CN-121986742-A - Closed loop feedback type accurate feeding method for circulating water culture

CN121986742ACN 121986742 ACN121986742 ACN 121986742ACN-121986742-A

Abstract

The invention provides a closed-loop feedback type precise feeding method for circulating water aquaculture, which relates to the technical field of aquaculture and intelligent control, and aims to obtain water quality and fish swarm image data in real time through a multidimensional state sensing system, call a fish swarm health and space distribution model, accurately distinguish health and disease fish swarm and quantify fish swarm aggregation degree, thoroughly subvert the defect of traditional decision solidification, based on the method, the system differentially calculates standard and pathological compensation feeding quantity, combines regional feeding weight coefficient driving equipment to accurately throw according to the need and the region, effectively solves the blind feeding problem of an execution level, calculates feeding benefit evaluation index after feeding, and dynamically fine-adjusts a feed energy conversion coefficient and pathological nutrition enhancement coefficient in the model by taking the feed energy conversion index as a feedback signal, thereby realizing closed-loop self-adaptive optimization of strategies, solving the problem of disconnection system of sensing and executing links, and improving the scientificity, precision and resource utilization rate of feeding.

Inventors

HE RENCAI
KANG SHENGYUN
ZOU TAO
YANG LE
LU GUOXING

Assignees

江西省农业技术推广中心
海南快渔生物科技有限公司

Dates

Publication Date: 20260508
Application Date: 20251025

Claims (10)

1. The closed loop feedback type accurate feeding method for circulating water culture is characterized by comprising the following specific steps of: S1, constructing a culture Chi Duowei state sensing system, and collecting a culture environment data set for reflecting the current state of a culture pond in real time, wherein the culture environment data set comprises water quality state data and fish swarm image data; S2, calling a preset fish school health state identification model, analyzing and processing the fish school image data to identify and distinguish healthy fish school and patient fish school in a feeding area, and respectively counting the corresponding healthy fish school number and patient fish school number; S3, judging whether the current water quality is in a preset normal feeding threshold value range according to the acquired water quality state data, if not, preferentially executing a water quality regulation program until the water quality is recovered to be normal, and if so, executing the next step; s4, executing a differentiated feeding amount calculation strategy based on the fish school health condition determined in the S2; s5, generating feeding execution instructions according to the final feeding quantity generated in the S4, driving the intelligent feeding equipment to complete feeding operation at the time, and simultaneously calculating feeding benefit evaluation indexes in real time.
2. The closed-loop feedback type accurate feeding method of recirculating aquaculture of claim 1, wherein in S1, constructing a aquaculture Chi Duowei state sensing system comprises: At least one multi-parameter water quality sensor is arranged in the culture pond and used for monitoring and collecting water quality state data including pH value, dissolved oxygen, ammonia nitrogen concentration and water temperature in real time; And at least one high-definition image acquisition module is arranged above a preset feeding area of the culture pond and used for periodically or continuously shooting and acquiring image data of the fish shoal.
3. The closed loop feedback type accurate feeding method of circulating water culture according to claim 2, wherein in the step S2, the fish school health state identification model is a deep learning model based on a convolutional neural network, and the model can output health state labels of fish school individuals including health and patients by inputting the fish school image data; The judgment basis of the health state label comprises the body surface characteristics, the swimming posture and the outlier behavior of the fish.
4. The closed loop feedback type accurate feeding method for recirculating aquaculture according to claim 3, wherein in S2, a fish shoal space distribution analysis module is further called, the fish shoal image data are processed to obtain the fish shoal aggregation degree in each preset feeding subarea, and the fish shoal aggregation degree is an index for quantifying the concentration degree of fish shoals in a specific space.
5. The closed-loop feedback type precise feeding method for circulating water culture according to claim 4, wherein the method comprises the following steps of: The feeding amount calculation strategy specifically comprises the following steps: a) If the number of the patient fish shoals is zero, a standard feeding prediction model is called, and a standard feeding amount is calculated to be used as a final feeding amount according to the number of the healthy fish shoals and other culture parameters; b) If the number of the sick fish shoal is larger than zero, additionally calling a pathological compensation feeding model on the basis of calculating and obtaining a standard feeding amount, calculating and obtaining a sick fish compensation feeding amount according to the number of the sick fish shoal, and superposing the standard feeding amount and the sick fish compensation feeding amount to generate a final feeding amount.
6. The method for closed-loop feedback type precise feeding for recirculating aquaculture of claim 5, wherein in step a) of S4, the standard feeding prediction model is used for calculating the standard feeding amount by the following calculation process: and comprehensively considering the number of the healthy fish shoals, the average specification of the fish bodies, the current water temperature and the preset feed energy conversion coefficient, carrying out iterative calculation through a preset growth nutrition demand algorithm, and finally determining the feed amount capable of meeting the basic growth metabolism demand of the healthy fish shoals. .
7. The method for closed-loop feedback type precise feeding of recirculating aquaculture of claim 6, wherein in step b) of S4, the calculation process of the pathological compensation feeding model for calculating the compensation feeding amount of the diseased fish is as follows: Multiplying the number of identified patient fish groups by a pathological nutrition fortification factor preset according to the specific type of condition and treatment regimen, the factor being intended to provide additional, rehabilitation-facilitating nutritional or pharmaceutical carriers for the patient fish groups, thereby calculating the amount of feed that needs additional supplementation. .
8. The closed loop feedback type precise feeding method for recirculating aquaculture according to claim 7 is characterized in that a group of regional feeding weight coefficients are calculated and generated according to the shoal aggregation degree of each feeding subarea, and the final feeding amount generated in S4 is multiplied by the regional feeding weight coefficients to determine the specific feeding amount distributed to each feeding subarea, so that intelligent feeding equipment is driven to realize differential precise throwing according to the need and the region.
9. The method of claim 8, wherein in step S3, when it is determined that the current water quality is not within the normal feeding threshold range, the water quality control program specifically comprises automatically suspending step S4 and subsequent feeding steps by the system, and sending a start or adjustment instruction to water quality control equipment linked with the system, including an aerator, a water circulation pump and a medicament dosing pump, until the data fed back by the water quality sensor is restored within the normal feeding threshold range.
10. The closed loop feedback type precise feeding method for recirculating aquaculture according to claim 9, wherein the method comprises the following steps: In the step S5, continuously collecting water quality state data and fish school image data in a preset time window after each feeding operation is completed, and calculating a feeding benefit evaluation index, wherein the feeding benefit evaluation index integrates fluctuation degree of water quality after feeding and feeding activity change of fish school; And meanwhile, periodically taking the feeding benefit evaluation index as a feedback signal, and performing dynamic fine adjustment on the feed energy conversion coefficient in the standard feeding prediction model and the pathological nutrition enhancement coefficient in the pathological compensation feeding model so as to realize closed-loop self-adaptive optimization of feeding strategies.

