A method for monitoring the health status of eel fry based on multi-sensor data

Abstract

The present application relates to fry monitoring technical field, especially a kind of eel fry health state monitoring method based on multi-sensor data, comprising: obtaining environmental data, measured behavior data and breeding operation data, and gathering as monitoring data;Perform time correspondence analysis, and obtain response time delay data;Calibrate normal operation response benchmark;Match to generate expected behavior data;Separation obtains remaining deviation data;Identify the moment of influence subsides and analyze the remaining deviation, obtain recovery feature data;According to normal operation response benchmark, make recovery abnormality determination, and generate monitoring result.The present application is beneficial to reduce short-term fluctuation false alarm and reduce early continuous abnormality omission probability by jointly judging deviation continuous over-limit and recovery overtime after environmental influence subsides.

Description

Technical Field

[0001] This invention relates to the field of fish fry monitoring, and in particular to a method, electronic device and storage medium for monitoring the health status of eel fry based on multi-sensor data. Background Technology

[0002] During the rearing, acclimatization, and initial feeding stages of eel fry in high-density recirculating aquaculture ponds, aquaculture operations such as feeding, aeration, water exchange, and water flow regulation can cause short-term changes in the fry's swimming, aggregation, and feeding behaviors. At the same time, fluctuations in water quality and water flow conditions may also cause similar behavioral changes. Therefore, it is necessary to combine water quality data, actual behavioral data of the fry population, and aquaculture operation records to determine the actual source of abnormal fry behavior and identify potential health declines that may occur before water quality parameters exceed limits.

[0003] Existing technologies typically utilize sensors for temperature, dissolved oxygen, pH, and nitrogenous substances to collect water quality data from nursery ponds, and employ camera equipment to capture the swimming, gathering, and feeding behaviors of fish fry. Subsequently, the water quality data and measured behavioral data are correlated over time and features are extracted. Preset thresholds or classification models are then used to determine whether the fish fry exhibit any health abnormalities.

[0004] However, existing technologies typically link water quality data and measured behavioral data for the same period, without fully considering the short-term impact of aquaculture operations on fry behavior. This can easily lead to misjudging behavioral changes caused by normal operations as health abnormalities. Furthermore, existing technologies primarily rely on whether water quality parameters exceed limits or the degree of behavioral abnormality for judgment, lacking continuous analysis of the fry behavior recovery process after environmental disturbances. This makes it difficult to detect early health abnormalities in a timely manner, even when water quality parameters have not yet exceeded limits, but the fry's behavioral recovery speed has slowed down or the degree of recovery has decreased after environmental disturbances. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies that make it difficult to distinguish between short-term behavioral changes caused by aquaculture operations and true health abnormalities, and also make it difficult to identify early health abnormalities in a timely manner when water quality parameters have not exceeded limits but the fish fry's behavioral recovery ability has already declined. Therefore, this invention proposes a method, electronic device, and storage medium for monitoring the health status of eel fry based on multi-sensor data.

[0006] To at least solve some of the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] Firstly, this invention provides a method for monitoring the health status of eel fry based on multi-sensor data, including:

[0008] Environmental data, measured behavioral data, and aquaculture operation data are collected and aggregated into monitoring data.

[0009] The monitoring data is analyzed to determine the corresponding execution time, and the response delay data is obtained.

[0010] Based on the response delay data, the monitoring data is calibrated to obtain the normal operation response benchmark;

[0011] The aquaculture operation data and environmental data are matched with the normal operation response benchmark to obtain the expected behavior data;

[0012] Deviation separation is performed between the measured behavioral data and the expected behavioral data to obtain the remaining deviation data;

[0013] Based on the aquaculture operation data, environmental data, and normal operation response benchmark, the time of impact decline is identified, and the remaining deviation data after the time of impact decline is analyzed to obtain recovery characteristic data.

[0014] Based on the normal operation response benchmark, the recovery feature data is subjected to recovery anomaly determination, and monitoring results are generated.

[0015] Preferably, the monitoring data is analyzed according to the execution time to obtain response delay data, including:

[0016] Operational change extraction is performed on the aquaculture operation data to obtain the operation change sequence;

[0017] Environmental change extraction is performed on the environmental data to obtain an environmental change sequence;

[0018] Perform time delay correlation analysis on the operation change sequence and the environment change sequence to obtain time delay correlation results that include candidate time delays and correlation degrees;

[0019] The latency-related results are correlated and confirmed to obtain response latency data.

[0020] Preferably, the normal operation response benchmark is obtained by performing response benchmark calibration on the monitoring data based on the response delay data, including:

[0021] Acquire breeding status records, correct monitoring data according to the aforementioned response delay data, and filter health records;

[0022] Response feature summarization is performed on the health records to obtain normal response data;

[0023] The recovery process of health records is analyzed based on the normal response data to obtain recovery baseline data;

[0024] The response delay data, normal response data, and recovery baseline data are collected to obtain the normal operation response baseline.

[0025] Preferably, the aquaculture operation data and environmental data are matched with the normal operation response benchmark to obtain expected behavior data, including:

[0026] The aquaculture operation data is matched with the normal operation response benchmark to obtain a matching benchmark record, which includes a benchmark environmental response sequence and a benchmark behavioral response sequence.

[0027] The current response sequence is obtained by performing current response extraction on the environmental data using the matching benchmark record.

[0028] The response magnitude is compared with the current environmental response sequence and the matching baseline record to obtain the response adjustment ratio;

[0029] The baseline behavior response sequence in the matching baseline record is corrected using the aforementioned response adjustment ratio to obtain the expected behavior data.

[0030] Preferably, deviation separation is performed between the measured behavioral data and the expected behavioral data to obtain residual deviation data, including:

[0031] Perform behavior response conversion on the measured behavior data to obtain measured behavior response data;

[0032] Deviation calculation is performed on the measured behavioral response data and the expected behavioral data to obtain behavioral deviation data;

[0033] The behavioral deviation data is subjected to over-limit separation based on the normal operation response benchmark to obtain the remaining deviation data.

[0034] Preferably, based on the aquaculture operation data, environmental data, and normal operation response benchmark, the time of impact decline is identified, and the remaining deviation data after the time of impact decline is analyzed to obtain recovery characteristic data, including:

[0035] Based on the aforementioned aquaculture operation data, environmental data, and normal operation response benchmarks, impact regression identification is performed to obtain the impact regression time.

[0036] The remaining deviation data is extracted at the time when the influence fades to obtain candidate recovery data;

[0037] Perform recovery boundary identification on the candidate recovery data to obtain recovery analysis data and the current recovery duration;

[0038] A persistent deviation analysis is performed on the recovery analysis data, and the data is aggregated to form recovery feature data.

