Early warning method, device and equipment based on aquatic disease and storage medium
By training a normal baseline model of aquaculture species and analyzing environmental characteristics, the future trends of stress-inducing parameters are predicted, triggering early warning signals. This solves the problem of untimely disease early warning in aquaculture and achieves accuracy and cost-effectiveness in early warning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING TIMES KEHUA TECH CO LTD
- Filing Date
- 2025-10-13
- Publication Date
- 2026-07-03
AI Technical Summary
Current technologies often fail to provide timely disease warnings for aquaculture, which can easily lead to large-scale disease outbreaks or misjudgments that result in unnecessary interventions and increased operating costs.
A normal baseline model for aquaculture subjects is trained based on historical health data. Through multidimensional behavioral and environmental characteristic analysis, anomaly scores and stress trigger parameters are determined. Long short-term memory networks are used to predict future trends and trigger early warning signals.
It improves the accuracy and reliability of early warning of aquatic diseases, reduces unnecessary interventions, saves operating costs, and provides effective intervention time.
Smart Images
Figure CN121303565B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of aquaculture technology, specifically to an early warning method, device, equipment, and storage medium for aquatic diseases. Background Technology
[0002] Aquaculture is an agricultural activity that uses scientific breeding techniques to raise, raise, and manage aquatic organisms to achieve high-efficiency production and economic benefits. Early warning systems can improve the management efficiency of aquaculture. If early warnings for aquaculture species are not timely, the optimal time for prevention and control will be missed, leading to large-scale disease outbreaks and economic losses for farmers. Conversely, misinterpreting momentary behavioral fluctuations as urgent risks may prompt farmers to take unnecessary interventions, resulting in waste of electricity, materials, and manpower, and increasing operating costs. Therefore, overcoming these technical problems and shortcomings is a key issue that needs to be addressed. Summary of the Invention
[0003] To overcome the aforementioned problems in the prior art, this application provides a method, apparatus, device, and storage medium for early warning of aquatic diseases, employing the following technical solution:
[0004] Firstly, this application provides an early warning method for aquatic diseases, including:
[0005] Based on historical health data, a normal baseline model of aquaculture subjects is trained to determine the boundaries of normal behavior.
[0006] The real-time multidimensional behavioral characteristics of aquaculture objects are input into the normal baseline model to obtain the abnormality score of the aquaculture objects. When the abnormality score of the aquaculture objects exceeds the preset threshold, an abnormal behavior event is generated, which includes the time window in which the abnormal behavior occurs.
[0007] Environmental data from the N hours preceding the time window is obtained from environmental features. A subset of lagged environmental data is constructed. Based on feature importance analysis, the environmental parameters that contribute the most in the subset of lagged environmental data are obtained.
[0008] A causal relationship test is performed on the environmental parameters. If the causal relationship test is passed, the environmental parameters are used as stress precipitate parameters.
[0009] Based on the environmental characteristics of aquaculture species, the trend of stress-inducing parameters within a preset time period is predicted. When the stress-inducing parameters fall below the preset safe range, an early warning signal is triggered.
[0010] Furthermore, based on historical health data, a normal baseline model of aquaculture subjects is trained to determine the boundaries of normal behavior, including:
[0011] We acquire multidimensional behavioral characteristic data of aquaculture objects under health conditions, input the multidimensional behavioral characteristic data into the isolated forest model, and construct multiple random decision trees in parallel.
[0012] Each random decision tree randomly selects features and segments from the multidimensional behavioral feature data;
[0013] When a sample point is placed into the tree from the root node of the random decision tree, it moves along the branches until it is isolated and becomes a leaf node, and the path length of the sample point in this tree is obtained.
[0014] The sample point is passed through all the random decision trees to obtain the path length on each tree;
[0015] Calculate the average of all path lengths, and determine a score threshold based on the abnormal score distribution of all healthy samples. This score threshold is the normal behavior boundary of the multidimensional behavioral feature space.
