Intelligent aquaculture automatic control system

Abstract

The invention relates to the technical field of automatic control, in particular to an intelligent aquaculture automatic control system which comprises a feeding feedback control module, a feeding rhythm allocation module, a water body disturbance partition module, a control unit recognition module and a control path construction module. According to the method, the bait residue density is quantified through image recognition, a feeding state dynamic judgment mechanism is constructed, the accuracy and timeliness of feeding response are improved, the feeding speed is calculated through linkage of time difference and density change, a rhythm allocation strategy based on behavior feedback is formulated, the matching degree of the feeding rhythm to the actual feeding ability is enhanced, and the feeding speed is increased. A main flow path continuous block is constructed by combining flow velocity and flow direction, disturbance levels are marked in a grading manner, a key equipment action area is judged according to cross of fish school active density and dissolved oxygen trend, and an equipment linkage path is constructed by connectivity of a physical channel and an inhabitation area. And an intelligent control scheme of cooperative linkage of ingestion behavior driving, disturbance partition recognition and space layout optimization is realized.

Description

Technical Field

[0001] The present invention relates to the field of automatic control technology, and in particular to an intelligent aquaculture automatic control system. Background Art

[0002] The field of automation control technology includes the control operations of various mechanical equipment, industrial processes, production systems or environmental parameters. Its core content is to use the foundations of control theory, electrical engineering, computer science, etc., through the coordinated operation of sensors, actuators, and controllers to achieve automatic management and adjustment of equipment or processes. It mainly covers signal acquisition and control logic implementation, controller design and configuration, human-computer interaction interface construction, and remote or local control solution integration. It is widely used in multiple application scenarios such as industrial manufacturing, transportation, energy management, agriculture and aquaculture, and has high efficiency, high precision and high reliability. Its development trend is extending towards intelligence, systematization and multi-dimensional data-driven direction, emphasizing the integration of software and hardware and decision-making control optimization.

[0003] Among them, the intelligent aquaculture automatic control system refers to an automated control solution applied to the aquaculture production process, covering the real-time monitoring of environmental factors such as dissolved oxygen, temperature, and pH value of the aquaculture water body and their control and adjustment operations. Specifically, it is based on the deployment of water quality monitoring sensors to collect parameters and combines them with control circuits to execute aquaculture operations such as ventilation, oxygenation, and feeding. Usually, an industrial control host performs logical judgment on external sensory data, automatically decides and executes control instructions based on set parameters, and uses wired or wireless communication methods to complete the linkage of various devices. In closed or semi-closed water environments, aquaculture parameters are dynamically controlled to achieve continuous management of key environmental variables in the aquaculture process.

[0004] In existing aquaculture processes, automated processing relies solely on the collection and thresholding of environmental parameters. This lacks state feedback mechanisms for feeding behavior, making it impossible to dynamically adjust feeding operations based on real-time changes in feeding density. This creates a disconnect between static control and the actual needs of fish. Control timing settings fail to incorporate behavioral indicators such as feeding speed, instead scheduling feeding cycles based solely on parameter settings or timing mechanisms. This leads to lags in rhythmic control, resulting in feed accumulation or delayed feeding. Water flow modeling lacks a grid-continuous representation of disturbance paths, making it impossible to use disturbance intensity as a reference for dynamic control. This results in a mismatch between the activation of water flow control devices and the flow field structure, impairing water quality exchange efficiency in the aquaculture area. Control device identification relies solely on fixed layout locations and basic parameters, lacking the ability to couple and discriminate between fish activity distribution and oxygen demand. Consequently, control responses fail to accurately cover areas of low oxygen density and high oxygen density. Device connectivity paths are constructed without integrating the physical pathways between fish habitats and control units. Dynamic analysis of spatial layout rationality is lacking, making it difficult to form an automatic control network that synchronizes with environmental changes, impacting the adaptability and regulatory effectiveness of the overall control system. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent aquaculture automatic control system.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solutions: an intelligent aquaculture automatic control system comprising:

[0007] The feeding feedback control module obtains the bait residue image, calculates the current bait residue density, calculates the difference between the residual density and the preset feeding density threshold, determines the feeding completion status, and outputs the feeding feedback control signal for the breeding area;

[0008] The feeding rhythm adjustment module determines the feeding completion state of the breeding area based on the feeding feedback control signal of the breeding area, calculates the feeding interval value, calculates the feeding speed based on the change in feeding density before and after, determines the trigger time offset of the next round of directional feeding, and generates the feeding delay control instruction;

[0009] The water disturbance partitioning module collects the current water flow velocity and water flow direction based on the feeding delay control instruction, combines the standard coordinate grid in the aquaculture pond, identifies the disturbance area and binds the disturbance mark, calculates the disturbance level, marks the control linkage priority block, and generates the disturbance priority block label set;

[0010] The control unit identification module locates the water circulation equipment and the oxygenation device according to the disturbance priority block label set, combines the fish activity density image and the change trend of the dissolved oxygen collection point, determines whether the active density change area overlaps with the low oxygen area, determines the core control node of the aquaculture area, and generates an aquaculture control node set.

[0011] As a further solution of the present invention, the feeding feedback control signal of the breeding area includes a feeding status identifier, a bait residual density index, and a feeding threshold deviation; the bait feeding delay control instruction includes a feeding interval value, a feeding speed parameter, and a trigger time offset; the disturbance priority block label set includes a mainstream path continuity level, a disturbance level label, and a linkage priority block coordinate; the aquaculture control node set includes water flow regulation equipment positioning information, fish activity density analysis results, and a low oxygen area identification identifier.

[0012] As a further solution of the present invention, the feeding feedback control module includes:

[0013] The image acquisition submodule acquires the set water surface partition image through the camera, detects the pixel distribution of the bait residual area in the current image, marks and encodes it by area, extracts the bait coverage area data in each partition image, and obtains the partition bait image area value;

[0014] The residual density calculation submodule uses the formula:

[0015]

[0016] Calculate the bait residual density M under the jth water surface partition rj , comprehensively generate the residual density value of bait in the partition, where represents, A rj Represents the bait image area value in the jth water surface partition, A tj represents the total water surface area value of the jth water surface partition, and n represents the total number of partitions;

[0017] The feeding status determination submodule calculates the difference between the residual bait density value of the partition and the currently set feeding density threshold, and makes a judgment based on the difference. If the difference is positive, it is determined that the current feeding stage is in a state of excessive residual bait; if it is negative, it is determined to be a state of complete feeding, and a feeding feedback control signal is obtained.

