and extracting relevant production status data points within a preset time window after this time point in the time-sorted production status data sequence. For example, by treating a specific oxygenation operation as an event, and querying all dissolved oxygen sensor readings within 120 minutes of the operation's commencement, a direct temporal correlation is established between these production status data points and the oxygenation operation. This time-window-based correlation provides a clear and traceable data foundation for subsequent analysis of the response logic between operational events and status change events.

[0069] S2. Analyze the response logic between the actual management operation data and the corresponding production status data. When the response logic is inconsistent with the preset causal logic relationship, it is determined that a deviation trend has occurred. The specific implementation is as follows:

[0070] To establish a mapping relationship between management operation events and subsequent production status change events, the management operation data sequence and the production status data sequence are first defined and extracted as events. A management operation event corresponds to each independent management operation data record in step S1. For example, a feeding operation record constitutes a feeding event, which includes the event type attribute "feeding," the event occurrence time attribute "execution time in the record," and the event intensity attribute "operation intensity in the record." Identification of production status change events requires detection from the continuous production status data sequence obtained in step S1. For a specific production status parameter, such as dissolved oxygen concentration, a minimum change threshold is set. This minimum change threshold is determined based on the analysis of the parameter's normal fluctuation range during historical stable periods. For example, it is determined by calculating the standard deviation of the difference between adjacent time points of the parameter in historical data, and setting the minimum change threshold to one standard deviation, such as 0.5 mg per liter. A minimum duration, such as 30 minutes, is also set to exclude instantaneous noise interference. When the system scans the production status data sequence, if it finds that the dissolved oxygen concentration value has been continuously rising or falling from a point in time, and the cumulative magnitude of its continuous change exceeds the set minimum change threshold, and the duration of this change exceeds the minimum duration threshold, then this segment of continuously changing data is identified as a production status change event, and its start time, end time, peak time, direction of change, and cumulative change magnitude are recorded. After defining and identifying the two types of events, the mapping relationship is established. For each management operation event, a backward search response time window is defined based on its occurrence time. The length of this window is set according to the operation type and experience; for example, for oxygenation operations, the response time window is set to within 180 minutes after the start. Then, within this time window, all production status change events whose start time falls within this window are searched. If the start time of a production status change event is after the occurrence time of a management operation event, and the time difference between the two is greater than a minimum response delay, such as 2 minutes, then this pair of management operation events and production status change events is established as a candidate mapping relationship. This step provides a clear set of event pairs for subsequent analysis.

[0071] After obtaining the set of candidate mapping event pairs, actual response feature parameters are extracted from each pair to quantitatively describe the characteristics of the response. These actual response feature parameters aim to objectively measure the quantitative relationship between the input of management operations and the resulting changes in production status. For a mapping pair, the extracted feature parameters typically include response time characteristics and response intensity characteristics. The response time characteristic refers to the time interval, measured in minutes, from the occurrence of the management operation event to the peak concentration reached by the associated production status change event. The calculation method for the response intensity characteristic varies depending on the nature of the operation and status parameters. For example, for a pairing of an oxygenation operation event and a dissolved oxygen concentration increase event, the response intensity characteristic can be quantified as the ratio between the cumulative increase in dissolved oxygen concentration during the production status change event (in milligrams per liter) and the energy consumption value obtained by multiplying the operating power by the duration of the oxygenation operation event (in kilowatt-hours). This ratio reflects the dissolved oxygen increase effect per unit of energy consumption. By traversing all candidate mapping event pairs and calculating response time and response intensity features according to the above rules, a set of actual response feature parameters is obtained, where each parameter corresponds to a quantitative description of the response to a specific operation and state change.

[0072] Comparing the actual response characteristic parameters with the expected response characteristic parameters in the pre-defined causal logic relationship is the core step in determining whether the response logic is consistent. The pre-defined causal logic relationship is established during historical normal aquaculture phases by analyzing a large amount of historical management operation data and production status data from normal periods. Using the same mapping relationship as described above, rules and characteristic parameter extraction methods are established, and the stable range of response characteristic parameters between different types of operations and specific state changes is statistically calculated. For example, regarding the relationship between aeration operations and increased dissolved oxygen concentration, by analyzing 100 aeration events that occurred under normal conditions in multiple past aquaculture cycles and their successfully mapped dissolved oxygen change events, the average historical response time characteristic was calculated to be 60 minutes, with a standard deviation of 15 minutes; the average historical response intensity characteristic was an increase of 1.2 mg / L in dissolved oxygen per kilowatt-hour, with a standard deviation of 0.2 mg / L per kilowatt-hour. Based on these statistical data, the expected response characteristic parameters for the oxygenation-dissolved oxygen relationship in the preset causal logic are set as a numerical range, typically defined as the historical statistical average plus or minus two standard deviations. That is, the expected response time characteristic range is 30 to 90 minutes, and the expected response intensity characteristic range is 0.8 mg / L to 1.6 mg / L. The comparison process involves comparing a currently extracted actual response characteristic parameter value with the expected characteristic parameter ranges for the corresponding operation type and state parameter type in the preset causal logic. For example, after the current oxygenation operation, the extracted actual response time characteristic is 120 minutes, and the actual response intensity characteristic is 0.5 mg / L.