[0016] Furthermore, by using a time window, environmental data from the previous N hours is obtained from environmental characteristics, and a lagged environmental data subset is constructed, including:
[0017] Determine the start and end times of abnormal behavior events based on time windows;
[0018] Based on biological experience, the lag duration corresponding to different environmental parameters is preset, and the lag interval is extracted based on the start time.
[0019] The query retrieves data from environmental time series data whose timestamps fall within the lag interval. The queried data is then sorted according to time order to obtain a subset of the lagged environmental data.
[0020] Furthermore, based on feature importance analysis, the environmental parameters that contribute the most to the lagged environmental data subset are obtained, including:
[0021] An anomalous indicator from an abnormal behavioral event is selected as the explanatory variable, serving as the first data. Time series of all environmental parameters to be tested are extracted from a subset of lagged environmental data, serving as the second data. The first and second data are time-aligned. The cross-correlation function for each pair of first and second data is calculated. For each environmental parameter, the point with the highest absolute value in the cross-correlation function curve is obtained. When the point with the highest absolute value appears in the leading region of the environmental parameter, and within that leading region, the absolute value of the correlation coefficient obtained based on the cross-correlation function exceeds a preset significance threshold, then the environmental parameter is considered to have the greatest contribution.
[0022] Furthermore, a causal relationship test is performed on the environmental parameters. When the causal relationship test is passed, the environmental parameters are used as stress trigger parameters. This includes: outputting a causal graph and a causal weight matrix containing all environmental and behavioral parameters based on a preset causal model; taking the abnormal indicators in the abnormal behavioral events as target variables, tracing back along all directed edges directly pointing to the abnormal indicator nodes in the causal graph, and outputting a list of direct cause nodes, which contains the environmental parameters on the causal graph that directly drive the changes in the abnormal indicators; extracting the driving weight of each environmental parameter on the abnormal indicator from the list of direct cause nodes from the causal weight matrix, and determining the environmental parameters corresponding to the driving weights that meet the preset threshold as stress trigger parameters.
[0023] Furthermore, based on the environmental characteristics of aquaculture species, the trend of stress-inducing parameters over a preset time period is predicted. When the stress-inducing parameters fall below a preset safe range, an early warning signal is triggered, including:
[0024] The established stress trigger parameters are used as the target sequence.
[0025] Obtain the historical sequence of the identified stress trigger parameters over a preset time period, as well as the sequence of other environmental parameters that have a direct influence on the identified stress trigger parameters;
[0026] The historical sequence and other environmental parameter sequences are divided into several input sequences using a sliding window, and then divided into several target sequences based on the time step of the sliding window.
[0027] Data pairs are generated by combining several input sequences and several target sequences. Several input sequences are then fed into a pre-defined long short-term memory network to obtain a corresponding number of predicted values. Based on the loss between the predicted values and the target sequences, the parameters of the long short-term memory network are adjusted so that the predicted values continuously approach the target sequences. When the loss between the predicted values and the target sequences meets a preset threshold, the training of the long short-term memory network is completed.
[0028] Determine the warning threshold of the stress trigger parameter, input the real-time observation data of the current time period at a preset time step into the long short-term memory network, and obtain the predicted trend curve for the future preset time period;
[0029] When the predicted value at any point on the trend curve falls below the preset range of the warning threshold, an advance warning is immediately triggered.
[0030] The spatial clustering degree of anomalous individuals in the water body segmentation network is obtained, anomaly heatmaps are generated, hotspot areas are located, and hotspot areas are associated with their corresponding local environments.
[0031] Secondly, this application also provides an early warning device for aquatic diseases, comprising:
[0032] The abnormal score acquisition module is used to train a normal baseline model of aquaculture objects based on historical health data and determine the boundaries of normal behavior.
[0033] The abnormal behavior event generation module is used to input the real-time multidimensional behavioral characteristics of aquaculture objects into the normal baseline model to obtain the abnormal score of aquaculture objects. When the abnormal score of aquaculture objects exceeds a preset threshold, an abnormal behavior event is generated, which includes the time window in which the abnormal behavior occurs.