[0018] As a further solution of the present invention, the feeding rhythm adjustment module includes:

[0019] The interval calculation submodule selects the area that has completed feeding in the current stage based on the feeding feedback control signal of the breeding area, extracts the feeding end time of the previous stage and the feeding start time of the current stage of the corresponding area, calculates the difference between the two time values, and obtains the feeding interval value;

[0020] The speed measurement submodule uses the formula according to the feeding interval value and the bait residual density value of the current and previous stages, combined with the regional feeding intensity data corresponding to each stage:

[0021]

[0022] Calculate the feeding density change rate V s , generate feeding speed data, where M c Indicates the residual density of bait at the current stage, M p Indicates the residual density of bait in the previous stage, I c Indicates the feeding intensity value at the current stage, T c Indicates the current feeding start time, T p Indicates the end time of the previous stage, T k represents the feeding interval value of the kth region, M k represents the current bait residual density value in the kth area, and K represents the total number of areas in the feeding completion state;

[0023] The delay control submodule calculates the time offset for triggering the next round of directional feeding of bait according to the feeding speed data, superimposes the current time point, sets it as the feeding execution time, records the waiting period before the next round of operation, and establishes the bait feeding delay control instruction.

[0024] As a further solution of the present invention, the water disturbance zoning module includes:

[0025] The flow field identification submodule collects the current water velocity value and water flow direction transmitted by the flow velocity probe based on the waiting period set by the feeding delay control instruction, and combines the standard coordinate grid structure in the aquaculture pond to identify the coordinate blocks with continuous mainstream paths and aggregate them to obtain the mainstream path block set;

[0026] The disturbance calculation submodule obtains the flow velocity value, flow direction angle and number of path connections in each block according to the set of mainstream path blocks, and performs joint calculations through flow velocity difference, direction offset and connectivity scale, using the formula:

[0027]

[0028] The disturbance level value L is calculated, and the disturbance intensity of continuous blocks is sorted by the disturbance level value and a label is generated, where represents the disturbance level value, v represents the mean flow velocity of the current mainstream path, u represents the set reference flow velocity value, and α g represents the flow angle of the gth coordinate block, μ represents the average flow angle of all coordinate blocks, G represents the number of coordinate blocks in the mainstream path, Q g Indicates the number of connected paths in the g-th block;

[0029] The priority screening submodule determines whether each coordinate block meets the disturbance intensity condition according to the disturbance level value, sets a control flag for the block that meets the condition, binds it as a control linkage priority block, and establishes a disturbance priority block label set.

[0030] As a further solution of the present invention, the control unit identification module includes:

[0031] The equipment positioning submodule searches for the water circulation equipment and aeration devices deployed at each coordinate range, confirms the spatial overlap between the equipment coordinate position and the block range, and obtains the regional equipment index set.

[0032] The activity matching submodule collects fish activity density images within the corresponding coordinate area based on the regional device index set, monitors the oxygen content change trend of the current dissolved oxygen collection point, determines whether the activity density distribution of the coordinates of each device is synchronized with the oxygen content change trend, and obtains the hypoxia activity overlap value;

[0033] The node extraction submodule selects the coordinate units where water flow regulation equipment has been deployed in the blocks where the active density change coincides with the hypoxic area based on the hypoxia-active overlap value, records the corresponding equipment number as the core control equipment of the corresponding aquaculture area, and establishes an aquaculture control node set.

[0034] As a further aspect of the present invention, the system further includes a control path construction module;

[0035] The control path construction module sequentially identifies the physical connection channels between the auxiliary oxygenation unit and the circulation device based on the spatial layout information of each core control device in the aquaculture control node set. By combining the connectivity conditions between the fish habitat and activity area and the device location, it constructs an integrated automatic control network of devices in series that forms a continuous water body regulation link, and generates an automatic control plan for the aquaculture area.

[0036] The automatic control scheme for the breeding area includes the physical connection channel information of the equipment, the connection path of the habitat activity, and the automatic control network topology.

[0037] As a further solution of the present invention, the control path construction module includes:

[0038] The structure recognition submodule identifies adjacent auxiliary oxygenation units and water circulation devices within the control area based on the spatial layout information of each core control device in the aquaculture control node set, extracts physical location coordinates and performs spatial connection judgment to obtain a set of device connection channels;

[0039] The path connection submodule determines whether the devices meet the continuous action range, adjacent coordinate mapping relationship and linkage response logic based on the device connection channel set and the overlap relationship between the fish habitat and activity area in the aquaculture area and the device distribution. It integrates all the connected units that meet the conditions to obtain the continuous adjustment path network value.

[0040] The solution output submodule classifies and summarizes the control device numbers covered by each path according to the continuous adjustment path network value, and associates the automatic response conditions and trigger cycle requirements to establish an automatic control solution for the breeding area.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are:

[0042] In the present invention, the distribution of bait residues is obtained through image recognition, and the quantitative expression of feeding density is realized by combining with area ratio calculation. A dynamic feeding state recognition mechanism is established to avoid feeding deviation caused by misjudgment of static parameters and improve the accuracy of feeding judgment. An interval calculation mode is constructed based on the time difference between the end of feeding and the new round of feeding. The feeding speed parameter is introduced in conjunction with the density change to form a rhythm adjustment logic based on behavioral feedback, thereby enhancing the adaptability of feeding rhythm to the feeding ability of fish. The water flow velocity and flow direction are integrated within the set waiting period, and the flow field path is subjected to coordinate grid analysis to realize the continuity recognition of the mainstream path and the division of disturbance levels, enriching the water resources. Based on the dimension of body dynamic information expression, the system is based on the overlapping analysis of disturbance areas and dissolved oxygen trends, combined with the active density image of fish schools, and establishes a judgment standard for identifying the matching relationship between environmental anomalies and key equipment responses, supporting the differentiated extraction of control nodes. According to the spatial connectivity relationship between the control equipment and the fish habitat, a device series network is constructed based on physical connection channels to strengthen the continuity of the adjustment link and ensure that the equipment linkage response covers the key areas of the aquaculture environment. The overall structure drives the control cycle setting through state feedback, guides equipment identification through spatial disturbance data, and optimizes path connectivity through habitat distribution coupling, thereby realizing highly adaptable automated aquaculture control. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a system flow chart of the present invention;

[0044] Figure 2 This is a flow chart of the feeding feedback control module of the present invention;

[0045] Figure 3 This is a flow chart of the feeding rhythm adjustment module of the present invention;

[0046] Figure 4 This is a flow chart of the water disturbance zoning module of the present invention;

[0047] Figure 5 This is a flow chart of the control unit identification module of the present invention;

[0048] Figure 6 This is a flow chart of the control path construction module of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0050] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0051] See also Figure 1 , an intelligent aquaculture automatic control system includes:

[0052] The feeding feedback control module obtains the bait residue image of the water surface partition set by the camera in the aquaculture area, calculates the current bait residue density according to the image area ratio, and calculates the difference between the residue density and the preset feeding density threshold. If the difference is positive, the current feeding stage is determined to be in a state of excessive bait residue. If the difference is negative, the feeding stage is determined to be in a state of complete feeding, and outputs a feeding feedback control signal for the aquaculture area;

[0053] The feeding rhythm adjustment module determines the feeding completion state based on the feeding feedback control signal of the breeding area, calculates the time difference between the actual feeding end time of the previous stage and the feeding start time of the current stage as the feeding interval value, calculates the feeding speed based on the change in feeding density before and after, determines the time offset for triggering the next round of targeted feeding, and sets it as the waiting period before the next round of operation, generating the feeding delay control instruction;

[0054] The water disturbance partitioning module combines the current water velocity value and water flow direction sent back by the flow probe with the standard coordinate grid in the aquaculture pond within the waiting period set by the feeding delay control instruction. It merges the coordinate blocks with continuous mainstream paths into disturbance areas, binds the control units in the disturbance area as disturbance identifiers, calculates the disturbance level according to the continuity of the mainstream path, selects the blocks that meet the disturbance intensity conditions, marks them as control linkage priority blocks, and generates a disturbance priority block label set.