[0039] Preferably, based on the normal operation response benchmark, a recovery anomaly determination is performed on the recovery feature data to generate monitoring results, including:

[0040] Recovery determination parameters are extracted from the normal operation response benchmark to obtain the recovery determination benchmark;

[0041] The recovery feature data is compared with the recovery judgment benchmark in terms of execution time to obtain the recovery comparison result;

[0042] A joint determination is performed on the recovery comparison results to obtain the recovery anomaly determination result;

[0043] Based on the recovery anomaly determination results, monitoring results are generated.

[0044] Preferably, the latency-related results are correlated and confirmed to obtain response latency data, including:

[0045] From the monitoring data, stable periods in which no aquaculture operations were recorded and environmental data were within the permissible range of aquaculture management were selected, and stable environmental change sequences were extracted;

[0046] Perform correlation analysis on the stable environment change sequence and the operation change sequence to obtain a set of background correlation degrees;

[0047] Perform quantile statistics on the set of background relevance to obtain the association confirmation threshold;

[0048] The environmental indicators and candidate delays corresponding to the highest relevant records that exceed the association confirmation threshold are respectively determined as the associated environmental indicators and response delay values, forming response delay data.

[0049] Preferably, the response magnitude is compared with the matching baseline record to obtain the response adjustment ratio, including:

[0050] The absolute value of the current environmental response value in the current environmental response sequence is taken, and the maximum value is obtained to obtain the current environmental response amplitude.

[0051] Extract the baseline environmental response sequence from the matching baseline record, take the absolute value of the values ​​and the maximum value to obtain the baseline environmental response amplitude;

[0052] The response adjustment ratio is obtained by dividing the current environmental response amplitude by the baseline environmental response amplitude; when the baseline environmental response amplitude is 0, the response adjustment ratio is set to 1.

[0053] Preferably, a joint determination is performed on the recovery comparison results to obtain a recovery anomaly determination result, including:

[0054] Extract the deviation persistence comparison result and the recovery duration comparison result from the recovery comparison result;

[0055] When the deviation persistence comparison result is that the deviation persists beyond the limit and the recovery time comparison result is that the recovery timeout occurs, a result indicating that the risk of persistent behavior decay is established is generated.

[0056] If at least one of the above two conditions is not met, the result "the risk of persistent behavior decay is not met" is generated.

[0057] The result of whether the risk of persistent behavior decay is established or not is determined as the result of recovery anomaly determination.

[0058] In a second aspect, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method in the first aspect.

[0059] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method in the first aspect.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] I. This invention confirms the true response relationship between aquaculture operations and environmental indicators from actual monitoring data, and corrects the time relationship of monitoring data according to the response delay value. This helps to reduce time mismatch caused by water propagation and avoids misalignment of similar response curves due to different water propagation times.

[0062] Second, this invention corrects the normal behavioral response according to the degree of environmental change actually caused by the current aquaculture operation, and deducts the normal fluctuations that are allowed to occur in healthy fish fry under similar aquaculture operations, retaining only the behavioral differences that cannot be fully explained by the normal operation response, which helps to reduce false alarms caused by normal aquaculture operations and short-term behavioral fluctuations.

[0063] Third, by combining the judgment of persistent deviation and recovery timeout after the environmental impact caused by aquaculture operations subsides, this invention helps to eliminate short-term behavioral fluctuations that recover quickly, and to identify situations where environmental indicators are still within the management range, but the fish fry behavior continues to deviate and the recovery time is abnormally prolonged, thereby reducing false alarms and lowering the probability of missing early persistent anomalies. Attached Figure Description

[0064] Figure 1 This is a flowchart of the method for monitoring the health status of eel fry according to the present invention.

[0065] Figure 2 This is a comparison diagram of the effects of the present invention and the prior art. Detailed Implementation

[0066] To make the technical means, creative features, objectives, and effects of this invention easier to understand, the invention is further described below with reference to specific embodiments. However, the following embodiments are merely preferred embodiments of this invention and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments described herein without creative effort are all within the protection scope of this invention.

[0067] A specific embodiment of the present invention discloses a method for monitoring the health status of eel fry based on multi-sensor data. Please refer to [link to relevant documentation]. Figure 1 ,include:

[0068] S1. Acquire environmental data, measured behavioral data, and aquaculture operation data, and compile them into monitoring data;

[0069] S2. Perform time-correspondence analysis on the monitoring data to obtain response delay data;

[0070] S3. Based on the response delay data, perform response benchmark calibration on the monitoring data to obtain the normal operation response benchmark;

[0071] S4. Match the aquaculture operation data and environmental data with the normal operation response benchmark to obtain the expected behavior data;

[0072] S5. Perform deviation separation on the measured behavior data and the expected behavior data to obtain the remaining deviation data;

[0073] S6. Identify the time of impact decline based on the aquaculture operation data, environmental data and normal operation response benchmark, analyze the remaining deviation data after the time of impact decline, and obtain recovery characteristic data;

[0074] S7. Based on the normal operation response benchmark, perform recovery anomaly determination on the recovery feature data and generate monitoring results.

[0075] As an optional implementation, the water quality acquisition devices, camera equipment, and aquaculture control equipment in the high-density recirculating aquaculture pond have different data formats and recording periods. Directly splicing these data can easily cause the same aquaculture operation to correspond to environmental changes and fry behavior at different times. This implementation can solve the above problems, specifically:

[0076] S1. Acquire environmental data, measured behavioral data, and aquaculture operation data, and compile them into monitoring data;

[0077] In this context, aggregation refers to unifying the time reference and arrangement order of data without changing the original sampled values.

[0078] The water quality sampling device collects dissolved oxygen, turbidity, and water flow velocity, and records the sampling time, seedling pond number, and sensor location to form environmental data.

[0079] The camera continuously captures images of the fish fry group, and the processing terminal uses background subtraction to extract the foreground region of the fry. The number of foreground pixels that change in two adjacent images is divided by the average number of foreground pixels in the two images to obtain the group activity intensity; the number of foreground pixels in the pool wall area is divided by the total number of foreground pixels in the entire image to obtain the pool wall aggregation ratio; the group activity intensity and pool wall aggregation ratio are arranged according to the acquisition time to form measured behavioral data. By using a group image statistical method, it is not necessary to continuously identify individual fry, and it can still reflect the overall behavioral changes when the fry occlude each other.

[0080] The aquaculture control equipment records the operation type, start time, end time, operation characteristic value, nursery pond number, and aquaculture batch stage for feeding, aeration, water exchange, and water flow regulation, forming aquaculture operation data. For aquaculture operations with continuously adjustable execution intensity, the equipment setpoint is used as the operation characteristic value; for aquaculture operations with only start / stop states, the start / stop state value is used as the operation characteristic value. Manually performed aquaculture operations are recorded using the same information entered through the operation terminal.

[0081] After receiving the above data, the processing terminal uses a unified clock to correct the time stamps in the environmental data, measured behavior data, and aquaculture operation data, and then collects them according to the nursery pond number and time sequence to form monitoring data.