[0073] When the difference between the actual response characteristic parameters and the expected response characteristic parameters exceeds a preset difference threshold, the response logic is determined to be inconsistent with the preset causal logic. The preset difference threshold is a pre-set critical standard used to quantify whether the difference reaches a significant deviation level. For scenarios comparing with an expected value range, the difference is usually quantified as the relative degree to which the actual value deviates from the expected range boundary. The preset difference threshold can be set as an allowed percentage deviation, such as 20%. The specific judgment logic is as follows: First, check whether the actual value falls within the expected range. If the actual value falls within the expected range, it is determined that there is no difference. If the actual value falls outside the expected range, calculate the difference between it and the nearest range boundary value, and divide the absolute value of the difference by the absolute value of the boundary value to obtain the deviation percentage. For example, for response intensity characteristics, the expected range is 0.8 mg / L to 1.6 mg / L. If the actual value is 0.5 mg / L, which is lower than the lower boundary value of 0.8, the difference is 0.3, and the deviation percentage is 0.3 / 0.8 × 100% = 37.5%. The calculated percentage deviation is compared to a preset difference threshold. Since 37.5% is greater than 20%, the actual response intensity characteristic is determined to differ from the expected value beyond the preset difference threshold. The preset difference threshold is set based on a further assessment of the volatility of characteristic parameters in historical normal data. For example, it can be set as an empirical proportion of the fluctuation range of historical characteristic parameters during a stable period, or a critical value distinguishing normal fluctuations from abnormal deviations can be determined through statistical methods. When, in a specific event analysis, one or more actual response characteristic parameter values ​​deviate from their corresponding expected range beyond the limits specified by the preset difference threshold, it is comprehensively determined that the response logic for this management operation is inconsistent with the preset causal logic relationship established based on historical normal data. This indicates that a potential operational deviation trend has been identified.

[0074] S3. When a deviation trend is detected, a causal relationship network is constructed and compared based on historical and current production status data using a causal relationship analysis algorithm to identify causal relationships that have changed significantly. Specifically, this is implemented as follows:

[0075] To distinguish between historical normal-period production status data and current deviation-period production status data from production status data, it is first necessary to determine the deviation trend judgment time point for this division. This time point is determined in step S2 above by analyzing the response logic between management operation data and production status data. It is the moment when the judgment response logic is inconsistent with the preset causal logic relationship, marking the starting point when the system considers the operating status to have begun to deviate from expectations. Based on this judgment time point, the production status data continuously collected and stored in chronological order is divided. Specifically, all production status data within a continuous and stable time period before the judgment time point is defined as historical normal-period production status data. The length of this time period must be sufficient to reflect the normal steady-state mode of the system. Its setting can be based on historical experience or the stage of the breeding cycle, for example, selecting data from 60 consecutive days before the judgment time point, ensuring that no deviation trend instances appear within these 60 days as judged by step S2. All production status data after the judgment time point until the latest moment at the time of analysis in this step is defined as current deviation-period production status data. The current analysis time can be a fixed time point after the judgment time, such as the 7th day after the judgment, or it can be the latest data collection time during continuous real-time analysis. Through this division, two time-continuous datasets are obtained, which respectively carry the complete sequence of production state parameters under historical normal steady state and current possible deviation state, providing a clear data foundation for subsequent construction and comparison of causal relationship networks at different times.

[0076] To construct a causal relationship network by processing two datasets using a causal relationship analysis algorithm, the specific structure of the causal relationship network in this method needs to be defined first. A causal relationship network is a graph model composed of nodes and directed edges. Each node corresponds to a production state variable defined and continuously monitored in step S1, such as water temperature, dissolved oxygen concentration, ammonia nitrogen concentration, pH, and average body weight. Each directed edge connects two nodes, with the direction indicating the direction of the causal influence. For example, an edge from the water temperature node to the dissolved oxygen concentration node indicates that a change in water temperature is considered a cause of a change in dissolved oxygen concentration. Each edge is accompanied by a causal strength value, used to quantify the strength of the influence of the causal variable on the outcome variable. To construct a historical causal relationship network from historical normal-period production state data, a causal relationship discovery algorithm suitable for time series data and known in the field is required. For example, a method based on the Granger causality test can be used. Before processing, the historical normal-period production status data needs to be preprocessed. This includes handling potential missing values; for example, for data collected at 5-minute intervals, missing values ​​at a certain time point can be filled with the arithmetic mean of the parameter values ​​at the two preceding and following time points. It also includes stabilizing the time series of each variable, for example, by calculating the first difference between values ​​at adjacent time points to eliminate trends in the data. After preprocessing, for any pair of production status variables A and B, a maximum time lag order L is set. The value of L is related to the data collection frequency; for example, for data collected every 5 minutes, L can be set to 6, representing the consideration of the lag effect within the past 30 minutes. The core of the algorithm lies in constructing two linear regression models to predict the current value of variable B: the first is a restricted model, using only the L past historical values ​​of variable B as predictors; the second is a complete model, using both the L past historical values ​​of variable B and the L past historical values ​​of variable A as predictors. By comparing the sum of squared predicted residuals of the two models, an F-statistic is calculated, resulting in a probability value P. If the p-value is less than a preset significance level threshold, such as 0.05, then statistically, the historical values ​​of variable A are considered to contain unique information predicting the future values ​​of variable B, meaning variable A is considered a Granger cause of variable B. In this case, the causal strength can be quantified by calculating the increase in the explanatory power of the model, for example, by using the difference in the squared coefficient of determination R between the complete model and the restricted model. By iterating through all pairs of production state variables and performing the aforementioned bidirectional test on each pair, the causal relationships that pass the significance test are recorded as directed edges, along with the calculated causal strength values, ultimately forming a historical causal relationship network. Using the exact same algorithm, parameter settings, and preprocessing steps, the current deviation period production state data is processed to obtain a structurally comparable current causal relationship network, reflecting the causal influence relationships between various production state variables in the current period.