[0034] The environmental parameter acquisition module is used to obtain environmental data from environmental features within the previous N hours through a time window, construct a lagged environmental data subset, and obtain the environmental parameters that contribute the most in the lagged environmental data subset based on feature importance analysis.
[0035] The stress trigger parameter determination module is used to test the causal relationship of environmental parameters. When the causal relationship test is passed, the environmental parameters are used as stress trigger parameters.
[0036] The early warning signal triggering module is used to predict the trend of stress causative parameters within a preset time period based on the environmental characteristics of aquaculture objects. When the stress causative parameters fall below the preset safe value range, an early warning signal is triggered.
[0037] Thirdly, this application provides an electronic device, comprising:
[0038] One or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions that, when executed by the device, cause the device to perform the method as described in the first aspect.
[0039] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to perform the method described in the first aspect.
[0040] Fifthly, this application provides a computer program that, when executed by a computer, performs the method described in the first aspect.
[0041] In one possible design, the program in the fifth aspect can be stored wholly or partially on a storage medium packaged with the processor, or it can be stored wholly or partially on a memory not packaged with the processor.
[0042] This application has the following beneficial effects:
[0043] 1. This application trains a normal baseline model of aquaculture subjects based on historical health data to determine the boundaries of normal behavior; inputs the real-time multidimensional behavioral characteristics of aquaculture subjects into the normal baseline model to obtain the abnormal scores of aquaculture subjects; when the abnormal scores of aquaculture subjects exceed a preset threshold, an abnormal behavior event is generated, wherein the abnormal behavior event includes the time window of the abnormal behavior. This application determines abnormal behavior events by using time windows, eliminating abnormal behavioral fluctuations caused by short-term factors, reducing the frequency of early warning and forecasting, and providing effective intervention time for early warning before the aquaculture subject group shows abnormal behavior.
[0044] 2. This application obtains environmental data from the preceding N hours of a time window from environmental characteristics, constructs a lagged environmental data subset, and, based on feature importance analysis, identifies the environmental parameter that contributes the most to the lagged environmental data subset. A causal relationship test is performed on the environmental parameter; if the causal relationship test is passed, the environmental parameter is used as a stress-inducing parameter. Based on the environmental characteristics of the aquaculture species, the trend of the stress-inducing parameter within a preset time period is predicted. When the stress-inducing parameter falls below a preset safe value range, an early warning signal is triggered. This application uses dual verification to determine the stress-inducing parameter, further ensuring the accuracy of early warning. Attached Figure Description
[0045] Figure 1 This is a flowchart of an early warning method for aquatic diseases based on an embodiment of this application;
[0046] Figure 2 This is a flowchart illustrating the determination of normal behavior boundaries in an early warning method for aquatic diseases based on an embodiment of this application.
[0047] Figure 3 This is a flowchart illustrating the process of acquiring a subset of lagging environmental data for an early warning method for aquatic diseases based on an embodiment of this application.
[0048] Figure 4 The flowchart for obtaining environmental parameters that contributes most to the embodiments of this application is shown.
[0049] Figure 5 This is a flowchart of the apparatus according to an embodiment of this application;
[0050] Figure 6 This is a schematic diagram of a computer device according to an embodiment of this application. Detailed Implementation
[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.
[0052] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0053] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.
[0054] Please refer to Figure 1 The following is a flowchart of an early warning method for aquatic diseases provided in this application embodiment. The specific implementation steps are as follows:
[0055] Step S1: Based on historical health data, train a normal baseline model for aquaculture subjects to determine the boundaries of normal behavior.
[0056] In this embodiment, a normal baseline model of aquaculture organisms is trained based on historical health data to determine the boundaries of normal behavior. Please refer to [reference needed]. Figure 2 ,include:
[0057] Step 11: Obtain multidimensional behavioral characteristic data of aquaculture objects under health status, input the multidimensional behavioral characteristic data into the isolated forest model, and construct multiple random decision trees in parallel;
[0058] Step 12: Each random decision tree randomly selects features and segments from the multidimensional behavioral feature data;
[0059] Step 13: When a sample point is placed into the tree from the root node of the random decision tree, it moves along the branches until it is isolated and becomes a leaf node, and obtains the path length of the sample point in this tree.