[0055] The control unit identification module locates the coordinate range of the control linkage priority block based on the disturbance priority block label, locates the water circulation equipment and oxygenation device, obtains the active density image of the fish school in the corresponding area, and combines the current trend of dissolved oxygen sampling points to determine whether the active density change area overlaps with the low-oxygen area. If there is a deployed water flow regulation device in the overlapping area, the corresponding device is determined to be the core control node of the current aquaculture area, and the aquaculture control node set is generated;

[0056] The control path construction module identifies the physical connection channels between the auxiliary oxygenation unit and the circulation device in sequence according to the spatial layout information of each core control device in the aquaculture control node set. By combining and constructing the connectivity conditions between the fish habitat and activity area and the equipment location, it constructs an integrated automatic control network of equipment series connection for a continuous water body regulation link, and generates an automatic control plan for the aquaculture area.

[0057] The feeding feedback control signals in the breeding area include feeding status identification, bait residual density index, and feeding threshold deviation. The bait feeding delay control instructions include feeding interval value, feeding speed parameter, and trigger time offset. The disturbance priority block label set includes mainstream path continuity level, disturbance level label, and linkage priority block coordinates. The aquaculture control node set includes water flow regulation equipment positioning information, fish activity density analysis results, and low-oxygen area identification. The breeding area automatic control plan includes equipment physical connection channel information, habitat activity connection path, and automatic control network topology.

[0058] See also Figure 2 , the feeding feedback control module includes:

[0059] The image acquisition submodule acquires the set water surface partition image through the camera, detects the pixel distribution of the bait residual area in the current image, marks and encodes it by area, extracts the bait coverage area data in each partition image, and obtains the partition bait image area value;

[0060] The camera in the feeding feedback control module acquires the set water surface partition image. The camera is arranged above the breeding area. Each camera covers a 3m × 3m breeding water surface area, which is divided into 9 equal-area water surface partitions according to the set coordinates. When acquiring images, a shooting frequency of 2 seconds per frame is used for continuous acquisition. The resolution of each frame image is set to 1920 × 1080 pixels. After image acquisition, each partition image is first extracted according to the preset image coordinate area. The process of executing image extraction is to call the image boundary coordinates xmin, xmax, ymin, ymax, and perform a nine-equal-area partition operation on the original image through the cropping function. The extracted image is binarized according to the pixel grayscale threshold. During the binarization operation, the pixel threshold is set to 95. The areas with pixel values ​​higher than the threshold are marked as bait, and those below the threshold are marked as background. Then the number of pixels in the bait marked area in each partition is counted, and the image area value of the bait in each partition is recorded by pixel value statistics. For example, the effective pixel area of ​​the first partition is 320×360. After binarization, the number of bait pixels is 14520, and the corresponding area ratio is 14520 / 115200, which is about 0.126, indicating that the bait image coverage rate of this partition is 12.6%. This type of operation is performed on the remaining partitions in sequence, and the bait image pixel areas of all nine partitions are recorded to finally obtain the partition bait image area value.

[0061] The residual density calculation submodule uses the formula:

[0062]

[0063] Calculate the bait residual density M under the jth water surface partition rj , comprehensively generate the residual density value of bait in the partition, where represents, A rj Represents the bait image area value in the jth water surface partition, A tj represents the total water surface area value of the jth water surface partition, and n represents the total number of partitions;

[0064] According to the area value of the partitioned bait image and the total water surface area value of the partition corresponding to the current image, the total pixel area of ​​each partition is fixed to 320×360 pixels, and the corresponding pixel area is 115200. The formula is used to calculate the current bait residual density value of each partition.

[0065] n=9, now take the first partition as an example, A r1 =14520,A t1 =115200, the residual density ratios of all partitions are as follows:

[0066] Partition number <![CDATA[Bait pixel area (A rj )]]> <![CDATA[Total pixel area of the partition (A tj )]]> <![CDATA[Ratio (A rj / A tj )]]> 1 14520 115200 0.126 2 17340 115200 0.1506 3 11800 115200 0.1025 4 16000 115200 0.139 5 13400 115200 0.1163 6 9800 115200 0.085 7 15200 115200 0.132 8 12600 115200 0.1094 9 14100 115200 0.1224

[0067] Table 1 Bait residue ratios for different zones

[0068] Refer to Table 1 and substitute the residual density of the first partition into the formula for calculation:

[0069] Item 1:

[0070]

[0071] Item 2:

[0072]

[0073] Item 3:

[0074]

[0075] but:

[0076] M1=0.126+120.49-0.1201=120.4959;

[0077] Thus, the bait residual density value of the first partition is 120.4959. The remaining partitions are calculated in the same way to obtain the partition bait residual density values ​​of all partitions.

[0078] The operational logic of this formula is reflected in the synergistic combination of multi-dimensional parameters to enhance the comprehensive judgment ability of the bait residual status. It represents the direct ratio of the bait image area in the current partition to the total water surface pixel area, quantifies the coverage density of bait in the partition, and is the most basic indicator for measuring the degree of residue; the second item The square root operation of the bait image area value is introduced to perform nonlinear amplification on the absolute area, so that when the bait area is large, its influence on the overall density value is increased, thereby enhancing the recognition ability of large-area residues; the third item A baseline term representing the average bait density across the entire region serves as a reference for the global residue background. Subtracting this term reveals the deviation between the current partition and the average state, reflecting the relative specificity of the bait residue in that region. The combination of these three terms forms a "local-nonlinear-global" joint expression model, enabling the system to not only identify the basic distribution of bait within a partition but also enhance the prominence of large residue areas. Furthermore, dynamic relative comparisons using the global average residue density as a reference allow for a more comprehensive and comprehensive residue density judgment structure.

