Multi-parameter coupled intelligent feeding calculation method and system for aquaculture

Abstract

The invention relates to the technical field of fish culture feeding, and discloses a multi-parameter coupled intelligent feeding calculation method and system for aquaculture, and the method comprises the steps: presetting a culture object classification result and a feed protein content parameter in a control processor, collecting water data through a water quality sensor, collecting a real-time image through an RGB image collection unit, and carrying out the calculation of the real-time image. The average biomass and the real-time number of the breeding objects are calculated through a deep learning target detection algorithm, the real-time breeding density is calculated according to a breeding density formula, and the unique corresponding calculation model is retrieved in the model library according to the breeding object classification result; and substituting the water body data, the average biomass, the real-time quantity, the real-time breeding density and the feed protein content parameters into the selected model for calculation to obtain the feed feeding amount. The problem of estimation deviation caused by an existing feeding method is solved, the calculation result of the feed feeding amount is made to be matched with the actual breeding state, and the accuracy of feeding decision making is improved.

Description

Technical Field

[0001] This invention relates to the field of fish farming feeding technology, specifically to a multi-parameter coupled intelligent feeding calculation method and system for aquaculture. Background Technology

[0002] As a core link in the aquaculture production chain, the scientific and precise nature of feed feeding is not only key to determining the growth rate of farmed animals and shortening the breeding cycle, but also directly dominates the input-output ratio of feed costs and affects the health and stability of the aquatic ecosystem.

[0003] However, existing feeding methods still face technical challenges in achieving precision and intelligence. Currently, feeding methods in aquaculture rely on experience-based judgment, mechanical timers, or metering systems. These are all based on the subjective experience of the farmers and do not automatically adjust the amount of feed to meet the actual needs of the fish. This can lead to overfeeding or underfeeding. Underfeeding will slow down the growth of fish and reduce economic returns, while overfeeding will lead to waste of fish feed, water pollution, and negative impacts on the health of fish.

[0004] Furthermore, aquaculture species are diverse, with significant differences in physiological habits. Warm-water and cold-water fish have completely different response mechanisms to water temperature, and their metabolic rates and feeding desires change differently with temperature. Similarly, carnivorous and herbivorous fish have different nutritional requirements. Carnivorous fish need high-protein feed to support rapid growth, while herbivorous fish need to adapt to low-protein, high-fiber food sources. If a single, universal model is used to calculate the feeding amount for all types of fish, it will produce a large deviation and will not achieve truly precise feeding. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multi-parameter coupled intelligent feeding calculation method and system for aquaculture. This solves the problem that existing feeding methods rely on experience or single parameters, leading to estimation errors that result in significant discrepancies between the calculated feed amount and the actual aquaculture conditions, thus causing insufficient accuracy in feeding decisions.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multi-parameter coupled intelligent feeding calculation method for aquaculture, comprising the following steps: S1. Pre-set the classification results of the aquaculture objects and the feed protein content parameters in the control processor; S2. The water quality sensor is triggered by the control processor to detect and collect water quality data in real time, thereby obtaining water body data; S3. The computer triggers the RGB image acquisition unit to acquire real-time images, and uses a deep learning object detection algorithm to calculate the average biomass and real-time quantity of the cultured objects. S4. The real-time stocking density is obtained by substituting the average biomass and real-time quantity into the stocking density formula. S5. By retrieving the classification results of the aquaculture objects from the model library, a calculation model that uniquely corresponds to the classification results of the aquaculture objects is obtained. S6. The feed input amount is obtained by substituting water body data, average biomass and real-time quantity, real-time stocking density and feed protein content parameters into the selected calculation model.

[0007] Specifically, through steps S1 and S5, a corresponding calculation model is first selected based on the inherent biological characteristics of the cultured species, thus achieving model targeting. Subsequently, through steps S2, S3, and S4, dynamic water data, biomass, quantity, and culture density are acquired in real time. Finally, in step S6, the static biological characteristic selection is coupled with dynamic multi-environmental and biological parameters for calculation. The above method integrates multiple key factors affecting fish feeding, making the feeding amount calculation results more consistent with the actual culture conditions and improving the accuracy of feeding.

[0008] Preferably, the classification results of the aquaculture objects are dietary characteristics and water temperature adaptability characteristics.

[0009] Specifically, dietary characteristics and water temperature adaptability are key biological parameters that determine the basal metabolic rate and feeding intention of aquaculture organisms. By presetting these characteristics, a basis is provided for selecting a matching calculation model in the subsequent step S5.