[0060] Step 14: Pass the sample point through all the random decision trees and obtain the path length on each tree;
[0061] Step 15: Calculate the average of all path lengths and determine the score threshold based on the abnormal score distribution of all healthy samples. This score threshold is the normal behavior boundary of the multidimensional behavioral feature space.
[0062] This application learns the dense distribution pattern of healthy sample data in the feature space by calculating the average path length of all sample points.
[0063] It should be noted that multidimensional behavioral feature data is a data record composed of all behavioral features at a specific point in time. Isolation forest determines whether a sample is normal or abnormal by analyzing the position of this multidimensional behavioral feature data in the feature space.
[0064] In this embodiment, historical health data includes video data and acoustic data. The video data can collect the behavior of aquaculture organisms, while the acoustic data can collect acoustic signals from them. The behavior of the aquaculture organisms includes swimming speed, aggregation index, activity area distribution, feeding behavior, and density. The acoustic signals include the frequency of abnormal friction sounds and breathing noises.
[0065] It should be noted that this application collects acoustic signals from aquaculture organisms by deploying hydrophones and collects their behavior by installing underwater high-definition cameras, which cover the main activity areas of the aquaculture organisms. Internet of Things (IoT) sensors are installed at different depths and key locations within the aquaculture water.
[0066] In this embodiment of the application, a normal baseline model of aquaculture objects is trained to determine normal behavioral boundaries, which include swimming boundaries, spatial distribution boundaries, and individual physiological boundaries.
[0067] Step S2: Input the real-time multidimensional behavioral features of the aquaculture object into the normal baseline model to obtain the abnormal score of the aquaculture object. When the abnormal score of the aquaculture object exceeds the preset threshold, generate an abnormal behavior event, wherein the abnormal behavior event includes the time window in which the abnormal behavior occurs.
[0068] In this embodiment, if the time window for the abnormal behavior does not meet the preset threshold, step S3 is not performed. This reduces the likelihood of misjudging instantaneous behavioral fluctuations as emergency risks, helping operators reduce operational costs.
[0069] It should be noted that the preset threshold in step S2 refers to a score threshold. This application determines an abnormal behavior event by defining a time window in which the event occurs, rather than a single point in time, thereby further improving the reliability of the early warning. This eliminates abnormal behavioral fluctuations caused by transient factors, reducing the frequency of early warnings.
[0070] Step S3: Using a time window, obtain environmental data from the environmental features for the previous N hours, construct a subset of lagged environmental data, and based on feature importance analysis, obtain the environmental parameters that contribute the most in the subset of lagged environmental data.
[0071] In this embodiment, environmental data from the previous N hours is obtained from environmental features using a time window to construct a subset of lagging environmental data. Please refer to [reference needed]. Figure 3 ,include:
[0072] Step 31: Determine the start and end times of the abnormal behavior event based on the time window.
[0073] Step 32: Based on biological experience, preset the lag time corresponding to different environmental parameters, and extract the lag interval based on the start time.
[0074] Step 33: Query the data in the environmental time series data whose timestamps are within the lag interval, sort the queried data according to time order, and obtain the lag environmental data subset.
[0075] In this embodiment of the application, based on feature importance analysis, the environmental parameters that contribute the most in the subset of lagging environmental data are obtained, as shown in Figure 4, including:
[0076] Step 41: Select an abnormal indicator from the abnormal behavior event as an explanatory variable and use it as the first data.
[0077] Step 42: Extract the time series of all environmental parameters to be tested from the lagged environmental data subset as the second data.
[0078] Step 43: Time-align the first and second data. Calculate the cross-correlation function for each pair of first and second data.
[0079] Step 44: For each environmental parameter, obtain the point with the highest absolute value in the cross-correlation function curve.