Description

Closed loop feedback type accurate feeding method for circulating water culture Technical Field The invention relates to the technical field of cultivation and intelligent control, in particular to a closed-loop feedback type accurate feeding method for circulating water cultivation. Background The industrial circulating water culture is a highly intensive modern aquaculture mode, is honored as one of the future directions of aquaculture, but the existing industrial circulating water culture equipment system mainly has the following defects: The traditional feeding devices are mostly open loop systems, and can only execute preset and fixed feeding amount and feeding position. The feed throwing machine lacks of fine mechanical control capability in the feeding process, and the throwing force and coverage area of the feed cannot be adjusted according to the requirements, so that the feed is locally piled up at the feeding point, and uniform distribution in space cannot be realized; the decision-level "solidification" problem-existing feeding strategies are typically based on a fixed schedule and estimated total fish population, a "one-cut" static mode. The mode completely ignores the real-time aggregation behavior and the dynamic change of the ingestion desire of the fish shoals caused by the factors of environment, time and the like, so that the feed waste and the water pollution in the sparse area of the fish shoals, and the feed deficiency and the uneven growth in the dense area of the fish shoals are caused; The problem of "broken links" at the system level is that the prior art lacks an effective "sense-decision-execution" closed loop feedback system. The real distribution state of the shoal of fish in the culture pond cannot be perceived in real time, intelligent decision cannot be made based on the state, and the feeding equipment is driven to perform accurate and dynamic response finally. The sensing and executing links are disjointed. Disclosure of Invention The invention aims to provide a closed-loop feedback type accurate feeding method for circulating water culture, which aims to solve the problems in the background technology. In order to achieve the above purpose, the present invention provides the following technical solutions: a closed loop feedback type accurate feeding method for circulating water culture comprises the following specific steps: S1, constructing a culture Chi Duowei state sensing system, and collecting a culture environment data set for reflecting the current state of a culture pond in real time, wherein the culture environment data set comprises water quality state data and fish swarm image data; S2, calling a preset fish school health state identification model, analyzing and processing the fish school image data to identify and distinguish healthy fish school and patient fish school in a feeding area, and respectively counting the corresponding healthy fish school number and patient fish school number; S3, judging whether the current water quality is in a preset normal feeding threshold value range according to the acquired water quality state data, if not, preferentially executing a water quality regulation program until the water quality is recovered to be normal, and if so, executing the next step; s4, executing a differentiated feeding amount calculation strategy based on the fish school health condition determined in the S2; s5, generating feeding execution instructions according to the final feeding quantity generated in the S4, driving the intelligent feeding equipment to complete feeding operation at the time, and simultaneously calculating feeding benefit evaluation indexes in real time. Further, in the step S1, constructing a state sensing system of the culture Chi Duowei includes: At least one multi-parameter water quality sensor is arranged in the culture pond and used for monitoring and collecting water quality state data including pH value, dissolved oxygen, ammonia nitrogen concentration and water temperature in real time; And at least one high-definition image acquisition module is arranged above a preset feeding area of the culture pond and used for periodically or continuously shooting and acquiring image data of the fish shoal. Further, in the step S2, the fish school health state identification model is a deep learning model based on a convolutional neural network, and by inputting the fish school image data, the model can output health state labels of fish school individuals, including health and patients; The judgment basis of the health state label comprises the body surface characteristics, the swimming posture and the outlier behavior of the fish. Further, in S2, the method further includes calling a fish farm space distribution analysis module, and processing the fish farm image data to obtain a fish farm aggregation degree in each preset feeding subregion, where the fish farm aggregation degree is an index for quantifying the degree of concentrat