[0082] This method enables the same aquaculture operation to accurately correspond to the corresponding environmental changes and fish fry behavior, which helps to reduce data mismatch caused by inconsistent recording times of different devices.

[0083] As an alternative implementation method, the effects of oxygenation, water exchange, and water flow regulation need to propagate through the water body to reach different pool areas. Simply unifying the equipment clock will still result in inconsistencies between the operating time and the environmental response time. Therefore, this invention proposes the following working method:

[0084] S2. Perform time-correspondence analysis on the monitoring data to obtain response delay data;

[0085] Time correspondence analysis refers to comparing the degree of correspondence between changes in aquaculture operations and environmental changes at different delayed times, starting from the moment when changes occur in aquaculture operations, in order to determine the associated environmental indicators and their response time delay values.

[0086] S2.1. Perform operation change extraction on the aquaculture operation data to obtain the operation change sequence;

[0087] The processing terminal reads a single aquaculture operation from the aquaculture operation data and establishes equally spaced time points according to the sampling time of the environmental data.

[0088] For aquaculture operations with continuously adjustable execution intensity, the equipment setting value corresponding to each time point is used as the operation characterization value; for aquaculture operations with only start and stop states, the non-executed state is recorded as 0 and the executed state is recorded as 1, which are used as the operation characterization values ​​at the corresponding time points; the current operation characterization value is used during the duration of the aquaculture operation, and the value corresponding to the non-executed state is restored after the aquaculture operation ends, and arranged in chronological order to form an operation characterization sequence consistent with the environmental data sampling period.

[0089] Subsequently, the operation change is obtained by subtracting the operation characterization value from the previous time point from the operation characterization value at the current time point; the operation change values ​​are then arranged in chronological order to form an operation change sequence.

[0090] This treatment can highlight the timing of the start, stop, and intensity adjustment of aquaculture operations, and reduce the interference of constant operational values ​​on time-correspondence analysis.

[0091] S2.2, Perform environmental change extraction on the environmental data to obtain an environmental change sequence;

[0092] Extract individual continuous records from the environmental data according to the data name. Each individual continuous record is an environmental indicator, which includes at least one of dissolved oxygen, turbidity, and water flow velocity.

[0093] By subtracting the previous sample value from the current sample value, the environmental change of the corresponding environmental indicator at the current sampling time is obtained. All environmental changes are arranged continuously according to the sampling time to form the environmental change sequence corresponding to the environmental indicator. The environmental change can highlight the moment when the environmental state begins to change, and avoid the original value of the environmental indicator from masking the actual response process.

[0094] S2.3 Perform time delay correlation analysis on the operation change sequence and the environment change sequence to obtain time delay correlation results including candidate time delays and correlation degrees;

[0095] Candidate delay represents the waiting time required for environmental indicators to respond after an aquaculture operation. The processing terminal sets candidate delays sequentially, starting from 0, based on the environmental data sampling period, with the maximum candidate delay being twice the water circulation time. The water circulation time is determined by recording the time required for water to flow from the inlet area to the outlet area.

[0096] For each candidate delay, the amount of operational change at each sampling time in the operational change sequence is matched with the amount of environmental change in the environmental change sequence that is delayed by the candidate delay, to form paired data under the current candidate delay.

[0097] Subsequently, the Pearson correlation coefficient was used to calculate the linear correlation of the paired data. First, the average values ​​of the operational and environmental changes were calculated separately. Then, the operational change deviation was obtained by subtracting the operational average value from each operational change value, and the environmental change deviation was obtained by subtracting the environmental average value from each environmental change value.

[0098] Under the same candidate time delay, the operational variation deviation and environmental variation deviation corresponding to each sampling time are multiplied separately, and all products are summed to obtain the deviation product sum. Then, the squares of each operational variation deviation are summed to obtain the operational deviation sum of squares, and the squares of each environmental variation deviation are summed to obtain the environmental deviation sum of squares. The square root of the product of the operational deviation sum of squares and the environmental deviation sum of squares is taken to obtain the normalized value; finally, the deviation product sum is divided by the normalized value to obtain the linear correlation degree under the current candidate time delay.

[0099] When the normalization value is 0, the correlation degree corresponding to the current candidate delay is recorded as 0, and this record is not used for the confirmation of associated environmental indicators.

[0100] The absolute value of the linear correlation is taken to obtain the correlation degree of the current candidate time delay. The larger the absolute value, the more obvious the correspondence between the change in aquaculture operation and the environmental change that occurs after the corresponding delay. It can identify both responses where environmental indicators increase as operation intensifies and responses where environmental indicators decrease as operation intensifies.

[0101] The candidate time delays and their correlations for each environmental indicator are written into the same record to form a time delay correlation result.

[0102] S2.4 Perform correlation confirmation on the latency-related results to obtain response latency data;

[0103] Association confirmation is used to exclude incidental correlations caused by natural environmental fluctuations.

[0104] S2.4.1. Select stable periods from the monitoring data where no aquaculture operations were recorded and the environmental data was within the allowable range of aquaculture management, and extract stable environmental change sequences;

[0105] The processing terminal selects stable periods from the monitoring data of the same nursery pond where no aquaculture operations were recorded and the environmental data remained within the allowable range for aquaculture management. Subsequently, it generates environmental change quantities according to the environmental change extraction method in S2.2, and extracts environmental change quantities of the same length as the operation change sequence from each stable period to form a stable environmental change sequence.

[0106] S2.4.2 Perform correlation analysis on the stable environment change sequence and the operation change sequence to obtain the background correlation set;

[0107] For each stable environmental change sequence, the absolute correlation between it and the operational change sequence at each candidate time delay is calculated according to the calculation method in S2.3, and the maximum value is taken as the background correlation for the corresponding stable time period. Subsequently, the background correlations for each stable time period are aggregated to form a background correlation set.

[0108] S2.4.3 Perform quantile statistics on the background relevance set to obtain the association confirmation threshold;

[0109] The background relevance was sorted from lowest to highest. The sample size was multiplied by 95% and rounded up. The corresponding value at each sorted position was then read to obtain the association confirmation threshold. This threshold represents the upper limit of the possible correlation that natural environmental fluctuations could reach without the corresponding aquaculture operation occurring.

[0110] S2.4.4. The environmental indicators and candidate delays corresponding to the highest relevant records that exceed the association confirmation threshold are determined as the associated environmental indicators and response delay values, respectively, to form response delay data;

[0111] The record with the highest absolute correlation is selected from the latency correlation results. If the absolute correlation of this record exceeds the correlation confirmation threshold, its corresponding environmental indicator is determined as the correlated environmental indicator, and its corresponding candidate latency is determined as the response latency value; if it does not exceed the correlation confirmation threshold, this aquaculture operation will not be used for response benchmark calibration.

[0112] Subsequently, the aquaculture operation type, associated environmental indicators, sensor location, and response delay value are written into the same record to form response delay data.