[0080] Step 45: When the highest absolute value appears in the leading region of the environmental parameter, if the absolute value of the correlation coefficient obtained based on the cross-correlation function exceeds the preset significance threshold within the leading region, then the environmental parameter is judged to have the greatest contribution.
[0081] In this embodiment, the cross-correlation function refers to the degree of similarity between the abnormal indicators and environmental parameters under different time lags. The first data is the time series of the behavioral abnormal indicators, and the second data is the time series of the environmental parameters.
[0082] It should be noted that the leading nature of environmental parameters refers to changes in environmental parameters preceding changes in behavioral characteristics. This application, by determining that environmental parameters lead behavioral characteristics, can issue an early warning when environmental parameters deviate from safe values, even before the aquaculture organisms exhibit obvious behavioral abnormalities, thus providing an effective intervention window for the aquaculture organisms.
[0083] Step S4: Perform a causal relationship test on the environmental parameters. If the causal relationship test is passed, the environmental parameters are used as stress precipitate parameters.
[0084] It should be noted that the causal relationship test on the environmental parameters is to further verify the environmental precedence determined in step S3, and to prove that the environmental parameters are significantly related to the abnormal behavior in time. Only then are the environmental parameters identified as stress trigger parameters.
[0085] In this embodiment of the application, a causal relationship test is performed on environmental parameters. When the causal relationship test is passed, the environmental parameters are used as stress precipitate parameters, including:
[0086] Based on the preset causal model, the output includes a causal graph and a causal weight matrix containing all environmental and behavioral parameters.
[0087] Using the anomalous indicators in the abnormal behavior events as target variables, we trace back along all directed edges that directly point to the anomalous indicator nodes in the causal graph, and output a list of direct cause nodes. The list contains environmental parameters on the causal graph that directly drive the changes in the anomalous indicators.
[0088] Extract the driving weight of each environmental parameter in the list of direct cause nodes from the causal weight matrix. The environmental parameters corresponding to the driving weights that meet the preset threshold are identified as stress trigger parameters.
[0089] Step S5: Based on the environmental characteristics of the aquaculture species, predict the trend of stress causative parameters within a preset time period in the future. When the stress causative parameters fall below the preset safe value range, trigger an early warning signal.
[0090] In this embodiment of the application, based on the environmental characteristics of the aquaculture species, the trend of stress-inducing parameters within a preset time period is predicted. When the stress-inducing parameters fall below a preset safe value range, an early warning signal is triggered, including:
[0091] The determined stress precipitating parameters are defined as the target sequence. Assuming the determined stress precipitating parameter is dissolved oxygen, each value in the target sequence is a predicted value of dissolved oxygen.
[0092] It should be noted that the determined stress trigger parameter in this application embodiment is a time series variable, while the target sequence is a segment of this time series variable extending from the current time to a preset time step.
[0093] The process involves acquiring historical sequences of identified stress-inducing parameters over a preset time period, as well as sequences of other environmental parameters directly related to these parameters. In this embodiment, assuming the identified stress-inducing parameter is dissolved oxygen, sequences of other environmental parameters directly related to dissolved oxygen include key data such as water temperature and pH. These other environmental parameter sequences can improve prediction accuracy.
[0094] The historical sequence and other environmental parameter sequences are divided into several input sequences using a sliding window, and then divided into several target sequences based on the time step of the sliding window.
[0095] Data pairs are generated by combining several input sequences and several target sequences. The input sequences are then fed into a pre-defined long short-term memory network to obtain a corresponding number of predicted values. Based on the loss between the predicted values and the target sequences, the parameters of the long short-term memory network are adjusted so that the predicted values continuously approach the target sequences. When the loss between the predicted values and the target sequences meets a preset threshold, the training of the long short-term memory network is completed.
[0096] The warning threshold of the stress trigger parameter is determined, and the real-time observation data of the current time period at a preset time step is input into the long short-term memory network to obtain the predicted trend curve for the future preset time period. When the predicted value of any prediction point of the predicted trend curve falls below the preset range of the warning threshold, an advance warning is immediately triggered.