[0077] The purpose of comparing the current causal relationship network with historical causal relationship networks is to identify causal relationships that have changed significantly in the two different periods. The comparison is performed at the edge set level of the two networks. First, identify causal relationships that exist simultaneously in both networks. This requires that there exists a directed edge in both networks connecting two identical production state variable nodes, and the direction of the edge is consistent. For such an edge that exists in both periods, its associated causal strength value needs to be compared. Calculate the absolute difference between the causal strength value of this edge in the current causal relationship network and the causal strength value of the corresponding edge in the historical causal relationship network. Compare this difference with a preset strength threshold. This preset strength threshold is set based on the stability analysis of historical normal-period causal relationship networks or the volatility analysis of historical data. For example, it can be set to 10% of the average causal strength value of all edges in the historical causal relationship network, or twice the standard deviation of the historical causal strength value in multiple sampling calculations. If the calculated strength difference is greater than the preset strength threshold, the strength of the causal relationship is considered to have changed significantly, meeting one of the comparison conditions. Second, identify causal relationships that exist only in one of the causal relationship networks. This includes directed edges that exist only in the current causal relationship network but not in the historical causal relationship network, as well as directed edges that existed historically but have disappeared in the current causal relationship network. The appearance or disappearance of these edges directly signifies a change in the causal relationship structure, and therefore directly meets the comparison criteria. When performing this comparison, it is necessary to ensure that the two networks are constructed on the same set of production state variables, i.e., the same set of nodes, to guarantee the comparability of the network structures.

[0078] The final output of this step is to identify causal relationships that meet the above comparison criteria as those that have undergone significant changes. Specifically, edges that exist in both networks and whose differences in causal strength exceed a preset strength threshold, as well as edges that appear in only one network, are summarized into a single result set. Each identified causal relationship that has undergone significant changes is recorded with its associated pair of production state variables, causal direction, causal strength value in the historical causal relationship network, causal strength value in the current causal relationship network, and change type. Change types can be categorized as significantly increased strength, significantly decreased strength, newly emerging causal relationships, and disappearance of causal relationships. This identification result directly reveals the specific and quantifiable structural changes in the driving relationships between key state variables within the aquaculture system from the historical normal period to the current deviation period, providing crucial information about the evolution of causal logic within the system for subsequent comprehensive analysis.

[0079] S4. Analyze the pattern of uniformity in individual size within the aquaculture population when deviations from the trend occur. The specific implementation is as follows:

[0080] To obtain sampling data on individual size within the aquaculture population at different time points before and after the occurrence of the deviation trend, the sampling time series must first be planned based on the deviation trend determination time point identified in step S2. This deviation trend determination time point is a specific date and time, marking the starting point when the system determines that the aquaculture process has begun to deviate from the expected logic. Based on this time point, several time points representing historical normal states are selected before it, such as the 30th day, 15th day, 7th day before this time point, and the day of the determination time point, as historical sampling time points. Several time points for tracking and monitoring are selected after it, such as the 3rd day, 7th day, 14th day, and 21st day after the determination, as current and subsequent sampling time points. At each determined sampling time point, on-site sampling of the aquaculture population is performed. Sampling must follow the principles of random sampling in statistics to ensure that the sample can unbiasedly represent the entire population in the aquaculture pond. In practice, for example, a standard hand-thrown net is used to randomly collect individuals from five different areas (east, west, south, north, and center) of the aquaculture pond. All individuals collected from these five collections are combined to form the total sample set for that sampling, requiring a minimum of 50 individuals. Obtaining individual size data within the aquaculture group involves measuring the specific dimensions of each individual in the sample set, typically using weight as a quantitative indicator. For example, a calibrated electronic scale with an accuracy of 0.1 grams is used to weigh each individual in the sample, accurately recording the value in grams. Ultimately, each sampling time point will generate a dataset containing the specific weight values ​​of all individuals measured in that sampling. Organizing and storing these datasets obtained from all selected time points strictly according to their sampling time order constitutes the complete original dataset for subsequent analysis of evenness changes.