[0010] Preferably, the dietary characteristics and water temperature adaptability characteristics are divided into four categories: carnivorous, herbivorous, warm-water, and cold-water.

[0011] Specifically, by combining the two dimensions of dietary characteristics and water temperature adaptability, four specific biological classifications were formed. This classification method covers the main physiological differences of common aquaculture species, enabling the subsequent selected computational models to more accurately reflect the feeding needs of different categories of aquaculture species.

[0012] Preferably, the water body data includes water temperature, dissolved oxygen, and pH value data of the aquaculture water body.

[0013] Specifically, water temperature, dissolved oxygen, and pH are key aquatic environmental factors that affect the physiological activities and feeding intensity of aquaculture species. These three core water body data are used as dynamic input parameters and incorporated into the calculation model in step S6 to adjust the feeding amount in real time.

[0014] Preferably, in step S3, the specific steps for performing calculations using a deep learning object detection algorithm include: S301. Analyze the fish image information in the real-time screen to identify and locate individual aquaculture objects in the screen; S302. Count the total number of individual aquaculture objects that have been identified and located to obtain the real-time quantity; S303. Estimate the corresponding individual biomass based on the image features of each identified and located individual aquaculture object; S304. Average all estimated individual biomass to obtain average biomass.

[0015] Specifically, non-contact biological information collection of aquaculture objects was achieved through deep learning target detection algorithms. Steps S301-S302 obtained the real-time quantity through image recognition and counting. Steps S303-S304 estimated the biomass by analyzing the outline and area of ​​individual objects and performed averaging to obtain the average biomass, thus avoiding the stress response and data lag caused by manual sampling measurement to aquaculture objects.

[0016] Preferably, the RGB image acquisition unit is used to acquire information about the aquaculture objects and generate real-time images. The RGB image acquisition unit includes a supplementary light, which is connected to a control processor.

[0017] Specifically, the supplementary light is set up to ensure that the RGB image acquisition unit can acquire clear real-time images under different lighting conditions. The supplementary light is connected to the control processor and can be triggered by the control processor according to the acquisition requirements, providing a high-quality image data source for the accurate calculation of the deep learning object detection algorithm.

[0018] Preferably, the stocking density formula is: ; in, For stocking density, This represents the average biomass. For real-time quantities, The volume of the aquaculture water.

[0019] Specifically, stocking density is an important parameter that affects the living space of cultured organisms, the concentration of dissolved oxygen and metabolites in the water, and thus affects the feeding rate. The average biomass and real-time quantity obtained in step S3 are combined with the known volume of the culture water to calculate the stocking density in real time, and the stocking density is substituted into the calculation model in step S6 as one of the dynamic parameters.

[0020] Preferably, the model library is used to store multiple computational models, including warm-water herbivorous fish models, warm-water carnivorous fish models, cold-water carnivorous fish models, and cold-water herbivorous fish models.

[0021] Specifically, the four calculation models stored in the model library correspond to four categories based on diet and water temperature adaptability, ensuring that different categories of aquaculture objects can call a corresponding formula for calculating the amount of feed.

[0022] Preferably, the calculation model corresponds one-to-one with the four categories, specifically including: Warm-water herbivorous fish model: ; Warm-water carnivorous fish model: ; Cold-water carnivorous fish model: ; Cold-water herbivorous fish model: ; in, This refers to the amount of feed given. For water temperature, Dissolved oxygen content in water bodies pH value of the water body For feed protein content, This represents the average biomass. This is the real-time quantity.

[0023] Specifically, all four calculation models are coupled with biomass parameters, population size parameters, stocking density parameters, water body parameters, and feed protein content parameters. Through these targeted models, all dynamic and static parameters are comprehensively calculated, and the feed feeding amount is finally output, realizing intelligent and precise feeding decisions.

[0024] A multi-parameter coupled intelligent feeding calculation system for aquaculture includes: Circulating water treatment system, variable frequency feeder, computer, breeding pond, supplemental lighting, RGB image acquisition unit, control processor, water quality sensor and variable frequency water pump; The RGB image acquisition unit and supplementary light are installed directly above the aquaculture pond, and the RGB image acquisition unit is connected to a computer. The water quality sensor is immersed in the water of the aquaculture pond and connected to the control processor; The computer is connected to the control processor, and the output of the control processor is connected to the variable frequency feeder, the supplemental lighting, and the variable frequency water pump, respectively. The RGB image acquisition unit is used to acquire image information of the aquaculture objects; The water quality sensor is used to monitor water quality data in aquaculture water bodies; The computer is used to receive RGB image information of the cultured objects to estimate the real-time quantity and average biomass. The control processor is used to receive water quality information from the aquaculture water body and process it to estimate the amount of feed to be given to the aquaculture organisms.