[0097] In this embodiment of the application, when the stress trigger parameter falls below the preset range of the safety value, the method further includes: obtaining the spatial aggregation degree of abnormal individuals in the water body division network, generating an abnormal heat map, locating hotspot areas, associating the hotspot areas with the corresponding local environment, and determining whether it is a global problem or a local problem.
[0098] It should be noted that the embodiments of this application provide early warning of impending risks, giving farmers time to intervene in advance and prevent the risks from developing further. This allows for early prevention of aquaculture, rather than waiting until the risks occur before taking treatment measures, thus further mitigating the occurrence of danger and helping farmers save aquaculture resources.
[0099] Example 2: Figure 5As shown, this application provides an early warning device for aquatic diseases, which can implement the early warning method for aquatic diseases in Embodiment 1. The device includes an abnormal score acquisition module 501, an abnormal behavior event generation module 502, an environmental parameter acquisition module 503, a stress trigger parameter determination module 504, and an early warning signal triggering module 505, wherein:
[0100] The abnormal score acquisition module 501 is used to train a normal baseline model of aquaculture objects based on historical health data and determine the boundaries of normal behavior.
[0101] The abnormal behavior event generation module 502 is used to input the real-time multidimensional behavioral characteristics of aquaculture objects into the normal baseline model to obtain the abnormal score of aquaculture objects. When the abnormal score of aquaculture objects exceeds a preset threshold, an abnormal behavior event is generated, wherein the abnormal behavior event includes the time window in which the abnormal behavior occurs.
[0102] The environmental parameter acquisition module 503 is used to acquire environmental data from environmental features within the previous N hours through a time window, construct a subset of lagged environmental data, and acquire the environmental parameters that contribute the most in the subset of lagged environmental data based on feature importance analysis.
[0103] The stress induced parameter determination module 504 is used to perform causal relationship testing on environmental parameters. When the causal relationship test is passed, the environmental parameters are used as stress induced parameters.
[0104] The early warning signal triggering module 505 is used to predict the trend of stress causative parameters within a preset time period based on the environmental characteristics of aquaculture objects. When the stress causative parameters fall below the preset safe value range, an early warning signal is triggered.
[0105] This application trains a normal baseline model of aquaculture subjects based on historical health data to determine the boundaries of normal behavior. Real-time multidimensional behavioral characteristics of the aquaculture subjects are input into the normal baseline model to obtain anomaly scores. When the anomaly score exceeds a preset threshold, an abnormal behavior event is generated. Environmental data from the preceding N hours are obtained from environmental features to construct a lagged environmental data subset. Based on feature importance analysis, the environmental parameter contributing the most to the lagged environmental data subset is obtained. A causal relationship test is performed on the environmental parameter; if the causal relationship test is passed, the environmental parameter is used as a stress trigger parameter. The trend of the stress trigger parameter within a preset time period is predicted. When the stress trigger parameter falls below a preset safe range, an early warning signal is triggered. This application provides an effective intervention time for early warning before the aquaculture subject group exhibits abnormal behavior.
[0106] Example 3: This application provides an electronic device, including: one or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions that, when executed by the device, cause the device to perform the method as described in Example 1.
[0107] Example 4: This application provides a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to perform the method described in Example 1.
[0108] Example 5: This application provides a computer program that, when executed by a computer, performs the method described in Example 1.
[0109] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 6 , Figure 6 This is a basic structural block diagram of the computer device in this embodiment.
[0110] The computer device 6 includes a memory 6a, a processor 6b, and a network interface 6c that are interconnected via a system bus. It should be noted that only the computer device 6 with components 6a-6c is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.
[0111] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.