[0081] Based on sampling data of individual size within a breeding population, a series of evenness indicators characterizing the dispersion of individual size within the population is calculated and formed. This requires selecting specific statistical indicators and clarifying their calculation process. In this embodiment, the evenness indicator chosen is the coefficient of variation, defined as the ratio of the standard deviation of the sample to the sample mean. It is a dimensionless measure of relative dispersion, suitable for comparing evenness differences between populations of different average sizes. For a dataset containing N individual weight values ​​obtained at a specific sampling time point, the specific steps for calculating the evenness indicator at that time point are as follows: First, calculate the arithmetic mean of the dataset, i.e., add up the weight values ​​of all individuals in the dataset, obtain the sum, and then divide it by the total number of individuals N. Second, calculate the standard deviation of the dataset, first calculating the difference between each individual weight value and the mean obtained in the first step, squaring each difference, then summing all the squared values, dividing the sum by the total number of individuals N minus one, and finally taking the square root of the quotient. Third, calculate the coefficient of variation, i.e., divide the standard deviation calculated in the second step by the mean calculated in the first step, usually multiplying the result by 100%, and expressing the evenness indicator value at that time point as a percentage. For example, if a sample of 50 shrimp yields a calculated average weight of 10.5 grams and a standard deviation of 2.1 grams, then the evenness index for this sample is 2.1 / 10.5 × 100% = 20%. This three-step calculation process is repeated for all planned sampling time points before and after the deviation from the trend, yielding a corresponding evenness index percentage value for each time point. These percentage values ​​are then arranged in chronological order according to their corresponding sampling time points, forming an ordered sequence called the evenness index sequence. This sequence, indexed by time, clearly presents the dynamic evolution of the size dispersion within the aquaculture population over time.

[0082] Analyzing the trend of the evenness index sequence before and after the occurrence of a deviation requires quantitative analysis of the trend direction and rate of change. First, the position of the deviation trend determination point is clearly marked on the time axis. This point divides the complete evenness index sequence into a historical normal period and a current deviation period. The core of trend analysis lies in quantitatively characterizing the overall direction of the sequence values' movement within the two periods before and after the determination point. A concrete and implementable method is to perform linear fitting on the evenness index values ​​of the historical normal period. For example, four historical sampling points before the determination point and their corresponding four evenness index values ​​are selected. With time as the independent variable and the index value as the dependent variable, a straight line is fitted using the least squares method. The slope of this fitted line quantifies the historical trend of the evenness index before the deviation occurred; a positive slope indicates an upward trend, a negative slope indicates a downward trend, and a slope close to zero indicates stability. Using the same fitting method, a straight line is also fitted to the sampling point data of the current deviation period to obtain the slope characterizing the current trend. By directly comparing the values ​​and signs of the two slopes, one can quantitatively determine whether the evenness index sequence continues the historical trend or whether a trend reversal or change has occurred after a deviation from the trend appears. Another auxiliary method is to calculate the difference between the index values ​​of adjacent sampling points in the sequence and observe the sequence of positive and negative signs of these differences to intuitively judge the persistence of the change.

[0083] To identify evenness change patterns that characterize population differentiation, stability, or homogenization based on the changing trends of evenness index sequences, a clear mapping rule from quantitative trends to qualitative patterns needs to be established. First, a trend change significance threshold needs to be set to determine whether a trend is significant. This threshold can be determined based on the inherent volatility of the evenness index sequence during historical normal periods. For example, the standard deviation of all evenness index values ​​during historical normal periods can be calculated, and the trend change significance threshold can be set as K times that standard deviation. K can be empirically set to a number between 1 and 2, such as 1.5. The pattern recognition rules are as follows: If the slope of the fitted line for the current deviation period is positive, and its absolute value is greater than the set trend change significance threshold, then the evenness change pattern is determined to be population differentiation. If the slope of the fitted line for the current deviation period is negative, and its absolute value is greater than the trend change significance threshold, then the evenness change pattern is determined to be population homogenization. If the absolute value of the slope of the fitted line for the current deviation period is less than or equal to the trend change significance threshold, then there is no significant trend change, i.e., the evenness change pattern is stability. In addition, if the difference between adjacent points reveals that the index value changes continuously in one direction for more than three consecutive sampling periods, this can also serve as a supplementary basis for pattern determination.

[0084] S5. Based on the significant changes in causal relationships and uniformity change patterns, generate and execute management adjustment instructions for subsequent stages, specifically as follows:

[0085] To identify potential key regulatory links based on the production state variables involved in significantly changed causal relationships, it is first necessary to analyze the set of significantly changed causal relationships identified in step S3. Each significantly changed causal relationship records two production state variables and their causal direction. For example, a causal relationship recorded as dissolved oxygen concentration pointing to ammonia nitrogen concentration indicates that dissolved oxygen concentration is identified as a causal variable affecting ammonia nitrogen concentration. The core of identifying key regulatory links lies in starting from these causal variables and, based on a pre-defined mapping relationship between causal variables and controllable links, associating them with operational links in actual aquaculture management that can be directly or indirectly controlled. This mapping relationship is pre-summarized based on the physical and biochemical principles and operational experience in the aquaculture field. For example, when the causal variable is water temperature, the potential key regulatory links that can be associated include adjusting water exchange operations and adjusting aeration operations, because water exchange directly changes the water temperature, while the water agitation during aeration affects the water temperature distribution. When the causal variable is dissolved oxygen concentration, the directly associated potential key regulatory link is adjusting aeration operations. When the causal variable is ammonia nitrogen concentration or nitrite concentration, the associated potential critical control points may include adjusting water exchange operations for dilution and adjusting feeding operations to reduce pollutant input. When the causal variable is average body weight, the main associated potential critical control point is adjusting feeding operations. In practice, the system iterates through all causal relationships that have changed significantly, extracts the production state variables that are the causes, and finds all corresponding potential control points based on the above mapping relationship. If a potential control point is pointed to by multiple causal relationships, its priority as a critical control point is increased. Finally, after summarizing and deduplicating, a list of potential critical control points is obtained; for example, the list may include adjusting aeration operations and adjusting feeding operations.