[0025] Specifically, in the system, the RGB image acquisition unit acquires biological image information and water body data respectively. The computer is responsible for running a deep learning object detection algorithm to process the images. The control processor is responsible for receiving water quality data, receiving average biomass and real-time quantity from the computer, performing model selection and final calculation of feeding amount, and outputting control signals to drive the variable frequency feeder, supplemental lighting, and variable frequency water pump. All components work together to form a complete closed loop of data acquisition, analysis and calculation, and automatic execution.

[0026] This invention provides a multi-parameter coupled intelligent feeding calculation method and system for aquaculture. It has the following beneficial effects: 1. This invention couples and calculates multiple parameters such as the biological classification results of the cultured organisms, real-time water data, real-time biomass and quantity, real-time stocking density, and feed protein content. It comprehensively considers multiple key factors affecting feeding, overcomes the estimation deviation caused by existing feeding methods relying on experience or single parameters, and makes the calculated feed amount match the actual breeding status, thus improving the accuracy of feeding decisions.

[0027] 2. This invention utilizes a computer-triggered RGB image acquisition unit to capture real-time images and employs a deep learning target detection algorithm for calculation, thereby achieving non-contact automatic acquisition of the average biomass and real-time quantity of aquaculture objects. This avoids the stress response caused to aquaculture objects by manual harvesting and sampling, ensures the real-time nature of biological data, and provides reliable data input for subsequent accurate calculations.

[0028] 3. In step S1, the present invention presets the classification results of the aquaculture objects, and in step S5, it retrieves a unique corresponding calculation model from the model library based on the classification results of the aquaculture objects. By adopting specific calculation models for different biological categories, the invention achieves differentiation and refinement of feeding strategies, ensures that the calculation logic conforms to the physiological habits of specific aquaculture objects, and improves the scientificity and applicability of the feeding plan. Attached Figure Description

[0029] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the system structure of the present invention; Figure 3 This is a structural diagram of the device of the present invention.

[0030] The components include: 1. Circulating water treatment system; 2. Variable frequency feeder; 3. Computer; 4. Aquaculture pond; 5. Supplemental lighting; 6. RGB image acquisition unit; 7. Control processor; 8. Water quality sensor; and 9. Variable frequency water pump. Detailed Implementation

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] See attached document Figure 1 This invention provides a multi-parameter coupled intelligent feeding calculation method for aquaculture, comprising the following steps: S1. Pre-set the classification results of the aquaculture objects and the feed protein content parameters in the control processor 7.

[0033] Step S1 is the basis for feeding calculation. The control processor 7 pre-stores the biological classification information of the cultured objects, including feeding characteristics and water temperature adaptability characteristics. These characteristics are set according to the physiological habits of the cultured objects to ensure that the subsequent selected feeding calculation model matches the biological characteristics of the cultured objects. At the same time, the protein content parameter of the feed to be fed is preset. The protein content parameter of the feed to be fed is an important indicator of the nutritional composition of the feed and directly affects the calculation of the feeding amount.

[0034] S2. The water quality sensor 8 is triggered by the control processor 7 to detect and collect water quality data in real time, thereby obtaining water body data.

[0035] According to the preset sampling interval or real-time trigger command, the control processor 7 activates the water quality sensor 8. The water quality sensor 8 is immersed in the water of the aquaculture pond 4 to monitor the water data in real time. The water quality sensor 8 transmits the collected water data to the control processor 7. The water data includes the water temperature, dissolved oxygen and pH value of the aquaculture water. These parameters directly reflect the real-time environmental conditions of the aquaculture water and affect the physiological activities and feeding intensity of the aquaculture organisms.

[0036] S3. The computer 3 triggers the RGB image acquisition unit 6 to acquire real-time images, and uses a deep learning object detection algorithm to calculate the average biomass and real-time quantity of the aquaculture objects.