[0079] The feeding status determination submodule calculates the difference between the residual bait density value of the partition and the currently set feeding density threshold, and makes a judgment based on the difference. If the difference is positive, it is determined that the current feeding stage is in a state of excessive residual bait; if it is negative, it is determined to be a state of complete feeding, and a feeding feedback control signal is obtained;

[0080] The difference between the residual density of bait in each partition and the currently set feeding density threshold is calculated. The feeding density threshold is set based on the statistical value of the average residual density in the image area obtained by multiple batches of cameras during the feeding cycle, and is combined with the total feed feeding amount, the volume of the aquaculture water body and the daily feeding time period to aggregate the average residual density per unit area per unit time, and then formulate a control standard. The value is set to 95, which fluctuates with the feed particle size, water flow disturbance frequency and water temperature. The smaller the particle size, the higher the water flow frequency or the lower the temperature, the lower the average feeding speed, and the threshold needs to be lowered. Conversely, it can be appropriately increased. The feeding upper limit value set according to the current feeding plan is calculated through the feeding history average, and the residual density of bait in each partition obtained in the second section is called. The difference between the degree value and the density threshold is judged. Taking the first partition as an example, its residual density value is 120.4959, which is greater than the set threshold 95, and the difference is 25.4959. Therefore, the current partition is judged to be in an excessive state of bait residue. The second partition judgment is continued. Its residual density value is 135.128, which is greater than the threshold and is judged to be in an excessive state. The residual density value of the sixth partition is 88.54, which is less than the threshold and the difference is -6.46, which is judged to be in a feeding completion state. All partitions are judged in turn, and recorded as "excessive" or "completed" according to the positive or negative difference. Finally, the feeding feedback control signal of each partition is generated and recorded in the control signal output table as a reference for system automatic control or subsequent feeding parameter adjustment to obtain the feeding feedback control signal.

[0081] See also Figure 3 , the feeding rhythm adjustment module includes:

[0082] The interval calculation submodule, based on the feeding feedback control signal of the breeding area, screens the area that has completed feeding in the current stage, extracts the feeding end time of the previous stage and the feeding start time of the current stage in the corresponding area, calculates the difference between the two time values, and obtains the feeding interval value;

[0083] Based on the feeding feedback control signal of the aquaculture area, the water area partition in the feeding completion state is first identified. The state is marked and identified by the status code in the feedback signal. For example, when the status code is "1", it means that feeding is completed. By reading the feedback signal field "feeding_status", the records with the status value of 1 are filtered, and the feeding end time of the previous stage of the corresponding area is extracted. The time field is recorded in the form of a timestamp in the format of "2024-04-1408:25:00". At the same time, the feeding start time of the current stage is extracted from the current feeding plan of the system. The format is the same as above. Then the two time fields are converted to the standard Unix time format in seconds to facilitate numerical calculations. For example, the feeding end time is 08:25:00 (corresponding to Unix time in 16 81466700 seconds), the feeding start time is 08:55:00 (corresponding to Unix time is 1681468500 seconds), the difference between the two is the feeding interval value, and the calculated interval is 1800 seconds, that is, 30 minutes. In order to ensure the uniformity of the data structure, the above processing is performed on all completed feeding partitions, and the results are stored in "interval_list". The list structure is in array form, where each element contains fields such as "z_id" and "interval_seconds". For example, the feeding interval corresponding to z1 is 1800 seconds, and that of z2 is 2700 seconds. For the setting of the interval status classification standard, refer to the bait residue concentration change trend curve recorded in the system and reduce the bait residue to the critical density (such as 2.5g / m 2 ) was used as the basic threshold, and the benchmark value was set to 1800 seconds. Combined with the offset characteristics of the response time of different fish schools after feeding, an upward offset of 600 seconds was set as the long-cycle limit value, i.e. 2400 seconds, and a downward offset of 600 seconds was set as the short-cycle limit value, i.e. 1200 seconds. Three types of cycle intervals were delineated in this way; among them, an interval value less than 1200 seconds indicated a rapid feeding response and a rapid decrease in residual concentration, which was classified as a short-cycle feeding state; an interval value between 1200 and 2400 seconds indicated that the feeding rhythm was consistent with the residual attenuation trend, which was a medium-cycle feeding state; an interval value exceeding 2400 seconds indicated a delayed feeding response, and the residual concentration hovered around the threshold for a longer time, which was a long-cycle feeding state. Moreover, this value fluctuated with the initial peak value of the residual concentration and the range of the fish school active window, showing an aggravated trend of delayed feeding behavior. In the implementation process, in order to clarify the interval time setting and calculation processing method of each partition, the following data table is listed to assist in the explanation:

[0084] Table 2 Example of feeding interval calculation

[0085] Partition number Last end time Current start time Time interval (seconds) zone_01 2025-04-1408:25:00 2025-04-1408:55:00 1800 zone_02 2025-04-1408:10:00 2025-04-1408:55:00 2700

[0086] As shown in Table 2, by converting the time field into a standard value and performing difference calculation, the feeding interval value of each area can be accurately obtained, which can eventually be used for subsequent cycle adjustment and feeding strategy correction.

[0087] The speed measurement submodule uses the formula based on the feeding interval value and the bait residual density value of the current and previous stages, combined with the regional feeding intensity data corresponding to each stage:

[0088]

[0089] Calculate the feeding density change rate V s , generate feeding speed data, where M c Indicates the residual density of bait at the current stage, M p Indicates the residual density of bait in the previous stage, I c Indicates the feeding intensity value at the current stage, T c Indicates the current feeding start time, T p Indicates the end time of the previous stage, T k represents the feeding interval value of the kth region, M k represents the current bait residual density value in the kth area, and K represents the total number of areas in the feeding completion state;

[0090] According to the feeding interval value, combined with the bait residual density value of the corresponding partition in the current and previous stages, the feeding speed is calculated. In actual operation, first read the residual density value field "current_density" of the current stage, and "previous_density" of the previous stage, both in g / m 2 Measurement, for example, zone_01 is currently 18.2g / m 2 , the previous stage is 25.0g / m 2 At the same time, extract the regional feeding intensity "intensity" field, its unit is kg / min, such as 0.6kg / min at present, call multiple partition data, and enter the value: current residual density M c =18.2, the residual density M in the previous stage p =25.0, set 3 partitions to participate in the calculation, namely:

[0091] z1: M1 = 18.2, T1 = 1800 seconds

[0092] z2: M2 = 20.5, T2 = 2700 seconds;

[0093] z3: M3=15.8, T3=1500 seconds;

[0094] Current feeding intensity I c =0.6, the time difference between the current and the last feeding is T c -Tp = 1800 seconds, substitute the calculation into the following:

[0095] Calculate the absolute value of the density difference:

[0096] |M c -M p |=|18.2-25.0|=6.8;

[0097] Compute the summation:

[0098]

[0099] Denominator operation:

[0100]

[0101] Calculate the integer result:

[0102]

[0103] This result indicates that the feeding speed of zone_01 at the current stage is 0.00379 g / m 2 ·s -1 , which is used to characterize the rate of change of feeding density. The benefit of the formula is that it introduces the ratio of multiple partitioned bait density values ​​to the corresponding interval time, and combines the feeding intensity with the time difference between stages to achieve comprehensive quantification of feeding speed, enhance the timeliness of speed evaluation and the linkage between partitions, and finally obtain feeding speed data.