[0113] This implementation method does not pre-determine that a certain type of aquaculture operation must correspond to a certain environmental indicator. It can confirm the true response relationship from actual monitoring data and reduce time mismatch caused by water body transmission.

[0114] As an optional implementation method, environmental data from different pond areas exhibit response time delays relative to the same aquaculture operation. Historical monitoring data may also contain periods of abnormal fish fry conditions and sensor malfunctions. Directly using this data to construct a normal baseline can easily lead to temporal misalignment of similar responses and expand the range of normal variations in abnormal records. This implementation method can solve the above problems, specifically:

[0115] S3. Based on the response delay data, perform response benchmark calibration on the monitoring data to obtain the normal operation response benchmark;

[0116] Response benchmark calibration refers to first correcting the time relationship of monitoring data according to the response delay value and screening health records, and then statistically analyzing the normal environmental response, normal behavioral response and normal recovery range corresponding to similar aquaculture operations.

[0117] S3.1 Obtain breeding status records, correct monitoring data according to the response delay data, and filter health records;

[0118] The processing terminal reads the response delay value corresponding to the current aquaculture operation type, associated environmental indicators, and sensor location from the response delay data. The corrected time is obtained by subtracting the response delay value from the original acquisition time of the corresponding environmental data, and the time position of the environmental data is adjusted according to the corrected time, while the environmental sampling value remains unchanged.

[0119] Environmental responses typically occur later than aquaculture operations. By using response delay values ​​to trace the time position of environmental data, the delayed environmental response can be mapped to the aquaculture operation that caused it. This ensures that aquaculture operations, environmental responses, and measured behaviors all use the time of aquaculture operation as a common reference, avoiding misalignment of similar response curves due to different water propagation times.

[0120] Subsequently, records of the fish's rearing status were acquired, corresponding to the monitoring period after time correction and belonging to the same nursery pond and the same rearing batch stage. These records included manual sampling records, feeding records, mortality statistics records, and sensor calibration records. Manual sampling records were entered by rearing personnel after inspecting the fry's body surface, gills, and swimming behavior; feeding records were entered by rearing personnel based on feeding behavior after feeding; mortality statistics records were entered by rearing personnel according to statistical periods; and sensor calibration records were generated after the water quality acquisition device was calibrated.

[0121] Then check each monitoring period after the completion of time correction; if the environmental data is within the management range of the corresponding breeding batch stage, no obvious abnormalities are found in manual sampling, feeding is within the historical normal range of the same breeding batch stage, the number of deaths has not increased continuously, and the sensor calibration record is within the validity period, the corresponding monitoring period is determined as a health record.

[0122] Each health record retains the seedling pond number, breeding batch stage, breeding operation data, environmental data, measured behavioral data, and associated environmental indicators, sensor locations, and response delay values ​​for the corresponding breeding operation within the monitoring period.

[0123] By eliminating the propagation misalignment of environmental responses through time correction and excluding periods of abnormal fish fry condition and sensor malfunctions using aquaculture status records, abnormal data can be prevented from being included in normal response statistics, enabling health records to reflect normal changes under similar aquaculture operations.

[0124] S3.2. Perform response feature summarization on the health records to obtain normal response data;

[0125] Within the same breeding batch, the processing terminal groups health records according to the operation type of feeding, oxygenation, water exchange, and water flow regulation, and sets the start time of each breeding operation as a relative time starting point.

[0126] In each health record, the period during which the most recent water circulation time before the operation began, no other aquaculture operations occurred, and the relevant environmental indicators were within the aquaculture management range was determined as the stable period before the operation. This period is used to represent the basic environmental and behavioral status before the operation.

[0127] The statistical period is defined as the time between the start of the aquaculture operation and the time corresponding to two water cycle durations after the operation is completed. The statistical period includes the period during which the aquaculture operation is performed and the recovery observation period after the operation.

[0128] For each health record, the average value of the associated environmental indicators during the stable period before the operation is calculated. Then, the environmental response value at the corresponding relative time position is obtained by subtracting the average value from each environmental sample value during the statistical period.

[0129] Next, sort the environmental response values ​​of each health record in the group at the same relative time position from smallest to largest. When the number of records is odd, take the median value; when the number of records is even, take the average of the two median values. Arrange the results of each relative time position continuously to form a baseline environmental response sequence. Here, the same relative time position means the position after the same amount of time has elapsed since the start of each of the same type of breeding operation. Therefore, multiple response values ​​corresponding to multiple health records can be obtained in the group.

[0130] The average value of each measured behavior during the stable period before the operation was calculated in the same way. The behavior response value was obtained by subtracting the corresponding average value from each measured behavior value during the statistical period. Then, the median of the behavior response values ​​of each health record in the group at the same relative time position was taken and arranged according to the relative time to form a baseline behavior response sequence.

[0131] All health records were state-screened. The baseline environmental response sequence and baseline behavioral response sequence represent the changes that healthy fish fry typically exhibit under similar aquaculture operations. Using the median of multiple health records can reduce the impact of small, sudden fluctuations on the normal response curve.

[0132] For each behavioral feature, the absolute difference between the behavioral response value in each health record and the corresponding value in the baseline behavioral response sequence is calculated. Then, all absolute differences of the same behavioral feature are sorted from smallest to largest, and the sample size is multiplied by 95% and rounded up. The value at the corresponding sort position is read to obtain the upper limit of the residual deviation for that behavioral feature.

[0133] Then, calculate the absolute difference between the associated environmental indicators and the average value of that period during the stable period before the operation of each health record, and use the same percentile method to obtain the upper limit of environmental stability.

[0134] The baseline environmental response sequence, the baseline behavioral response sequence, the upper limit of the residual deviation corresponding to each behavioral characteristic, and the upper limit of environmental stability together form the normal response data. This data not only reflects the normal change process of healthy fish fry under similar operations, but also limits the range of normal fluctuations that are allowed in the health records, which helps to avoid directly identifying a small amount of random fluctuations as abnormal.

[0135] S3.3. Based on the normal response data, analyze the recovery process of the health record to obtain recovery baseline data;

[0136] The processing terminal first determines the continuous confirmation duration. For each health record, the environmental deviation value is obtained by subtracting the average value of the stable period before the operation from the associated environmental indicator value at each sampling time after the aquaculture operation, and taking the absolute value. An environmental fluctuation process is defined as the process where the environmental deviation value continuously exceeds the environmental stability limit and then returns to within the environmental stability limit without new aquaculture operations. The duration for which the environmental deviation value continuously exceeds the environmental stability limit during each environmental fluctuation process is counted, and the longest duration is added by one environmental data sampling period to obtain the continuous confirmation duration. If no environmental fluctuation process is identified, the duration corresponding to three environmental data sampling periods is determined as the continuous confirmation duration, thereby eliminating misjudgments caused by a single sampling point or short-term natural fluctuations.