[0112] The memory 6a includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 6a may be an internal storage unit of the computer device 6, such as the hard disk or memory of the computer device 6. In other embodiments, the memory 6a may also be an external storage device of the computer device 6, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 6. Of course, the memory 6a may also include both the internal storage unit and its external storage device of the computer device 6. In this embodiment, the memory 6a is typically used to store the operating system and various application software installed on the computer device 6, such as program code based on early warning methods for aquatic diseases. In addition, the memory 6a can also be used to temporarily store various types of data that have been output or will be output.
[0113] In some embodiments, the processor 6b may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 6b is typically used to control the overall operation of the computer device 6. In this embodiment, the processor 6b is used to run program code stored in the memory 6a or process data, for example, to run the program code for the early warning method for aquatic diseases.
[0114] The network interface 6c may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 6 and other electronic devices.
[0115] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, 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 is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.
[0116] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.
Claims
1. A method for early warning of aquatic diseases, characterized in that, include: Based on historical health data, a normal baseline model of aquaculture subjects is trained to determine the boundaries of normal behavior. Historical health data includes video data and acoustic data; normal behavioral boundaries include: swimming boundaries, spatial distribution boundaries, and individual physiological boundaries. The real-time multidimensional behavioral characteristics of aquaculture objects are input into the normal baseline model to obtain the abnormality score of the aquaculture objects. When the abnormality score of the aquaculture objects exceeds the preset threshold, an abnormal behavior event is generated, which includes the time window in which the abnormal behavior occurs. Environmental data within N hours prior to the time window is obtained from environmental characteristics. A lagged environmental data subset is constructed. Based on feature importance analysis, the environmental parameters that contribute the most to the lagged environmental data subset are identified. The lagged environmental data subset includes: determining the start and end times of abnormal behavior events based on the time window; pre-setting the lag duration corresponding to different environmental parameters based on biological experience; extracting the lag interval based on the start time; querying data in the environmental time series data whose timestamps are within the lag interval; sorting the queried data according to time order; and obtaining the lagged environmental data subset. A causal relationship test is performed on environmental parameters. When the causal relationship test is passed, the environmental parameters are used as stress trigger parameters. This includes: outputting a causal graph and a causal weight matrix containing all environmental and behavioral parameters based on a preset causal model; taking the abnormal indicators in the abnormal behavioral events as target variables, tracing back along all directed edges directly pointing to the abnormal indicator nodes in the causal graph, and outputting a list of direct cause nodes, which contains the environmental parameters on the causal graph that directly drive the changes in the abnormal indicators; extracting the driving weight of each environmental parameter on the abnormal indicator from the list of direct cause nodes from the causal weight matrix, and determining the environmental parameters corresponding to the driving weights that meet the preset threshold as stress trigger parameters. Based on the environmental characteristics of aquaculture species, the trend of stress-inducing parameters within a preset time period is predicted. When the stress-inducing parameters fall below the preset safe range, an early warning signal is triggered.
2. The early warning method for aquatic diseases according to claim 1, characterized in that, Based on historical health data, a normal baseline model of aquaculture subjects is trained to determine the boundaries of normal behavior, including: We acquire multidimensional behavioral characteristic data of aquaculture objects under their health status, input the multidimensional behavioral characteristic data into the isolated forest model, and construct multiple random decision trees in parallel. Each random decision tree randomly selects features and segments from the multidimensional behavioral feature data; When a sample point is placed into the tree from the root node of the random decision tree, it moves along the branches until it is isolated and becomes a leaf node, and the path length of the sample point in this tree is obtained. The sample point is passed through all the random decision trees to obtain the path length on each tree; Calculate the average of all path lengths, and determine a score threshold based on the abnormal score distribution of all healthy samples. This score threshold is the normal behavior boundary of the multidimensional behavioral feature space.
3. The method for early warning of aquatic diseases according to claim 1, characterized in that, Based on feature importance analysis, the environmental parameters that contribute the most to the lagged environmental data subset are obtained, including: An anomalous indicator from an abnormal behavior event is selected as an explanatory variable and used as the first data; the time series of all environmental parameters to be tested are extracted from the lagged environmental data subset and used as the second data; the first and second data are then time-aligned. Calculate the cross-correlation function for each pair of first and second data; For each environmental parameter, obtain the point with the highest absolute value in the cross-correlation function curve; When the highest absolute value occurs in the leading region of an environmental parameter, if the absolute value of the correlation coefficient obtained based on the cross-correlation function exceeds the preset significance threshold within the leading region, then the environmental parameter is judged to have the greatest contribution.