[0086] To integrate the population differentiation status indicated by the evenness change pattern and formulate management adjustment instructions for key control points, the qualitative judgment of the evenness change pattern output from step S4 needs to be incorporated into the decision-making logic. There are three main evenness change patterns: intensified population differentiation, maintaining stability, and trending towards evenness. This pattern provides additional information about the internal structural state of the population for the formulation of adjustment instructions. For each identified key control point, based on its operation type and the overall evenness change pattern, the specific timing and intensity of adjustment are determined to form a complete instruction. The formulation logic is implemented through a pre-set, rule-based empirical decision table. Taking the adjustment of feeding as an example, the key control point is explained below. First, the operation type is determined to be feeding. The timing of the adjustment needs to refer to the current system time and the routine feeding time set in the aquaculture production plan. It is usually set to be executed at the next nearest standard feeding time. For example, if the current analysis completion time is 10:00 AM, and the next feeding in the aquaculture plan is at 2:00 PM, then the adjustment timing is set at 2:00 PM. Determining the intensity of adjustments requires considering the overall pattern of evenness changes: If the evenness change pattern indicates increased group differentiation, it suggests widening differences in individual size within the group. In this case, the adjustment directive should aim to promote evenness. Specific adjustment strategies could include changing the planned single-point concentrated feeding to multi-point dispersed feeding, and increasing the proportion of additives promoting even growth in the feed by a certain percentage, such as 10%. If the evenness change pattern is to maintain stability, the adjustment intensity should be fine-tuned based on the severity of the problem revealed by the causal relationship of the significant change, such as reducing the feeding amount by 5%. If the pattern is to tend towards evenness, the original intensity can be maintained or only a symbolic adjustment of 1% to 2% can be made. Taking the adjustment of aeration as a key control point as an example, the operation type is determined to be aeration. The timing of the adjustment may be set to immediate execution or at the beginning of the next planned aeration period. Determining the adjustment intensity: If the uniformity change pattern indicates increased population differentiation, studies have shown that a more uniform distribution of dissolved oxygen in the water helps reduce stress differences between individuals. Therefore, adjustment instructions may include increasing the operating power of aerators or extending the duration of single aeration sessions, such as increasing the nighttime aeration duration from the planned 6 hours to 8 hours. If the pattern is otherwise, the basic aeration intensity is adjusted primarily based on the results of causal relationship analysis. This logic generates a management adjustment instruction for each key control step, containing a clear operation type, adjustment timing, and specific adjustment intensity parameters.

[0087] To output management adjustment instructions to the aquaculture management system and drive the actuators to perform corresponding operations in subsequent stages, it is necessary to complete the formatted encapsulation and communication transmission of the instructions. Each pre-defined management adjustment instruction is encapsulated according to the aquaculture management system's preset recognizable instruction data protocol. This protocol stipulates that each instruction must include at least the following fields: a unique instruction sequence number, a target execution device identifier, an operation type code, a planned execution timestamp, and an operation intensity parameter value. For example, an instruction regarding aeration might be encapsulated as follows: a unique instruction sequence number of 20251027001, a target execution device identifier of Aerator_Controller_01, an operation type code of AERATION, a planned execution timestamp of 2025-10-28 22:00:00, and an operation intensity parameter value of 1.5 kW continuous operation for 480 minutes. After encapsulation, the instruction data packet is sent to the aquaculture management system's central control server via a wired local area network or wireless communication network deployed in the aquaculture farm. After receiving the instruction data packet, the central control server of the aquaculture management system parses and verifies it, then stores it in the queue of instructions to be executed, while simultaneously performing time scheduling and resource conflict detection. When the system clock reaches the planned execution timestamp, the aquaculture management system generates corresponding underlying device control signals based on the target execution device identifier code and operation intensity parameter values ​​in the instruction. For example, for an aerator controller with the identifier code Aerator_Controller_01, an on signal is sent along with parameters setting the operating power to 1.5 kW and the duration to 480 minutes. Upon receiving this control signal, the actuator, i.e., the aerator controller, drives the connected physical aerator motor to start operating according to the set parameters, thus completing the physical execution of the aeration operation adjustment instruction. For feeding operation instructions, the target execution device is the automatic feeder controller, which drives the feeder motor and feeding mechanism; for water change operation instructions, the water pump and electric valve are driven. Through this series of steps, the intelligent decision-making based on causal analysis and population evenness diagnosis is transformed into specific, executable, and traceable physical intervention actions on the aquaculture environment and process.