[0137] Subsequently, health records are analyzed one by one. After the aquaculture operation is completed, the environmental deviation value corresponding to the associated environmental indicators is continuously calculated. When the environmental deviation value does not exceed the upper limit of environmental stability in the normal response data and is continuously maintained for a certain period of time, the start time of the continuous period of time is determined as the time when the impact subsides. This judgment takes into account both whether the operation is completed and whether the environment has recovered, which can avoid starting the recovery statistics in advance when the environmental impact is still ongoing.

[0138] Starting from the moment the impact subsides, the behavioral deviation value corresponding to each behavioral characteristic is obtained by subtracting the corresponding value in the baseline behavioral response sequence from each behavioral response value and taking the absolute value. For each behavioral characteristic, the time period in which the behavioral deviation value continuously exceeds the corresponding remaining deviation upper limit is identified, and the duration of the longest time period is taken as the deviation duration of that behavioral characteristic; then, the maximum value among the deviation durations of all behavioral characteristics is taken as the normal deviation duration of that health record.

[0139] If the behavioral deviation values ​​of all behavioral characteristics do not exceed their respective residual deviation limits and are maintained for a continuous confirmation period, the normal recovery period of the health record is obtained by subtracting the time of the decline of the influence from the start time of the continuous period. Using the continuous maintenance condition can avoid prematurely identifying that the health has recovered when the behavioral deviation rises again after a brief decline.

[0140] Sort the duration of normal deviations in all health records from smallest to largest, multiply the sample size by 95% and round up, and read the value at the corresponding sort position to obtain the upper limit of the duration of normal deviations. Then, process all normal recovery times in the same way to obtain the upper limit of the normal recovery time.

[0141] The continuous confirmation duration, the upper limit of normal deviation duration, and the upper limit of normal recovery duration together form the recovery benchmark data. The recovery benchmark data is formed by statistical analysis of the actual recovery process of healthy fish fry under similar aquaculture operations. It can limit the allowable duration of normal deviation and the time range required for the recovery of normal behavior, which helps to eliminate short-term fluctuations and avoid directly using fixed experience values ​​from other fish species and other aquaculture stages.

[0142] S3.4. Collect the response delay data, normal response data, and recovery baseline data to obtain the normal operation response baseline;

[0143] The processing terminal writes the associated environmental indicators, sensor locations and response delay values ​​in the response delay data, the baseline environmental response sequence, baseline behavioral response sequence, the upper limit of the remaining deviation and the upper limit of environmental stability corresponding to each behavioral feature in the normal response data, and the continuous confirmation duration, the upper limit of normal deviation duration and the upper limit of normal recovery duration in the recovery baseline data into the same baseline record to form a normal operation response baseline.

[0144] The normal operation response benchmark records both the breeding batch stage and the breeding operation type, which is used to subsequently match the normal environmental response, normal behavioral response, allowable fluctuation range, and normal recovery range corresponding to the same breeding stage and similar breeding operations.

[0145] As an optional implementation method, the same aquaculture operation will produce different environmental changes under different operating intensities. Directly applying a fixed behavior curve can easily lead to the judgment of abnormal fish fry due to differences in operating range. This implementation method can solve the above situation, specifically as follows:

[0146] S4. Match the aquaculture operation data and environmental data with the normal operation response benchmark to obtain the expected behavior data;

[0147] Response matching refers to selecting a baseline record corresponding to the current aquaculture operation from the normal operation response baseline, and then correcting the normal behavior response according to the amplitude relationship between the current environmental response and the baseline environmental response.

[0148] S4.1 Match the aquaculture operation data with the normal operation response benchmark to obtain the matching benchmark record;

[0149] The processing terminal reads the operation type, start time, and batch stage of the current aquaculture operation from the aquaculture operation data. Then, it searches for the corresponding record in the normal operation response baseline according to the batch stage and operation type. Each record in the normal operation response baseline records the batch stage, operation type, associated environmental indicators, sensor location, response delay value, baseline environmental response sequence, and baseline behavioral response sequence.

[0150] Once a record with the same operation type is found, that record is designated as the matching baseline record. This avoids mixing the normal responses corresponding to oxygenation, water exchange, and water flow regulation, ensuring that the current aquaculture operation uses the normal response formed by the same type of health record.

[0151] S4.2. Using the matching benchmark record, perform current response extraction on the environmental data to obtain the current environmental response sequence;

[0152] The processing terminal reads the associated environmental indicators, sensor locations, and response delay values ​​from the matching benchmark record, and extracts the associated environmental indicators collected at the corresponding sensor locations from the environmental data.

[0153] Subsequently, the response delay value is added to the start time of the current aquaculture operation to obtain the current response start time. Starting from this time, environmental data is extracted according to the duration of the baseline environmental response sequence in the matching baseline record to form the current response period.

[0154] Then, select the period of the most recent water circulation time before the start of the current aquaculture operation, the period during which no other aquaculture operations have occurred and the relevant environmental indicators are within the aquaculture management range, as the stable period before the operation. Add up the environmental sampling values ​​within this period and divide by the number of sampling values ​​to obtain the average environmental value before the operation.

[0155] The current environmental response value at each sampling time is obtained by subtracting the average environmental value before the operation from the environmental sample value within the current response period. All current environmental response values ​​are then arranged according to the sampling time to form the current environmental response sequence.

[0156] This process can separate the actual environmental changes caused by the current aquaculture operation from the original environmental values, thus avoiding the influence of the original values ​​of environmental indicators on the judgment of the response magnitude.

[0157] S4.3. Compare the response magnitude of the current environmental response sequence with the matching benchmark record to obtain the response adjustment ratio;

[0158] The processing terminal calculates the absolute value of each current environmental response value in the current environmental response sequence and takes the maximum value as the current environmental response amplitude.

[0159] Subsequently, the absolute values ​​of each value in the baseline environmental response sequence in the matching baseline record are calculated, and the maximum value is taken as the baseline environmental response amplitude.

[0160] The response adjustment ratio is obtained by dividing the current environmental response magnitude by the baseline environmental response magnitude.

[0161] When the response adjustment ratio is greater than 1, it indicates that the environmental response caused by this operation is stronger than the baseline response; when the response adjustment ratio is less than 1, it indicates that the environmental response caused by this operation is weaker than the baseline response.

[0162] When the baseline environmental response amplitude is 0, the response adjustment ratio is set to 1.

[0163] By comparing the environmental response amplitude, we can reflect the actual effect of the current aquaculture operation in the current nursery pond, and avoid inferring changes in fish fry behavior based solely on the intensity of the equipment settings.

[0164] S4.4. Correct the baseline behavior response sequence in the matching baseline record using the aforementioned response adjustment ratio to obtain the expected behavior data;

[0165] The processing terminal multiplies the behavioral response value at each time position in the baseline behavioral response sequence by the response adjustment ratio to obtain the expected behavioral response value at the corresponding time position.