[0104] The calculation logic of this formula is based on the feeding speed value V s The multi-factor comprehensive construction idea reflects the composite effect relationship of factors such as density difference, multi-area density time ratio, feeding intensity and feeding time interval. In the numerator of the formula, the term |M c -M p | Represents the absolute difference between the current and previous stage bait residual density, used to quantify the overall density change trend, and the summation term added later The proportional relationship between the residual density of multiple regions and their corresponding feeding intervals is further introduced to reflect the comprehensive intensity of the feeding process in different partitions. Since the feeding speed is affected by the feeding balance of multiple regions, it is comprehensively considered in a linear superposition manner; in the denominator, the feeding intensity I c It reflects the intensity of feeding amount per unit time. The purpose is to compress and adjust the excessive strength value to reduce its nonlinear impact on the speed value; and |T c -T p| represents the time difference between stages, which is used to measure the span of the feeding cycle and normalize the feeding speed value. Therefore, the overall formula constructs a composite calculation structure for rationally regulating the feeding speed through the weighted integration of the residual density change trend, the local area ratio enhancement term, the intensity correction term, and the time correction term.

[0105] The delay control submodule calculates the time offset for triggering the next round of directional feeding based on the feeding speed data, adds the current time point, sets it as the feeding execution time, records the waiting period before the next round of operation, and establishes the feeding delay control instruction;

[0106] According to the feeding speed data, the feeding time offset can be further determined. First, the current system time is extracted as the reference trigger time point, for example, the current time is 2024-04-14 09:00:00, and the speed value partition boundary is set to determine the delay adjustment threshold. The partition boundary is set according to the relationship between the unit feed feeding rate and time response of the fish school in the aquaculture water body recorded in the system. By collecting the change curves between the feeding speed and the residual feed density of multiple water surface partitions in different time periods, the speed range corresponding to the stable rate of decrease of the unit feed density is calculated to form the partition judgment threshold; wherein, when the feeding speed value V s Between 0.002 and 0.004 g / m 2 ·s -1 The range indicates that the fish feeding rate and feeding rhythm are close to balance during this period, and the residual density decline slope fluctuation remains within ±0.03g / m 2 min -1 The threshold is used as the “stable feeding zone” judgment standard, and the corresponding offset value is set to +600 seconds; when V s When it is less than 0.002, it means that the feeding is relatively delayed, and the density drop curve has a delayed slope. Set the offset value to +1200 seconds; if V s If it is greater than 0.004, it is considered as rapid ingestion, and the slope fluctuation of the curve exceeds +0.06g / m 2 min -1 , there is a tendency to consume bait in advance, the corresponding offset value is set to +300 seconds. This value fluctuates linearly with the slope of the residual density change and the feeding response interval, which belongs to the dynamically adjustable interval setting. Combined with the V calculated above, s=0.00379, which falls into the middle interval. The offset is set to 600 seconds. After adding the current system time, the execution time of the next round of targeted feeding is 09:10:00, and it is set as the waiting time before the next cycle is triggered, that is, the waiting period is 600 seconds. A structure is established to record the instruction content. The fields include "zone_id", "delay_time", "next_feed_time" and other information. For example, the corresponding record of z1 is as follows: delay_time=600, next_feed_time=09:10:00, and finally the bait feeding delay control instruction is generated.

[0107] See also Figure 4 , the water disturbance zoning module includes:

[0108] The flow field identification submodule collects the current water velocity and direction transmitted by the flow probe based on the waiting period set by the feeding delay control instruction. Combined with the standard coordinate grid structure in the aquaculture pond, it identifies and aggregates the coordinate blocks of the continuous mainstream path to obtain the mainstream path block set.

[0109] Based on the waiting period set by the feeding delay control instruction, the water velocity value and the water flow angle information at the corresponding time point sent back by the flow probe during this period are collected. The data collection frequency, position layout and sensor parameters need to be reasonably set. For example, in the 5×5 standard coordinate grid of the aquaculture pond, a flow probe is arranged in each grid unit, and the flow velocity and direction values ​​are recorded in units of 2 seconds. The total recording time is 120 seconds, and each unit can collect 60 sets of time series data. For these data, the system continuously detects from coordinates (2, 3) to (2, 7) that the flow velocity is maintained at 0.23~0.25m / s and the flow angle is concentrated at 85°~90°. Continuous paths, combined with the standard coordinate grid, these grids with continuous mainstream paths are classified into the same block and merged into a mainstream path block set. If there are discontinuous intervals in some paths, the maximum allowable interval at the breakpoint needs to be judged. For example, a maximum of one interval unit can be considered as a connected state, otherwise the path is eliminated. The setting conditions can be explained by actual data. For example, if there is no flow velocity record at coordinates (3, 5), if there is only one breakpoint, it is connected to the data between (3, 4) and (3, 6), otherwise the path is judged to be broken, thereby merging multiple mainstream path blocks, completing the extraction of continuous disturbance areas and outputting the mainstream path block set.

[0110] The disturbance calculation submodule obtains the flow velocity value, flow direction angle and number of path connections in each block based on the set of mainstream path blocks, and performs joint calculations through flow velocity difference, direction offset and connectivity scale, using the formula:

[0111]

[0112] The disturbance level value L is calculated, and the disturbance intensity of continuous blocks is sorted by the disturbance level value and a label is generated, where represents the disturbance level value, v represents the mean flow velocity of the current mainstream path, u represents the set reference flow velocity value, and α g represents the flow angle of the gth coordinate block, μ represents the average flow angle of all coordinate blocks, G represents the number of coordinate blocks in the mainstream path, Q g Indicates the number of connected paths in the g-th block;

[0113] According to the above merged mainstream path block set, the coordinate units in each block are statistically analyzed for factors such as velocity mean, directional angle deviation, and path connectivity. The velocity value and flow angle of each coordinate point are extracted in turn. The data in the main path block are calculated. First, the velocity value sequence in each block is extracted and the average velocity v is calculated. The reference velocity u (such as the design flow velocity of the aquaculture pond is 0.20m / s) is used as the benchmark to calculate the average deviation |vu|. The directional angle deviation is the difference between the angle of each coordinate point and the overall average angle |α g -μ| is used as a reference, and the number of continuous connected path segments Q is extracted from the connectivity term. g , and calculate the disturbance level value L in combination with the number of coordinates G.

[0114] Take a main path block as an example: it contains 6 coordinate units, that is, G=6, the flow velocity values ​​at each point are [0.22, 0.24, 0.23, 0.25, 0.24, 0.23] m / s, the reference flow velocity u=0.20 m / s, and the direction angles are [88, 89, 87, 90, 88, 89] ° , the average angle μ=88.5°, and the connectivity of each point is [4,4,3,4,4,3], then:

[0115]

[0116] |vu|=|0.235-0.20|=0.035;

[0117]

[0118] Substitute into the formula:

[0119]

[0120] The resulting disturbance level value for this path block is 1.225.