[0088] S6. Update the aquaculture production plan based on the execution results of the management adjustment instructions, specifically as follows:

[0089] After management adjustment instructions are executed in subsequent stages, production status data for these stages is obtained to evaluate the effectiveness of the instructions. The core of this process is defining an evaluation time window that matches the type of instruction. The starting point of this evaluation time window is set at the exact moment the physical operation of the management adjustment instruction is completed by the executing agency in step S5. The ending point of the evaluation time window, also known as the evaluation cutoff time, is not arbitrarily set but is predefined based on the typical response times of the physiological and hydrodynamic processes of aquaculture organisms affected by different management operations. For example, for operations such as oxygenation or water exchange, which mainly cause changes in the physicochemical parameters of the water body, the aquaculture water body can usually reach a new dynamic equilibrium within 24 hours; therefore, the evaluation cutoff time can be set 24 hours after the instruction is executed. For feeding adjustments, since they affect growth and metabolism through ingestion, the organism's response requires a longer time to be reflected in measurable growth parameters; therefore, the evaluation cutoff time can be set 72 hours after the instruction is executed. Within the defined evaluation time window, the system continuously collects environmental parameters such as water temperature, dissolved oxygen concentration, and ammonia nitrogen concentration, strictly following the sensor monitoring and manual sampling specifications established in step S1. It also performs sampling and weighing at a fixed interval, for example, every three days, to obtain average body weight data. All production status data obtained within the evaluation window are organized chronologically into a new dataset specifically for this effectiveness evaluation. The essence of the effectiveness evaluation is to compare the actual observations in this dataset with the expected production status sequence pre-set for the exact same future time period in the current aquaculture production plan. For example, if the instruction aims to control ammonia nitrogen concentration, the readings of all ammonia nitrogen sensors are extracted from the evaluation window dataset, their average value and fluctuation range are calculated, and compared with the expected upper limit of ammonia nitrogen concentration for that period specified in the aquaculture production plan, for example, 0.5 mg / L. This comparison forms the basis for subsequent quantitative analysis.

[0090] To analyze the degree to which the execution of instructions corrects deviations from the trend, a quantifiable measurement method is needed. The degree of correction measures the extent to which the abnormal response logic that initially led to the deviation judgment is corrected after the instruction is executed. In practice, the system re-invokes the complete logic for judging the deviation trend in step S2, but replaces the analysis object with the management adjustment instruction executed this time and the subsequent observed changes in production status. First, the instruction is treated as an independent management operation event. Within the production status data within the evaluation time window, production status change events related to the instruction's expected target are identified. For example, if the instruction is oxygenation, an increase in dissolved oxygen concentration is identified within the window. Next, according to the algorithm defined in step S2, the actual response characteristic parameters between the instruction event and the actual status change event are calculated. Then, the calculated actual response characteristic parameters are compared again with the expected response characteristic parameters of the corresponding operation type in the preset causal logic relation library to calculate a new difference value. This new difference value is compared with the original difference value calculated and stored when step S2 initially judged the deviation trend. The degree of correction is quantified as the proportion of the reduction in the original difference value. The specific calculation method is as follows: subtract the new difference value from the original difference value, divide the difference by the original difference value, and finally multiply by 100% to obtain a percentage value. To determine whether this percentage value represents a significant correction, it needs to be compared with a preset correction degree judgment threshold. This correction degree judgment threshold is set based on statistical analysis of successful correction cases in historical aquaculture data. For example, analysis of a large number of past cases has shown that when the difference value reduction rate exceeds 60%, the probability of subsequent production returning to normal exceeds 95%. Therefore, the correction degree judgment threshold can be set at 60%. If the calculated correction degree percentage is greater than or equal to the correction degree judgment threshold, the correction degree is judged as high; if it is between 30% and 60%, the correction degree is judged as medium; and if it is below 30%, the correction degree is judged as low.

[0091] The analysis of the improvement in evenness change patterns is conducted in parallel with the analysis of the degree of correction, but focuses on population structure indicators. Assessing the improvement requires acquiring new evenness data reflecting the population state after the implementation of instructions. Within a timeframe overlapping with or adjacent to the production status data assessment window, the population size is resampled and measured using the exact same sampling scheme as specified in step S4, such as the same number of sampling points and random sampling method. Based on the newly collected individual size data, a new evenness index sequence is recalculated using the exact same formula as in step S4, namely the coefficient of variation. Subsequently, this new sequence is applied to a trend analysis method identical to that in step S4, such as linear fitting, to identify the new evenness change pattern after instruction implementation. The assessment of the improvement is determined by comparing the old and new patterns. The pattern comparison follows a pre-defined qualitative rule: if the old pattern shows increased group differentiation while the new pattern remains stable or tends towards uniformity, the improvement is considered significant; if the old pattern remains stable while the new pattern tends towards uniformity, the improvement is considered partial; if the old and new patterns are the same, the improvement is considered no; if the new pattern is worse than the old pattern (e.g., changing from remaining stable to showing increased differentiation), it is considered a deterioration. Combining the quantitative level of correction with the qualitative judgment of improvement forms a comprehensive evaluation conclusion of the instruction's effectiveness.