[0166] Subsequently, the expected behavioral response values ​​are arranged according to their relative time position in the baseline behavioral response sequence based on the start time of the current aquaculture operation, generating expected behavioral data. Expected behavioral data includes the nursery pond number, aquaculture batch stage, aquaculture operation type, start time of the aquaculture operation, behavioral characteristic name, prediction time, and corresponding expected behavioral response value, used to represent the behavioral changes that healthy fish fry should typically exhibit under the current aquaculture operation and environmental response conditions.

[0167] This method corrects normal behavioral responses based on the actual degree of environmental change caused by the operation, which can distinguish normal behavioral changes caused by different operational intensities and helps reduce the misjudgment of normal aquaculture operations as abnormal fish fry.

[0168] As an optional implementation method, the measured behavioral data includes both changes caused by normal breeding operations and changes that cannot be explained by normal operations. Directly judging based on the measured values ​​is prone to false alarms. This implementation method can solve the above situation, specifically as follows:

[0169] S5. Perform deviation separation on the measured behavior data and the expected behavior data to obtain the remaining deviation data;

[0170] Deviation separation refers to first converting the measured behavioral data into actual behavioral changes relative to the steady state before the operation, then comparing it with the expected behavioral data, and deducting the normal fluctuations allowed in the health record.

[0171] S5.1 Perform behavior response conversion on the measured behavior data to obtain measured behavior response data;

[0172] The processing terminal uses the start time of the aquaculture operation corresponding to the expected behavior data as the time reference, and determines the period before the operation as the stable behavior period of the most recent water circulation time and the period during which no other aquaculture operations have occurred.

[0173] Subsequently, the average values ​​of each behavioral characteristic during the stable period before the operation were calculated. For each measured behavioral value after the operation began, the measured behavioral response value at the corresponding time was obtained by subtracting the average value of the same behavioral characteristic before the operation from the measured behavioral value.

[0174] Then, the measured behavioral response values ​​are arranged according to the behavioral feature name and the collection time to form the measured behavioral response data.

[0175] The above conversion can remove the influence of the original behavioral level at different breeding periods, so that the measured behavioral changes and the expected behavioral changes are at the same comparison benchmark.

[0176] S5.2 Perform deviation calculation on the measured behavioral response data and the expected behavioral data to obtain behavioral deviation data;

[0177] The processing terminal matches the measured behavior response value with the corresponding expected behavior value according to the behavior feature name and time location.

[0178] Subsequently, the expected behavior value is subtracted from the measured behavior response value and the absolute value is taken to obtain the behavior deviation value. The behavior deviation values ​​are then arranged according to the behavior feature name and time position to form behavior deviation data.

[0179] Behavioral deviation values ​​only indicate the degree to which actual behavioral changes deviate from expected changes, without distinguishing the direction of deviation. Using absolute values ​​can simultaneously identify situations where actual behavioral changes are higher or lower than expected, avoiding the omission of abnormal deviations in different directions.

[0180] S5.3. Perform over-limit separation on the behavior deviation data based on the normal operation response benchmark to obtain the remaining deviation data;

[0181] The processing terminal reads the corresponding upper limit of the remaining deviation from the normal operation response benchmark according to the behavior feature name, and compares each behavior deviation value with the upper limit of the remaining deviation of the same behavior feature.

[0182] When the behavior deviation value does not exceed the upper limit of the remaining deviation, the corresponding remaining deviation is recorded as 0; when the behavior deviation value exceeds the upper limit of the remaining deviation, the behavior deviation value is subtracted from the upper limit of the remaining deviation to obtain the corresponding remaining deviation.

[0183] Subsequently, the remaining deviations were arranged according to behavioral characteristic name and time position to form residual deviation data. This processing can remove the normal fluctuations that are allowed in healthy fish fry under similar farming operations, and only retain behavioral differences that cannot be fully explained by normal operation responses, which helps to reduce false alarms caused by normal farming operations and short-term behavioral fluctuations.

[0184] As an optional implementation method, the resulting changes in water quality and flow may persist after the aquaculture operation is completed. If the recovery of fish fry behavior is analyzed before the environmental impact has subsided, the ongoing environmental effects may be mistaken for slow recovery of the fish fry. Therefore, this invention proposes the following working method:

[0185] S6. Identify the time of impact decline based on the aquaculture operation data, environmental data and normal operation response benchmark, analyze the remaining deviation data after the time of impact decline, and obtain recovery characteristic data;

[0186] Recovery process analysis refers to first confirming that the environmental changes caused by aquaculture operations have returned to the normal fluctuation range, and then calculating the duration of the remaining deviation and the time required for behavioral recovery.

[0187] S6.1. Based on the aforementioned aquaculture operation data, environmental data, and normal operation response benchmark, perform impact regression identification to obtain the impact regression time;

[0188] The processing terminal reads the operation type, start time, end time, nursery pond number, and breeding batch stage of the current breeding operation from the breeding operation data. Subsequently, according to the breeding batch stage and operation type, it reads the associated environmental indicators, sensor locations, environmental stability upper limits, and continuous confirmation duration from the normal operation response benchmark.

[0189] Then, the associated environmental indicators collected by the corresponding seedling pond number and sensor location are extracted from the environmental data. The period of the most recent water circulation time before the start of the aquaculture operation, when no other aquaculture operations have occurred and the associated environmental indicators are within the aquaculture management range, is selected as the stable period before the operation. The environmental sampling values ​​within this period are added together and divided by the number of sampling values ​​to obtain the average environmental value before the operation.

[0190] After the aquaculture operation is completed, the environmental deviation value is obtained by subtracting the average environmental value before the operation from the associated environmental index value at each sampling time and taking the absolute value.

[0191] If the environmental deviation value does not exceed the upper limit of environmental stability and is maintained for a continuous confirmation period, the start time of the continuous period is determined as the time when the impact subsides after the confirmation conditions are met.

[0192] This method not only confirms that the aquaculture operation has been completed, but also confirms that the resulting environmental changes have returned to the normal fluctuation range, which helps to avoid prematurely judging the recovery status of fish fry while the effects of water quality and water flow are still ongoing.

[0193] S6.2. Extract the remaining deviation data at the time when the influence fades to obtain candidate recovery data;

[0194] The processing terminal uses the time when the impact fades as the starting point and the current analysis time as the candidate endpoint. When a new aquaculture operation occurs after the time when the impact fades, the start time of the new aquaculture operation is used as the candidate endpoint. Subsequently, the remaining deviation values ​​corresponding to each behavioral feature between the starting point and the candidate endpoint are read and arranged according to the behavioral feature name and the collection time to form candidate recovery data.

[0195] S6.3 Perform recovery boundary identification on the candidate recovery data to obtain recovery analysis data and the current recovery duration;

[0196] The processing terminal begins checking candidate recovery data from the moment the impact subsides. If the residual deviation of all behavioral characteristics is equal to 0 and this condition is maintained for a continuous confirmation duration, the start time of that continuous period is determined as the recovery completion time.