[0037] Computer 3 sends an image acquisition command to RGB image acquisition unit 6, which then acquires real-time images of the aquaculture pond 4. The acquired real-time images are transmitted to computer 3, which runs a deep learning object detection algorithm to analyze and calculate the aquaculture objects in the real-time images.

[0038] Deep learning object detection algorithms are used to identify, locate, and count individual farmed objects in an image, thereby estimating their individual biomass.

[0039] Specifically, step S3 includes the following sub-steps: S301. Analyze the fish image information in the real-time video, identify and locate the individual aquaculture object in the video; S302. Count the total number of individual aquaculture objects that have been identified and located to obtain the real-time quantity; S303. Estimate the corresponding individual biomass based on the image features of each identified and located individual aquaculture object; S304. Average all estimated individual biomass to obtain average biomass.

[0040] S4. The real-time stocking density is obtained by substituting the average biomass and real-time quantity into the stocking density formula.

[0041] Computer 3 or control processor 7 receives the average biomass and real-time quantity of the cultured objects calculated in step S3. At the same time, the volume of the culture water body has been stored in the system as a preset parameter. The average biomass, real-time quantity and volume of the culture water body are substituted into the culture density formula to calculate the real-time culture density.

[0042] The formula for stocking density is: ; in, For stocking density, This represents the average biomass. For real-time quantities, The volume of the aquaculture water.

[0043] S5. By retrieving the classification results of the aquaculture objects from the model library, a calculation model that uniquely corresponds to the classification results of the aquaculture objects is obtained.

[0044] According to the pre-set classification results of the aquaculture objects in step S1, the control processor 7 performs a matching search in the pre-stored model library. The classification results of the aquaculture objects are dietary characteristics and water temperature adaptability characteristics. The dietary characteristics and water temperature adaptability characteristics are divided into four categories: carnivorous, herbivorous, warm-water and cold-water. The model library is used to store multiple calculation models, including warm-water herbivorous fish models, warm-water carnivorous fish models, cold-water carnivorous fish models and cold-water herbivorous fish models. The control processor 7 retrieves the calculation model that uniquely corresponds to the current aquaculture object category.

[0045] S6. The feed input amount is obtained by substituting water body data, average biomass and real-time quantity, real-time stocking density and feed protein content parameters into the selected calculation model.

[0046] The control processor 7 inputs the water body data obtained in step S2, the average biomass and real-time quantity obtained in step S3, the real-time stocking density calculated in step S4, and the feed protein content parameters preset in step S1 into the calculation model selected in step S5.

[0047] The computational model corresponds one-to-one with four categories, specifically including: Warm-water herbivorous fish model: ; Warm-water carnivorous fish model: ; Cold-water carnivorous fish model: ; Cold-water herbivorous fish model: ; in, This refers to the amount of feed given. For water temperature, Dissolved oxygen content in water bodies pH value of the water body For feed protein content, This represents the average biomass. This is the real-time quantity.

[0048] The feed amount for this round is obtained through the above calculation. After the calculation is completed, the control processor 7 sends a control command to the variable frequency feeder 2, causing the variable frequency feeder 2 to perform the feeding operation according to the calculated feed amount.

[0049] Through the method steps S1-S6, the present invention realizes intelligent feeding based on multi-parameter coupling. The present invention comprehensively considers various dynamic and static factors such as the biological classification of the cultured objects, real-time water environment parameters, individual biomass of the cultured objects, population size, culture density, and feed nutrient composition.

[0050] Specifically, by using a deep learning object detection algorithm to analyze the real-time images acquired by the RGB image acquisition unit 6, the average biomass and real-time quantity of the aquaculture objects can be obtained non-contactly and without stress, replacing the existing manual sampling measurement method and improving the real-time performance and accuracy of data acquisition.

[0051] By matching the feeding characteristics and water temperature adaptability of cultured organisms with warm-water herbivorous fish models, warm-water carnivorous fish models, cold-water carnivorous fish models, and cold-water herbivorous fish models, the feeding calculations for cultured organisms with different biological characteristics are ensured to be targeted. The warm-water herbivorous fish model, warm-water carnivorous fish model, cold-water carnivorous fish model, and cold-water herbivorous fish model are coupled with water temperature, dissolved oxygen content, water pH value, feed protein content, real-time number of cultured organisms, average biomass, and real-time stocking density, making the feeding calculation results more accurate.