[0092] Based on a comprehensive analysis of the degree of correction and the improvement situation, adaptive adjustments are made to the expected production sequence of subsequent stages in the aquaculture production plan. This process relies on a set of pre-defined, parameterized adjustment rules. The core parameter of the adjustment is the basic adjustment coefficient K1, which is directly determined by the previously quantified correction level, and its mapping relationship is predefined. For example, the following mapping rule is established: if the correction level is high, the basic adjustment coefficient K1 is 0.9; if the correction level is medium, K1 is 0.6; if the correction level is low, K1 is 0.3. This mapping relationship is established based on the correlation analysis of the subsequent plan adjustment range and the final aquaculture success rate under different correction levels in historical adjustment cases, aiming to make the adjustment range positively correlated with the verified correction effect. The focus of the adjustment is controlled by the improvement situation. The system maintains a priority weight table to assign adjustment weights to environmental parameters such as dissolved oxygen and biological parameters such as average body weight for different improvement situations. For example, if the improvement is significant, the adjustment weight for biological parameters is set to 0.7, and the weight for environmental parameters is set to 0.3, encouraging growth based on a healthy population structure. If the improvement is not significant or worsens, the weight for environmental parameters is set to 0.8, and the weight for biological parameters is set to 0.2, prioritizing the stability of the basic environment. In specific adjustments, for all future stages starting from the next stage in the aquaculture production plan, each production state parameter requiring adjustment is iterated. The new target value for each parameter is calculated using an adjustment formula. The inputs to this formula include: the original planned target value for the parameter, the actual performance value or trend of the parameter observed within the evaluation window, the basic adjustment coefficient K1, the adjustment weight corresponding to the parameter type, and the remaining time factor until the end of the aquaculture period. Essentially, the calculation process involves compensating for the deviation between the observed actual performance and the original plan according to the coefficient K1 and the type weight, and then adding this compromise to the original planned target value, thus forming a new target value that is closer to reality. After all parameters have undergone this calculation, a complete adjusted expected production state sequence is generated.

[0093] Updating the adjusted expected production status sequence to the aquaculture production plan is a data persistence and version control operation. The adjusted sequence is a structured data table, where each row represents a future aquaculture stage, and each column represents a production status parameter and its adjusted target value. The update process is performed in an electronic database storing the aquaculture production plan. The system first queries the plan database table for all records whose stage start dates are later than the current date, using the current date as the boundary. Then, it replaces the original target value field in these records one by one with the calculated new target value. For example, the value of the target average weight field in the original record is updated from 25 grams to the calculated 23.5 grams. To ensure traceability, the system backs up the entire plan data to be replaced before performing the overwrite update, and associates it with the reason for the update, namely the management adjustment instruction number and the effect evaluation conclusion. At the same time, a new aquaculture production plan version number is generated, following a progressive rule, such as upgrading from V2.1 to V2.2. This updated aquaculture production plan version is marked as the current valid version. From this point onward, in subsequent stages of the aquaculture process, when step S1 mentions obtaining the preset aquaculture production plan, this latest version will be used. All comparisons, analyses, and decision-making cycles based on the expected values ​​of the plan will operate based on this updated target system. Through this step, the entire method achieves a complete adaptive closed loop from problem detection, intervention execution, effect feedback to plan optimization, enabling aquaculture management to dynamically respond to the actual state of the system.

[0094] Example 2

[0095] Figure 2 A schematic diagram of the structure of an intelligent digital aquaculture management system according to the present invention is provided. The intelligent digital aquaculture management system includes:

[0096] The data acquisition module is used to acquire the preset breeding production plan and continuously acquire production status data and management operation data executed in the actual breeding process;

[0097] The trend determination module is used to analyze the response logic between the actual management operation data and the corresponding production status data. When the response logic is inconsistent with the preset causal logic relationship, it is determined that a deviation from the trend has occurred.

[0098] The causal identification module is used to identify significantly changed causal relationships when a deviation trend is detected, based on a causal relationship analysis algorithm that constructs and compares a causal relationship network using historical and current production status data.

[0099] The pattern analysis module is used to analyze the pattern of uniformity changes in individual size within a breeding population when deviations from the trend occur.

[0100] The instruction generation module is used to synthesize the causal relationships and uniformity change patterns that have undergone significant changes, generate and execute management and adjustment instructions for subsequent stages;

[0101] The plan update module is used to update the aquaculture production plan based on the execution results of management adjustment instructions.