[0197] When determining the recovery completion time, candidate recovery data between the time of impact decay and the time of recovery completion are selected as recovery analysis data, and the current recovery duration is obtained by subtracting the time of impact decay from the time of recovery completion.

[0198] If the recovery completion time has not been determined as of the current analysis time, and no new aquaculture operations have occurred after the impact subsides, all candidate recovery data will be determined as recovery analysis data, and the current recovery duration will be obtained by subtracting the impact subsides time from the current analysis time.

[0199] If a new aquaculture operation occurs before the recovery completion time is determined, the recovery analysis of the current aquaculture operation is stopped, and the data interrupted by the new aquaculture operation is not used to generate recovery feature data.

[0200] S6.4 Perform persistent deviation analysis on the recovery analysis data and aggregate it to form recovery feature data;

[0201] The processing terminal identifies time periods in the recovery analysis data where the residual deviation is continuously greater than 0, based on the behavioral feature name. The duration of multiple discontinuous time periods is calculated separately and not accumulated.

[0202] Subsequently, the maximum value among the durations of deviations corresponding to all behavioral features is taken to obtain the current duration of deviation; if no period of continuous deviation greater than 0 is identified, the current duration of deviation is recorded as 0.

[0203] The seedling pond number, breeding batch stage, operation type, current deviation duration, and current recovery duration are written into the same record to form recovery characteristic data.

[0204] As an optional implementation method, simply determining whether a residual deviation exists at a certain moment can easily lead to the judgment of instantaneous behavioral fluctuations as continuous abnormalities, and it is also difficult to identify situations where the residual deviation persists and the behavioral recovery time exceeds the normal range. This implementation method can solve the above problems, specifically as follows:

[0205] S7. Based on the normal operation response benchmark, perform recovery anomaly determination on the recovery feature data and generate monitoring results.

[0206] The abnormal recovery determination refers to judging whether the duration of behavioral deviation and the time of behavioral recovery exceed the normal upper limit corresponding to the same type of breeding operation and the same breeding batch stage, and determining the risk of continuous behavioral decline when both of the two exceeding conditions are met.

[0207] S7.1 Extract recovery determination parameters from the normal operation response benchmark to obtain the recovery determination benchmark;

[0208] The processing terminal reads the aquaculture operation type and aquaculture batch stage from the recovery feature data, and extracts recovery judgment parameters from the normal operation response benchmark according to the aquaculture operation type and aquaculture batch stage to limit the duration of behavioral deviation and behavioral recovery time.

[0209] The parameter used to limit the duration of behavioral deviation is the upper limit of normal deviation duration, and the parameter used to limit the behavioral recovery time is the upper limit of normal recovery duration. These two parameters, along with their corresponding operation types and breeding batch stages, are recorded in the same record to obtain the recovery judgment benchmark.

[0210] S7.2. Compare the recovery feature data with the recovery judgment benchmark for execution time to obtain the recovery comparison result;

[0211] The processing terminal reads the current deviation duration, which represents the duration of behavioral deviation, from the recovery feature data, and compares it with the upper limit of normal deviation duration in the recovery judgment benchmark to obtain the deviation duration comparison result.

[0212] Subsequently, the current recovery time of the characterizing behavior recovery time is read and compared with the upper limit of the normal recovery time in the recovery judgment benchmark to obtain the recovery time comparison result.

[0213] If the current deviation duration exceeds the upper limit of the normal deviation duration, the deviation duration comparison result is "deviation duration exceeds the limit"; otherwise, the deviation duration does not exceed the limit. If the current recovery duration exceeds the upper limit of the normal recovery duration, the recovery duration comparison result is "recovery timeout"; otherwise, the recovery does not time out.

[0214] The results of the deviation persistence comparison and the recovery time comparison are aggregated to obtain the recovery comparison results.

[0215] S7.3 Perform a joint determination on the recovery comparison results to obtain the recovery anomaly determination result;

[0216] When the recovery comparison results include both persistent deviation exceeding the limit and recovery timeout, a result indicating that the risk of persistent behavior decay is established is generated.

[0217] If at least one of the two conditions, namely, persistent deviation exceeding the limit and recovery timeout, is not met, a result indicating that the risk of persistent behavior decay is not met is generated. The result indicating that the risk of persistent behavior decay is not met only means that the two exceeding conditions are not met simultaneously, and does not mean that there are no other abnormal farming conditions in the fish fry.

[0218] S7.4 Generate monitoring results based on the recovery anomaly determination results;

[0219] The processing terminal reads the seedling pond number, breeding batch stage, and operation type from the recovery feature data, and writes the above information, along with the judgment time and recovery anomaly judgment result, into the same record to form the monitoring result.

[0220] The monitoring results only indicate the risk of aquaculture status and do not output specific disease names or disease conclusions. By making the continuous deviation exceeding the limit and the recovery time exceeding the limit the judgment conditions that must be met simultaneously, short-term behavioral fluctuations that recover quickly can be excluded, and situations can be identified where environmental indicators are still within the management range, but the fish fry behavior continues to deviate and the recovery time is abnormally prolonged. This helps to reduce false alarms and lower the probability of missing early persistent anomalies.

[0221] like Figure 2As shown, this invention demonstrates superior performance in terms of time mismatch control, false alarm control, continuous anomaly identification, and recovery analysis accuracy. Specifically, it first establishes a response time delay relationship between changes in aquaculture operations and environmental responses to reduce time mismatches caused by water propagation; then, it corrects expected behavior based on the current environmental response amplitude and separates behavioral deviations within the normal fluctuation range of healthy fish fry to reduce false alarms caused by normal aquaculture operations; finally, after the environmental impact subsides, it statistically analyzes the duration and recovery time of behavioral deviations, and outputs risk when both deviation exceeding limits and recovery timeouts are simultaneously established, thereby improving the accuracy of continuous anomaly identification and recovery status analysis.

[0222] The embodiments of the present invention described above are subject to modification and change of method by those skilled in the art without departing from the embodiments and broader aspects of the present invention. The appended claims are intended to include all such modifications and changes of method that do not depart from the present invention.

[0223] Those skilled in the art should understand that the embodiments of the present invention can be implemented using a pure hardware architecture, a pure software architecture, or an integrated hardware and software architecture. The present invention can be prepared as a computer program product, which can be stored in various non-volatile computer-readable storage media, including but not limited to solid-state drives, flash memory chips, mobile storage devices, optical discs, cloud storage servers, and other standardized storage media, and is not limited to traditional storage media.

[0224] Based on the foregoing description in conjunction with the accompanying drawings, those skilled in the art will understand that the embodiments of this application can also be implemented by software programs. Therefore, this application also provides a computer-readable storage medium. This computer-readable storage medium stores computer-readable instructions thereon, which, when executed by one or more processors, implement the method described above in conjunction with the accompanying drawings.