[0052] See attached document Figure 2 and attached Figure 3 This invention provides a multi-parameter coupled intelligent feeding calculation system for aquaculture, comprising: The circulating water treatment system consists of 1. a variable frequency feeder, 2. a computer, 3. a breeding pond, 4. a supplemental light, 5. an RGB image acquisition unit, 6. a control processor, 7. a water quality sensor, and 9. a variable frequency water pump.

[0053] The RGB image acquisition unit 6 and the supplementary light 5 are installed directly above the breeding pond 4. The RGB image acquisition unit 6 is connected to the computer 3 via a data cable for image data transmission. The supplementary light 5 is connected to the control processor 7 via a control cable and is controlled by the instructions of the control processor 7.

[0054] The water quality sensor 8 is immersed in the water of the aquaculture pond 4. The water quality sensor 8 is connected to the control processor 7 through a communication interface for real-time transmission of water quality data.

[0055] Computer 3 and control processor 7 are connected through a communication interface to realize data exchange and instruction coordination. Computer 3 transmits the image processing results to control processor 7.

[0056] The output of the control processor 7 is connected to the variable frequency feeder 2, the supplementary light 5, and the variable frequency water pump 9 respectively. The control processor 7 drives the variable frequency feeder 2 to feed the feed, controls the switching of the supplementary light 5, and controls the operation of the variable frequency water pump 9 through control signals.

[0057] The circulating water treatment system 1 is connected to the aquaculture pond 4 through a pipeline to realize the circulation treatment of water. The variable frequency water pump 9 is used to drive the water to flow between the aquaculture pond 4 and the circulating water treatment system 1.

[0058] The RGB image acquisition unit 6 is used to acquire image information of the cultured objects in the culture pond 4 and generate real-time images. The RGB image acquisition unit 6 is equipped with an image sensor, an optical lens and an image processing module, and can capture high-resolution RGB color image data.

[0059] The supplementary light 5 is installed near the RGB image acquisition unit 6 to provide auxiliary lighting when the lighting conditions in the aquaculture pond 4 are insufficient. The supplementary light 5 is controlled by the control processor 7 and can be adjusted in brightness, turned on or off as needed to ensure that the RGB image acquisition unit 6 always obtains clear image data.

[0060] The water quality sensor 8 is an integrated multi-parameter sensor used to monitor key environmental parameters in aquaculture water. The water quality sensor 8 can measure water temperature, dissolved oxygen content and pH value in real time. The water quality sensor 8 converts the real-time measured water temperature, dissolved oxygen content and pH value into digital signals and transmits them to the control processor 7 through a communication protocol.

[0061] Computer 3 receives real-time image data from RGB image acquisition unit 6 and runs a pre-installed deep learning object detection algorithm. The deep learning object detection algorithm performs image analysis on the real-time image to identify and locate the aquaculture objects, and then calculates the real-time number and average biomass of the aquaculture objects. The calculation results are transmitted to control processor 7 through the communication interface.

[0062] The control processor 7 pre-stores parameters such as the feeding characteristics, water temperature adaptability, and feed protein content of the cultured organisms. The control processor 7 receives water data from the water quality sensor 8 and the average biomass and real-time number of cultured organisms from the computer 3. Based on the water data, average biomass and real-time number, real-time stocking density, and feed protein content parameters, the control processor 7 selects and performs specific calculations for the feeding amount calculation model to obtain the final feed amount. After the calculation is completed, the control processor 7 sends control commands to the variable frequency feeder 2, the supplemental lighting 5, and the variable frequency water pump 9. In this embodiment, the control processor 7 can be implemented using a DSP processor.

[0063] After receiving the instruction from the control processor 7, the variable frequency feeder 2 feeds the aquaculture pond 4 according to the feed amount set in the instruction. The variable frequency feeder 2 has variable frequency control capability and can adjust the feeding rate and dosage.

[0064] The variable frequency water pump 9, as the actuator, is controlled by the instructions of the control processor 7 to regulate the circulation flow of the water body and ensure effective circulation between the aquaculture water body and the circulating water treatment system 1.

[0065] The circulating water treatment system 1 is responsible for filtering, purifying, and disinfecting the aquaculture water to maintain its quality and provide a suitable growth environment for the aquaculture organisms.

[0066] The system and method provided by this invention can adjust the amount of feed in real time according to the actual needs of the cultured organisms and environmental changes, thereby avoiding feed waste and water quality deterioration caused by overfeeding in existing feeding methods, as well as the problem of slow growth of cultured organisms caused by underfeeding. Through precise feeding, this invention helps to optimize feed utilization, reduce breeding costs, and at the same time help to maintain the stability of the culture water body and reduce the water treatment load.