[0102] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0103] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0104] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. Computer-readable storage media can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0105] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0106] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0107] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0108] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0109] If a function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0110] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0111] In conclusion, the above are merely preferred embodiments of the present invention and are 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 smart digital aquaculture management method, characterized in that, include: S1. Obtain the preset breeding production plan and continuously obtain production status data and management operation data during the actual breeding process; S2. Analyze the response logic between the actual management operation data and the corresponding production status data. When the response logic is inconsistent with the preset causal logic relationship, it is determined that a deviation trend has occurred. S3. When a deviation trend is detected, a causal relationship network is constructed and compared based on historical and current production status data using a causal relationship analysis algorithm to identify causal relationships that have changed significantly. S4. Analyze the pattern of uniformity of individual size within the breeding population when deviations from the trend occur; S5. Based on the significant changes in causal relationships and uniformity change patterns, generate and execute management adjustment instructions for subsequent stages; S6. Update the aquaculture production plan based on the execution results of the management adjustment instructions; S2 includes: Based on the temporal relationship between management operation data and production status data, a mapping relationship between management operation events and subsequent production status change events is established. Extract the actual response characteristic parameters between management operation data and production status data from the mapping relationship; The actual response feature parameters are compared with the expected response feature parameters in the preset causal logic relationship; When the difference between the actual response characteristic parameters and the expected response characteristic parameters exceeds the preset difference threshold, it is determined that the response logic is inconsistent with the preset causal logic relationship. The pre-defined causal logic relationship is established in the following way: analyze the management operation data and production status data during the historical normal period, extract the stable temporal correlation and quantitative response characteristics between management operation events and subsequent production status change events, and establish the pre-defined causal logic relationship based on the extracted temporal correlation and quantitative response characteristics. S3 includes: Based on the time point for determining the deviation trend from the production status data, historical normal period production status data and current deviation period production status data are divided. Based on the causal relationship analysis algorithm, the production status data of the historical normal period and the production status data of the current deviation period are processed respectively to obtain the historical causal relationship network reflecting the causal influence relationship between production status variables in the historical period, and the current causal relationship network reflecting the causal influence relationship between production status variables in the current period. The current causal relationship network is compared with the historical causal relationship network. The specific comparison includes: identifying causal relationships that exist in both causal relationship networks but whose strength changes of the causal edges connecting the same pair of production state variables exceed a preset strength threshold; and identifying causal relationships that exist only in one of the causal relationship networks. Causal relationships that meet the comparison criteria are identified as causal relationships that have undergone significant changes.

2. The intelligent digital aquaculture management method according to claim 1, characterized in that, S1 includes: Obtain an aquaculture production plan that includes a phased sequence of expected production states; Continuous monitoring of aquaculture environment parameters and biological parameters of the aquaculture population through sensors to generate production status data; Continuously record the types, times, and intensity of feeding, aeration, and water exchange operations performed during the aquaculture process to form management operation data.

3. The intelligent digital aquaculture management method according to claim 2, characterized in that, Production status data is correlated with the expected production status sequence in the aquaculture production plan, and management operation data is correlated with production status data in time.

4. The intelligent digital aquaculture management method according to claim 1, characterized in that, S4 includes: Obtain sampled data on individual size within the aquaculture population at different time points before and after the occurrence of deviation from the trend; Based on the sampled data of individual size within the breeding population, a uniformity index sequence characterizing the dispersion of individual size within the population is calculated and formed. Analyze the changing trends of the evenness index sequence before and after the occurrence of deviation trends; Based on the changing trends of the evenness index sequence, evenness change patterns that characterize the intensification of population differentiation, the maintenance of stability, or the tendency towards evenness can be identified.

5. The intelligent digital aquaculture management method according to claim 1, characterized in that, S5 include: Based on the production state variables involved in the significant changes in causal relationships, identify potential key control points; Based on the population differentiation status indicated by the uniformity change pattern, management adjustment instructions including operation type, adjustment timing and adjustment intensity are formulated for key control links; The management adjustment instructions are output to the aquaculture management system to drive the execution mechanism to perform the corresponding feeding, oxygenation or water change operations in the subsequent stages.

6. The intelligent digital aquaculture management method according to claim 1, characterized in that, S6 include: After management adjustment instructions are executed in subsequent stages, production status data for those stages is obtained to evaluate the effectiveness of the instructions. The analysis examines the degree to which the execution of instructions corrects deviations from the trend and the improvement in uniformity variation patterns. Based on the comprehensive analysis of the degree of correction and the improvement, adaptive adjustments are made to the expected production status sequence of the subsequent stages in the aquaculture production plan; The adjusted expected production status sequence is updated to the aquaculture production plan, forming an updated aquaculture production plan to guide subsequent aquaculture processes.

7. An intelligent digital aquaculture management system, used to implement the intelligent digital aquaculture management method according to any one of claims 1-6, characterized in that, include: The data acquisition module is used to acquire the preset breeding production plan and continuously acquire production status data and management operation data executed in the actual breeding process; The trend determination module is used to analyze the response logic between the actual management operation data and the corresponding production status data. When the response logic is inconsistent with the preset causal logic relationship, it is determined that a deviation from the trend has occurred. The causal identification module is used to identify significantly changed causal relationships when a deviation trend is detected, based on a causal relationship analysis algorithm that constructs and compares a causal relationship network using historical and current production status data. The pattern analysis module is used to analyze the pattern of uniformity changes in individual size within a breeding population when deviations from the trend occur. The instruction generation module is used to synthesize the causal relationships and uniformity change patterns that have undergone significant changes, generate and execute management and adjustment instructions for subsequent stages; The plan update module is used to update the aquaculture production plan based on the execution results of management adjustment instructions.

Images