[0121] The benefit of this formula is that by introducing the directional offset term and the path connectivity term to integrate the disturbance trend analysis, the disturbance level value can reflect the stability and concentration of the mainstream path. Among them, the offset of the flow velocity from the reference value measures the disturbance intensity, the angular offset expresses the directional consistency, and the inverse of the connectivity reflects the structural stability, thereby enhancing the rationality of the disturbance area assessment in the system.

[0122] Table 3 Main path block parameters

[0123] Coordinate number Flow rate (m / s) Direction angle (°) Number of connections 1 0.22 88 4 2 0.24 89 4 3 0.23 87 3 4 0.25 90 4 5 0.24 88 4 6 0.23 89 3

[0124] As shown in Table 3, the flow velocity and direction at the sampling points exhibited a stable trend within the specified time period, with small angular deviations and a number of connections concentrated between 3 and 4, making them suitable for disturbance level determination. The disturbance level value of 1.225 falls within the system's preset disturbance determination range of [1.0, 1.5], indicating that this block has a medium to high disturbance intensity and is eligible for the subsequent screening stage.

[0125] The operational logic design of the above formula is to quantify the comprehensive stability of the disturbance region through multi-dimensional characteristics, among which the two indicators |vu| and ∑|α in the numerator are g -μ| is used to reflect the overall velocity deviation and flow direction consistency of the main path block. The former is used as a macro flow intensity indicator to measure the deviation between the current state and the set reference velocity, while the latter reflects the flow direction change within the path by accumulating the absolute deviation between the direction angle and the mean. The two are directly added to represent the linear cumulative contribution of the disturbance intensity and directional dispersion. The denominator part is The scale factor representing the number of coordinate units in the main path uses square root operation to limit the nonlinear amplification effect of scale on the disturbance level value, avoiding the exponential growth deviation caused by the simple accumulation of large blocks; It is used to introduce an inhibition factor for connectivity, where the greater the number of connections in each path unit, the smaller its reciprocal value, thereby reducing the denominator value overall and amplifying the weight of the numerator in the disturbance level, reflecting a significant improvement in the disturbance level when the connectivity is higher and the structure is more stable; the overall structure is a weighted ratio formula, in which the participating items construct a comprehensive quantitative expression framework of the disturbance state through arithmetic relationships such as absolute value, summation and square root.

[0126] The priority screening submodule determines whether each coordinate block meets the disturbance intensity condition based on the disturbance level value, sets a control flag for the block that meets the condition, binds it as a control linkage priority block, and establishes a disturbance priority block label set;

[0127] According to the obtained disturbance level value, the disturbance intensity condition threshold is set in the system. Assuming that the disturbance level threshold is set to 1.1, the setting basis of this threshold is the distribution range of each disturbance level value in the main path block in the system during the continuous multi-cycle stable acquisition process. Specifically, it is based on the disturbance level value distribution sequence calculated in the six daily water disturbance status monitoring cycles in the aquaculture pond. After excluding the lowest 10% and highest 10% fluctuation abnormal samples, the disturbance level concentration interval of the central dense segment is extracted. The median value of 1.1 is extracted from the interval [0.9, 1.3] as the threshold. This value is consistent with the main path connectivity number. The disturbance level is closely related to the quantity, especially when the number of connections is greater than 5, where the disturbance level typically exceeds 1.1. Therefore, this threshold fluctuates as the number of effectively connected units within the block increases. Its setting goal is to capture high-response units corresponding to high-intensity continuous disturbance paths. The disturbance level values ​​corresponding to all mainstream path blocks are called and compared with the threshold. Blocks with a value greater than 1.1 are screened out, and the control units of these blocks are set as priority control objects. At the same time, their corresponding coordinate numbers are numbered and marked. The system extracts this batch of control units using coordinate indexing and simultaneously binds the disturbance identifier to form a tag set, namely the disturbance priority block label set. The block disturbance level value calculated by the previous formula is 1.225, which is greater than the system-set threshold of 1.1. Therefore, this block is successfully selected as a control linkage priority block and included in the control cycle disturbance label set.

[0128] See also Figure 5 , the control unit identification module includes:

[0129] The equipment positioning submodule searches for the water circulation equipment and aeration devices deployed at each coordinate range, identifies the coordinate range of the control linkage priority block based on the disturbance priority block label, and confirms the spatial overlap between the equipment coordinate position and the block range to obtain the regional equipment index set.

[0130] The tag range in the disturbance priority block tag set corresponds to the coordinate grid number in the aquaculture area. In actual operation, the central control system usually derives a block index in the form of "[A1, B3, C5]". The corresponding two-dimensional coordinates can be located to the physical coordinate area of ​​the aquaculture pond through the coordinate conversion table. The control system obtains the layout coordinate information of the water circulation equipment and the oxygenation device by reading the equipment layout layer. In actual application, the equipment types may include aeration oxygenation pumps, underwater thrusters, fixed circulation pumps, etc., which are registered in the control platform in the form of equipment ID and coordinates "(x, y)". The system first performs coordinate matching processing, that is, for each coordinate area in the disturbance block, one by one A control device is deployed at each location. If the coordinates of the device center point fall within the boundary of the disturbance coordinate grid, it is considered a match. The grid boundary overlap deviation problem needs to be considered during the matching process. Therefore, a boundary buffer setting value of ±0.5 meters is allowed to avoid errors caused by device deployment offset. For example, in a control grid number A1, an aeration device with coordinates (12.1, 25.4) is deployed, and the grid boundary is (11.5, 13.0, 24.5, 26.0). Since the device position is within the boundary, it is confirmed as a successfully matched device. Finally, the system counts the numbers of all successfully matched devices and records them as the preliminary target devices of the control unit. The result is the regional device index set.

[0131] The activity matching submodule collects fish activity density images within the corresponding coordinate area based on the regional device index set, monitors the oxygen content change trend at the current dissolved oxygen collection point, and determines whether the activity density distribution at the coordinates of each device is synchronized with the oxygen content change trend to obtain the hypoxia activity overlap value;

[0132] Based on the obtained regional device index set, the system calls the real-time image of the underwater camera from each matching coordinate area, performs grayscale processing and regional segmentation on the image, and extracts the group movement density area in the image. If the grayscale intensity change exceeds the set threshold of 15, it is considered that there is an active fish school. When the image frame rate is 25fps, the continuous frame detection results are averaged to obtain the activity index value. At the same time, dissolved oxygen point data is collected in the corresponding area to obtain the oxygen content change sequence. For example, the oxygen content in a certain block is recorded as [6.8, 6.5, 6.2, 5.9, 5.7] mg / L for 5 consecutive minutes, indicating that the oxygen content in the area is decreasing. The average oxygen content in this data sequence is 6.22 mg / L. The system makes a judgment based on the preset hypoxia threshold of 5.5 mg / L, and compares the image activity index with the oxygen content change trend. If the two decrease synchronously, that is, the grayscale change range gradually shrinks, the oxygen content continuously decreases, and the change trends are highly overlapped (Pearson correlation coefficient is greater than 0.85), then it is considered that there is a hypoxia and high-density overlap phenomenon in the area, and the hypoxia active overlap value of such areas is set to a positive value. For example, in area A1, the activity density decreases at a trend of -0.2 / minute, the dissolved oxygen decreases at a trend of -0.15 mg / L / minute, and the correlation coefficient is 0.88. Therefore, the hypoxia active overlap value of this area is set to 1. The system summarizes all coordinate units that meet the conditions and generates a hypoxia active overlap value.