[0067] The control processor 7 integrates the image analysis results of the computer 3 and the monitoring data of the water quality sensor 8 to uniformly select models and calculate feeding amounts, and directly controls the variable frequency feeder 2 to perform feeding, realizing the automation and intelligence of the feeding process. The data acquisition, processing, decision-making and execution of the entire system form a closed loop, improving the level of precision in aquaculture management.

Claims

1. A multi-parameter coupled aquaculture intelligent feeding calculation method, characterized in that, The method comprises the following steps: S1, setting the breeding object classification result and the feed protein content parameter of the breeding object in advance in the control processor (7); S2, triggering the water quality sensor (8) to detect and collect the water body in real time through the control processor (7) to obtain water body data; S3, triggering the RGB image acquisition unit (6) to collect real-time pictures through the computer (3), and calculating by using a deep learning target detection algorithm to obtain the average biomass and real-time quantity of the breeding object; S4, calculating the real-time breeding density by substituting the average biomass and real-time quantity into the breeding density formula; S5, searching the model library by using the breeding object classification result to obtain a calculation model corresponding to the breeding object classification result uniquely; S6, calculating the feed feeding amount by substituting the water body data, the average biomass and real-time quantity, the real-time breeding density, and the feed protein content parameter into the selected calculation model.

2. The multi-parameter coupled intelligent feeding calculation method for aquaculture according to claim 1, characterized in that, The breeding object classification result is the feeding habit feature and the water temperature adaptability feature.

3. The multi-parameter coupled intelligent feeding computing method for aquaculture according to claim 2, characterized in that, The feeding habit feature and the water temperature adaptability feature are divided into four categories of carnivorous, herbivorous, warm water and cold water.

4. The multi-parameter coupled intelligent feeding computing method for aquaculture according to claim 1, characterized in that, The water body data includes water temperature, dissolved oxygen and pH value data of the breeding water body.

5. The multi-parameter coupled intelligent feeding computing method for aquaculture according to claim 1, characterized in that, In step S3, the specific steps of calculating by using the deep learning target detection algorithm include: S301, analyzing the fish school image information in the real-time picture, identifying and locating a single breeding object in the picture; S302, counting the total number of the identified and located single breeding object to obtain the real-time quantity; S303, estimating the individual biomass corresponding to each identified and located single breeding object according to the image feature of the single breeding object; S304, averaging all estimated individual biomasses to obtain the average biomass.

6. The multi-parameter coupled intelligent feeding computing method for aquaculture according to claim 1, characterized in that, The RGB image acquisition unit (6) is used to acquire the image information of the breeding object and generate real-time pictures, and the RGB image acquisition unit (6) comprises a fill light (5), and the fill light (5) is connected with the control processor (7).

7. The multi-parameter coupled intelligent feeding computing method for aquaculture according to claim 3, characterized in that, The model library is used to store a plurality of calculation models, and the calculation models include a warm water herbivorous fish model, a warm water carnivorous fish model, a cold water carnivorous fish model and a cold water herbivorous fish model.

8. A multi-parameter coupled aquaculture intelligent feeding computing system applied to the multi-parameter coupled aquaculture intelligent feeding computing method of any one of claims 1-7, characterized in that, It comprises: a circulating water treatment system (1), a variable frequency feeding machine (2), a computer (3), a breeding pond (4), a fill light (5), an RGB image acquisition unit (6), a control processor (7), a water quality sensor (8) and a variable frequency water pump (9); The RGB image acquisition unit (6) and the fill light (5) are installed directly above the breeding pond (4), and the RGB image acquisition unit (6) is connected with the computer (3); The water quality sensor (8) is immersed in the water body of the breeding pond (4) and connected with the control processor (7); The computer (3) is connected with the control processor (7), and the output end of the control processor (7) is connected with the variable frequency feeding machine (2), the fill light (5) and the variable frequency water pump (9) respectively; The RGB image acquisition unit (6) is used to acquire the image information of the breeding object; The water quality sensor (8) is used for monitoring water body data of the aquaculture water body; The computer (3) is used for receiving RGB image information of the aquaculture object to estimate real-time quantity and average biomass; The control processor (7) is used for receiving water quality information of the aquaculture water body and processing to estimate the feed feeding amount of the aquaculture object.

Images