4. The early warning method for aquatic diseases according to claim 1, characterized in that, Based on the environmental characteristics of aquaculture species, the trend of stress-inducing parameters over a preset time period is predicted. When the stress-inducing parameters fall below a preset safe range, an early warning signal is triggered, including: The established stress trigger parameters are used as the target sequence. Obtain the historical sequence of the identified stress trigger parameters over a preset time period, as well as the sequence of other environmental parameters that have a direct influence on the identified stress trigger parameters; The historical sequence and other environmental parameter sequences are divided into several input sequences using a sliding window, and then divided into several target sequences based on the time step of the sliding window. Data pairs are generated by combining several input sequences and several target sequences. Several input sequences are then fed into a pre-defined long short-term memory network to obtain a corresponding number of predicted values. Based on the loss between the predicted values and the target sequences, the parameters of the long short-term memory network are adjusted so that the predicted values continuously approach the target sequences. When the loss between the predicted values and the target sequences meets a preset threshold, the training of the long short-term memory network is completed. Determine the warning threshold of the stress trigger parameter, input the real-time observation data of the current time period at a preset time step into the long short-term memory network, and obtain the predicted trend curve for the future preset time period; When the predicted value at any point on the trend curve falls below the preset range of the warning threshold, an advance warning is immediately triggered.
5. An early warning device for aquatic diseases, used to implement the early warning method for aquatic diseases according to any one of claims 1-4, characterized in that, include: The abnormal score acquisition module is used to train a normal baseline model of aquaculture objects based on historical health data and determine the boundaries of normal behavior. Historical health data includes video data and acoustic data, and the boundaries of normal behavior include: swimming boundaries, spatial distribution boundaries, and individual physiological boundaries. The abnormal behavior event generation module is used to input the real-time multidimensional behavioral characteristics of aquaculture objects into the normal baseline model to obtain the abnormal score of aquaculture objects. When the abnormal score of aquaculture objects exceeds a preset threshold, an abnormal behavior event is generated, which includes the time window in which the abnormal behavior occurs. The environmental parameter acquisition module is used to obtain environmental data from environmental features within the preceding N hours of a time window, construct a lagged environmental data subset, and, based on feature importance analysis, identify the environmental parameters that contribute the most to the lagged environmental data subset. The lagged environmental data subset includes: determining the start and end times of abnormal behavioral events based on a time window; pre-setting the lag duration corresponding to different environmental parameters based on biological experience; extracting lag intervals based on the start time; querying data in the environmental time series data whose timestamps fall within the lag intervals; sorting the queried data according to time order; and obtaining the lagged environmental data subset. The stress trigger parameter determination module is used to perform causal relationship testing on environmental parameters. When the causal relationship test is passed, the environmental parameters are used as stress trigger parameters. This includes: outputting a causal graph and a causal weight matrix containing all environmental and behavioral parameters based on a preset causal model; taking the abnormal indicators in the abnormal behavioral events as target variables, tracing back along all directed edges directly pointing to the abnormal indicator nodes in the causal graph, and outputting a list of direct cause nodes, which contains the environmental parameters on the causal graph that directly drive the changes in the abnormal indicators; extracting the driving weight of each environmental parameter on the abnormal indicator from the causal weight matrix, and determining the environmental parameters corresponding to the driving weights that meet the preset threshold as stress trigger parameters. The early warning signal triggering module is used to predict the trend of stress causative parameters within a preset time period based on the environmental characteristics of aquaculture objects. When the stress causative parameters fall below the preset safe value range, an early warning signal is triggered.
6. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-4.
7. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, this instruction implements the steps of the method as described in any one of claims 1-4.