[0225] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0226] It should be noted that although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart can be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0227] It should be understood that when the terms "first," "second," "third," and "fourth," etc., are used in the claims, specification, and drawings of this application, they are used only to distinguish different objects and not to describe a specific order. The terms "comprising" and "including" as used in the specification and claims of this application indicate the presence of the described features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof.

[0228] It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this specification and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this specification and claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations.

[0229] Although the embodiments of this application are described above, the content is merely an example adopted for the purpose of facilitating understanding of this application and is not intended to limit the scope and application scenarios of this application. Any person skilled in the art described in this application may make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in this application, but the scope of patent protection of this application shall still be determined by the scope defined in the appended claims.

Claims

1. A method for monitoring the health status of eel fry based on multi-sensor data, characterized in that, include: Environmental data, measured behavioral data, and aquaculture operation data are collected and aggregated into monitoring data. The monitoring data is analyzed to determine the corresponding execution time, and the response delay data is obtained. Based on the response delay data, the monitoring data is calibrated to obtain the normal operation response benchmark; The aquaculture operation data and environmental data are matched with the normal operation response benchmark to obtain the expected behavior data; Deviation separation is performed between the measured behavioral data and the expected behavioral data to obtain the remaining deviation data; Based on the aquaculture operation data, environmental data, and normal operation response benchmark, the time of impact decline is identified, and the remaining deviation data after the time of impact decline is analyzed to obtain recovery characteristic data. Based on the normal operation response benchmark, the recovery feature data is subjected to recovery anomaly determination, and monitoring results are generated.

2. The method for monitoring the health status of eel fry according to claim 1, characterized in that, The monitoring data is analyzed to determine the corresponding execution time, resulting in response delay data, including: Operational change extraction is performed on the aquaculture operation data to obtain the operation change sequence; Environmental change extraction is performed on the environmental data to obtain an environmental change sequence; Perform time delay correlation analysis on the operation change sequence and the environment change sequence to obtain time delay correlation results that include candidate time delays and correlation degrees; The latency-related results are correlated and confirmed to obtain response latency data.

3. The method for monitoring the health status of eel fry according to claim 1, characterized in that, Based on the response delay data, the monitoring data is calibrated to obtain a normal operation response benchmark, including: Acquire breeding status records, correct monitoring data according to the aforementioned response delay data, and filter health records; Response feature summarization is performed on the health records to obtain normal response data; The recovery process of health records is analyzed based on the normal response data to obtain recovery baseline data; The response delay data, normal response data, and recovery baseline data are collected to obtain the normal operation response baseline.

4. The method for monitoring the health status of eel fry according to claim 1, characterized in that, The aquaculture operation data and environmental data are matched with the normal operation response benchmark to obtain expected behavior data, including: The aquaculture operation data is matched with the normal operation response benchmark to obtain a matching benchmark record, which includes a benchmark environmental response sequence and a benchmark behavioral response sequence. The current response sequence is obtained by performing current response extraction on the environmental data using the matching benchmark record. The response magnitude is compared with the current environmental response sequence and the matching baseline record to obtain the response adjustment ratio; The baseline behavior response sequence in the matching baseline record is corrected using the aforementioned response adjustment ratio to obtain the expected behavior data.

5. The method for monitoring the health status of eel fry according to claim 1, characterized in that, The measured behavioral data and the expected behavioral data are separated to obtain the remaining deviation data, including: Perform behavior response conversion on the measured behavior data to obtain measured behavior response data; Deviation calculation is performed on the measured behavioral response data and the expected behavioral data to obtain behavioral deviation data; The behavioral deviation data is subjected to over-limit separation based on the normal operation response benchmark to obtain the remaining deviation data.

6. The method for monitoring the health status of eel fry according to claim 1, characterized in that, Based on the aforementioned aquaculture operation data, environmental data, and normal operation response benchmark, the time of impact decline is identified. The remaining deviation data after the time of impact decline is analyzed to obtain recovery characteristic data, including: Based on the aforementioned aquaculture operation data, environmental data, and normal operation response benchmarks, impact regression identification is performed to obtain the impact regression time. The remaining deviation data is extracted at the time when the influence fades to obtain candidate recovery data; Perform recovery boundary identification on the candidate recovery data to obtain recovery analysis data and the current recovery duration; A persistent deviation analysis is performed on the recovery analysis data, and the data is aggregated to form recovery feature data.

7. The method for monitoring the health status of eel fry according to claim 1, characterized in that, Based on the normal operation response benchmark, the recovery feature data is subjected to recovery anomaly determination, and monitoring results are generated, including: Recovery determination parameters are extracted from the normal operation response benchmark to obtain the recovery determination benchmark; The recovery feature data is compared with the recovery judgment benchmark in terms of execution time to obtain the recovery comparison result; A joint determination is performed on the recovery comparison results to obtain the recovery anomaly determination result; Based on the recovery anomaly determination results, monitoring results are generated.

8. The method for monitoring the health status of eel fry according to claim 2, characterized in that, Perform correlation confirmation on the latency-related results to obtain response latency data, including: From the monitoring data, stable periods in which no aquaculture operations were recorded and environmental data were within the permissible range of aquaculture management were selected, and stable environmental change sequences were extracted; Perform correlation analysis on the stable environment change sequence and the operation change sequence to obtain a set of background correlation degrees; Perform quantile statistics on the set of background relevance to obtain the association confirmation threshold; The environmental indicators and candidate delays corresponding to the highest relevant records that exceed the association confirmation threshold are respectively determined as the associated environmental indicators and response delay values, forming response delay data.

9. The method for monitoring the health status of eel fry according to claim 4, characterized in that, The current environmental response sequence is compared with the matching baseline record to obtain the response adjustment ratio, including: The absolute value of the current environmental response value in the current environmental response sequence is taken, and the maximum value is obtained to obtain the current environmental response amplitude. Extract the baseline environmental response sequence from the matching baseline record, take the absolute value of the values ​​and the maximum value to obtain the baseline environmental response amplitude; The response adjustment ratio is obtained by dividing the current environmental response amplitude by the baseline environmental response amplitude; when the baseline environmental response amplitude is 0, the response adjustment ratio is set to 1.

10. The method for monitoring the health status of eel fry according to claim 7, characterized in that, A joint determination is performed on the recovery comparison results to obtain a recovery anomaly determination result, including: Extract the deviation persistence comparison result and the recovery duration comparison result from the recovery comparison result; When the deviation persistence comparison result is that the deviation persists beyond the limit and the recovery time comparison result is that the recovery timeout occurs, a result indicating that the risk of persistent behavior decay is established is generated. If at least one of the above two conditions is not met, the result "the risk of persistent behavior decay is not met" is generated. The result of whether the risk of persistent behavior decay is established or not is determined as the result of recovery anomaly determination.

Images