[0133] The node extraction submodule selects the coordinate units where water flow regulation equipment has been deployed in the blocks where the active density change coincides with the hypoxic area based on the hypoxia activity overlap value. The corresponding equipment number is recorded as the core regulation equipment of the corresponding aquaculture area, and an aquaculture regulation node set is established.

[0134] Based on the obtained hypoxia active overlap value, the system traverses the corresponding coordinate block, calls the deployment information of all its subordinate devices, and compares the coordinates of the overlapping area with the device deployment points in the device index set. If there is both a positive overlap value judgment result and a deployed regulation device (such as an underwater propeller or a water circulation device) in a certain coordinate area, the device is directly marked as a core control device. In the example, if the coordinate block A1 corresponds to the device number S001, and the device type is a circulation oxygen pump, the system will enter the device S001 into the control node list. If the subsequent device number S002 matches the coordinate B3 and the overlap value of the B3 block is 0, the system will skip the device, and finally obtain the results listed in Table 1:

[0135] Table 4 Example table of control node identification

[0136] Block number Equipment Number Device Type Active overlap value Whether to mark it as a control node A1 S001 Circulation oxygen pump 1 yes B3 S002 Aeration pipe installation 0 no C5 S003 underwater thrusters 2 yes

[0137] As shown in Table 4, only in areas where the active overlap value is positive and there are deployed devices, the system identifies the device as a core control unit. The system finally aggregates all identified devices to form an aquaculture control node set.

[0138] See also Figure 6 , the control path building blocks include:

[0139] The structure recognition submodule identifies the adjacent auxiliary oxygenation units and water circulation devices within the control area based on the spatial layout information of each core control device in the aquaculture control node set, extracts the physical location coordinates, performs spatial connection judgment, and obtains the device connection channel set;

[0140] According to the spatial layout information of the core control equipment listed in the aquaculture control node set, the coordinate position of the control node is first spatially mapped to identify its precise position in the three-dimensional coordinate system. For example, the oxygenation devices numbered A01 and B03 are located at 25 meters and 42 meters on the X-axis, 16 meters on the Y-axis, and 3 meters and 4 meters on the Z-axis, respectively. After obtaining the node information, the distance between each group of devices is measured by Euclidean distance by traversing the auxiliary devices configured in the peripheral device pool area. If the connection distance of a pair of devices is less than the set connection threshold of 7 meters, it is judged that its physical position meets the linkage layout requirements and is set to the "connectable" state. For example, the distance between A01 and A02 devices is calculated as meters, which is less than the threshold, so it is connectable; then, all device number pairs that meet the connectable relationship are combined to establish a device connection channel set, and based on this, a control network input structure is constructed to form a device connection channel set.

[0141] Table 5 Equipment connection channel data table

[0142] Equipment Number Device Type Connection distance (m) Connection Status A01 Oxygen enrichment device 5.6 Connectable A02 Water circulation 3.8 Connectable B03 Oxygen enrichment device 6.1 Connectable C04 Water circulation 4.2 Connectable

[0143] As shown in Table 5, the connection relationship between devices has been judged according to the spatial distance to form a preliminary set of device connection channels for subsequent path construction operations.

[0144] The path connection submodule determines whether the devices meet the requirements of continuous action range, adjacent coordinate mapping relationship and linkage response logic based on the set of device connection channels and the overlap between the fish habitat and activity areas in the aquaculture area and the device distribution. It then integrates all the connected units that meet the conditions to obtain the continuous adjustment path network value.

[0145] Based on the device connection channel set, it is necessary to further analyze whether the relative positions between the devices and fish habitats meet the path construction conditions. First, the fish activity distribution area is mapped and modeled. For example, the high-activity area of ​​a certain aquaculture area is a rectangular grid with a coordinate range of 15 to 35 meters on the X axis and 10 to 20 meters on the Y axis. This corresponds to the deployment location of devices A01 and A02 within this area. If the coordinates corresponding to the device numbers fall within the range of the fish activity grid, then the device is judged to have the function of directly connecting to the fish activity area. Then, the connectivity of all device-centered connection paths is determined in sequence. When the two devices are not only physically connectable, but the intermediate path is not blocked by obstacles in the coordinate map, and the channel line does not cross isolated water areas or closed boundaries, the path is considered traversable. If the line connecting A01 and A02 in a certain area is within the Y axis range of 15 to 16 meters and there is no water boundary blocking the area, then the path is considered continuous. Combining the data of multiple connected paths, a device chain with spatial continuity and complementary functions is established to serve as the continuous adjustment path network value within the aquaculture pond area.

[0146] The solution output submodule classifies and summarizes the control device numbers covered by each path based on the continuous adjustment path network value, and associates the automatic response conditions and trigger cycle requirements to establish an automatic control solution for the breeding area;

[0147] Based on the network value of the continuous adjustment path, the control device information in all cascaded paths is integrated, and their corresponding control logic numbers and response rules are listed one by one. In this example, the path A01→A02→C04 meets the path conditions. A01 is the oxygenation node, A02 is the water flow rate regulation node, and C04 is the water direction guidance node. The three are connected to form a complete automated adjustment link. The startup logic of the three must be uniformly set to a periodic collaborative relationship through the control center. For example, if A01 is set to run for 5 minutes every 30 minutes, A02 is delayed to start 10 seconds after the end of A01 and continues to run for 4 minutes, and C04 takes over and runs for 3 minutes. In this way, a control response sequence is constructed between the three. This sequence is recorded in the control data structure to form a joint structure of path and control logic. This structure is exported through the system and submitted to the controller as the actual operation configuration, and the final output forms the automatic control plan for the aquaculture area.

[0148] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. An intelligent aquaculture automatic control system, characterized in that: The system comprises: The feeding feedback control module obtains the bait residue image, calculates the current bait residue density, calculates the difference between the residual density and the preset feeding density threshold, determines the feeding completion status, and outputs the feeding feedback control signal for the breeding area; The feeding rhythm adjustment module determines the feeding completion state of the breeding area based on the feeding feedback control signal of the breeding area, calculates the feeding interval value, calculates the feeding speed based on the change in feeding density before and after, determines the trigger time offset of the next round of directional feeding, and generates the feeding delay control instruction; The water disturbance partitioning module collects the current water flow velocity and water flow direction based on the feeding delay control instruction, combines the standard coordinate grid in the aquaculture pond, identifies the disturbance area and binds the disturbance mark, calculates the disturbance level, marks the control linkage priority block, and generates the disturbance priority block label set; The control unit identification module locates the water circulation equipment and the oxygenation device according to the disturbance priority block label set, combines the fish activity density image and the change trend of the dissolved oxygen collection point, determines whether the active density change area overlaps with the low oxygen area, determines the core control node of the aquaculture area, and generates an aquaculture control node set.

2. The intelligent aquaculture automatic control system according to claim 1 is characterized in that: The feeding feedback control signal of the breeding area includes a feeding status identifier, a bait residual density index, and a feeding threshold deviation; the bait feeding delay control instruction includes a feeding interval value, a feeding speed parameter, and a trigger time offset; the disturbance priority block label set includes a mainstream path continuity level, a disturbance level label, and a linkage priority block coordinate; the aquaculture control node set includes water flow regulation equipment positioning information, fish activity density analysis results, and a low-oxygen area identification identifier.

3. The intelligent aquaculture automatic control system according to claim 1 is characterized in that: The feeding feedback control module includes: The image acquisition submodule acquires the set water surface partition image through the camera, detects the pixel distribution of the bait residual area in the current image, marks and encodes it by area, extracts the bait coverage area data in each partition image, and obtains the partition bait image area value; The residual density calculation submodule uses the formula: Calculate the bait residual density M under the jth water surface partition rj , comprehensively generate the residual density value of bait in the partition, where represents, A rj Represents the bait image area value in the jth water surface partition, A tj represents the total water surface area value of the jth water surface partition, and n represents the total number of partitions; The feeding status determination submodule calculates the difference between the residual bait density value of the partition and the currently set feeding density threshold, and makes a judgment based on the difference. If the difference is positive, it is determined that the current feeding stage is in a state of excessive residual bait; if it is negative, it is determined to be a state of complete feeding, and a feeding feedback control signal is obtained.

4. The intelligent aquaculture automatic control system according to claim 1, characterized in that: The feeding rhythm adjustment module includes: The interval calculation submodule selects the area that has completed feeding in the current stage based on the feeding feedback control signal of the breeding area, extracts the feeding end time of the previous stage and the feeding start time of the current stage of the corresponding area, calculates the difference between the two time values, and obtains the feeding interval value; The speed measurement submodule uses the formula according to the feeding interval value and the bait residual density value of the current and previous stages, combined with the regional feeding intensity data corresponding to each stage: Calculate the feeding density change rate V s , generate feeding speed data, where M c Indicates the residual density of bait at the current stage, M p Indicates the residual density of bait in the previous stage, I c Indicates the feeding intensity value at the current stage, T c Indicates the current feeding start time, T p Indicates the end time of the previous stage, T k represents the feeding interval value of the kth region, M k represents the current bait residual density value in the kth area, and K represents the total number of areas in the feeding completion state; The delay control submodule calculates the time offset for triggering the next round of directional feeding of bait according to the feeding speed data, superimposes the current time point, sets it as the feeding execution time, records the waiting period before the next round of operation, and establishes the bait feeding delay control instruction.

5. The intelligent aquaculture automatic control system according to claim 1 is characterized in that: The water disturbance zoning module includes: The flow field identification submodule collects the current water velocity value and water flow direction transmitted by the flow velocity probe based on the waiting period set by the feeding delay control instruction, and combines the standard coordinate grid structure in the aquaculture pond to identify the coordinate blocks with continuous mainstream paths and aggregate them to obtain the mainstream path block set; The disturbance calculation submodule obtains the flow velocity value, flow direction angle and number of path connections in each block according to the set of mainstream path blocks, and performs joint calculations through flow velocity difference, direction offset and connectivity scale, using the formula: The disturbance level value L is calculated, and the disturbance intensity of continuous blocks is sorted by the disturbance level value and a label is generated, where represents the disturbance level value, v represents the mean flow velocity of the current mainstream path, u represents the set reference flow velocity value, and α g represents the flow angle of the gth coordinate block, μ represents the average flow angle of all coordinate blocks, G represents the number of coordinate blocks in the mainstream path, Q g Indicates the number of connected paths in the g-th block; The priority screening submodule determines whether each coordinate block meets the disturbance intensity condition according to the disturbance level value, sets a control flag for the block that meets the condition, binds it as a control linkage priority block, and establishes a disturbance priority block label set.

6. The intelligent aquaculture automatic control system according to claim 1, characterized in that: The control unit identification module includes: The equipment positioning submodule searches for the water circulation equipment and aeration devices deployed at each coordinate range, confirms the spatial overlap between the equipment coordinate position and the block range, and obtains the regional equipment index set. The activity matching submodule collects fish activity density images within the corresponding coordinate area based on the regional device index set, monitors the oxygen content change trend of the current dissolved oxygen collection point, determines whether the activity density distribution of the coordinates of each device is synchronized with the oxygen content change trend, and obtains the hypoxia activity overlap value; The node extraction submodule selects the coordinate units where water flow regulation equipment has been deployed in the blocks where the active density change coincides with the hypoxic area based on the hypoxia-active overlap value, records the corresponding equipment number as the core control equipment of the corresponding aquaculture area, and establishes an aquaculture control node set.

7. The intelligent aquaculture automatic control system according to claim 1, characterized in that: The system also includes a control path construction module; The control path construction module sequentially identifies the physical connection channels between the auxiliary oxygenation unit and the circulation device based on the spatial layout information of each core control device in the aquaculture control node set. By combining the connectivity conditions between the fish habitat and activity area and the device location, it constructs an integrated automatic control network of devices in series that forms a continuous water body regulation link, and generates an automatic control plan for the aquaculture area. The automatic control scheme for the breeding area includes the physical connection channel information of the equipment, the connection path of the habitat activity, and the automatic control network topology.

8. The intelligent aquaculture automatic control system according to claim 7, characterized in that: The control path building module includes: The structure recognition submodule identifies adjacent auxiliary oxygenation units and water circulation devices within the control area based on the spatial layout information of each core control device in the aquaculture control node set, extracts physical location coordinates and performs spatial connection judgment to obtain a set of device connection channels; The path connection submodule determines whether the devices meet the continuous action range, adjacent coordinate mapping relationship and linkage response logic based on the device connection channel set and the overlap relationship between the fish habitat and activity area in the aquaculture area and the device distribution. It integrates all the connected units that meet the conditions to obtain the continuous adjustment path network value. The solution output submodule classifies and summarizes the control device numbers covered by each path according to the continuous adjustment path network value, and associates the automatic response conditions and trigger cycle requirements to establish an automatic control solution for the breeding area.

Images