Ammonia nitrogen early warning method and system for recirculating aquaculture water body based on fish population behavior feedback

Abstract

The application discloses a kind of based on fish population behavior feedback's circulating water aquaculture water body ammonia nitrogen early warning method and system.The method mainly utilizes computer vision technology and deep learning algorithm to obtain the foreground target of aquaculture object, constructs the model reflecting the change characteristics of water body ammonia nitrogen, and timely early warning to the abnormal concentration of water body ammonia nitrogen through quantitative analysis of fish population behavior.The system structure of the application is simple, and the method can effectively solve the problem that the accuracy of the commonly used water quality detection equipment is not high and the reliability is low, to ensure that the ammonia nitrogen of aquaculture water is within the range of adaptation in a non-invasive form, which is beneficial to fish welfare.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ammonia nitrogen concentration monitoring and fish population behavior analysis, and relates to a warning method and system, in particular to a circulating water aquaculture water body ammonia nitrogen warning method and system based on fish population behavior feedback. BACKGROUND

[0002] With the improvement of people's living standards, the aquaculture industry has experienced its supply and demand dividend and has developed vigorously, and the circulating water aquaculture has become a promising aquaculture mode because of its high breeding density, controllable breeding environment and sustainable development benefits. In order to ensure the healthy growth of the breeding organisms and the quality of aquatic products, the circulating water aquaculture mode relies on various water treatment equipment to accurately control the water quality parameters such as temperature and dissolved oxygen of the breeding environment. However, the characteristics of high breeding density also have higher requirements for water quality monitoring and water pollution treatment. Especially in the breeding process, if the ammonia nitrogen and other pollutants generated cannot be effectively degraded, it will have many adverse effects on the environment and production. Fish living in water bodies with too high ammonia nitrogen content will also have some physiological discomfort, and even lead to abnormal daily behavior, including but not limited to swimming, feeding and population behavior, etc.

[0003] Based on the above problems, the present application provides a circulating water aquaculture water body ammonia nitrogen warning method and system based on fish population behavior feedback, which obtains the foreground target of the breeding object through a deep learning algorithm, real-time acquires the behavior and growth condition of the fish population, and combines the correlation between the water quality ammonia nitrogen concentration to simulate the results as the basis for warning the abnormal ammonia nitrogen concentration of the breeding water body, so as to enable the breeding personnel to make timely adjustments and equipment control, and to ensure suitable water quality conditions for fish growth. SUMMARY

[0004] The purpose of the present application is to provide a circulating water aquaculture water body ammonia nitrogen warning method and system based on fish population behavior feedback, which can obtain the foreground target of the breeding object by using computer vision technology and deep learning algorithm according to the population behavior change characteristics of the fish in the breeding pond and the factors of water temperature and dissolved oxygen in the pond, construct a model reflecting the ammonia nitrogen change characteristics of the water body, and timely warn the abnormal ammonia nitrogen concentration of the water body through quantitative analysis of the population behavior of the fish, so as to ensure that the ammonia nitrogen of the breeding water body is within the adaptive range in a non-intrusive form (non-intrusive means that the technical solution of the present application will not affect the internal body and physiological behavior of the fish), which is beneficial to the welfare of fish breeding.

[0005] The technical solution adopted by the present application is as follows:

[0006] The circulating water aquaculture water body ammonia nitrogen warning method based on fish population behavior feedback specifically comprises:

[0007] The real-time picture of the culture pond is taken by a camera and transmitted to a computer, and the foreground target of the cultured object is obtained through a deep learning algorithm;

[0008] The sensitivity coefficient Γ of fish to ammonia nitrogen is calculated by using the foreground target of the cultured object;

[0009] The standard deviation of the speed of all fish in the field of view is calculated to obtain the coefficient Ω representing the dispersion degree of the fish population s ;

[0010] The values of water temperature and dissolved oxygen are read in real time, and the quantitative relationship between the fish population behavior feedback and the median lethal concentration of fish to ammonia nitrogen is calculated according to the sensitivity coefficient Γ of fish to ammonia nitrogen and the coefficient Ω representing the dispersion degree of the fish population s When the ammonia nitrogen concentration in the culture water body reaches the preset concentration, the alarm is triggered for early warning. The preset concentration is set according to experience, and is generally set to half of the median lethal concentration according to specific circumstances.

[0011] Further, the foreground target of the cultured object includes the body length of the fish, the caudal peduncle length of the fish, the swing frequency of the fish tail, the swing amplitude of the fish tail, and the opening and closing frequency of the fish gill.

[0012] Further, the sensitivity coefficient Γ of fish to ammonia nitrogen is calculated according to the following formula:

[0013]

[0014] Wherein, is the opening and closing frequency of the fish gill of the fish when no abnormality occurs, is the swing frequency of the fish tail of the fish when no abnormality occurs, g is the opening and closing frequency of the fish gill of the fish when abnormality occurs, t is the swing frequency of the fish tail of the fish when abnormality occurs, λ is a coefficient related to the change of the body color of the fish, λ ∈ [0, 1], the smaller the influence degree of the body color of the fish under abnormal ammonia nitrogen concentration, the closer λ is to 0, when the body color almost has no change, λ = 0, when the surface of the fish has significant change, λ is closer to 1.

[0015] Further, the coefficient Ω representing the dispersion degree of the fish population s is calculated according to the following formula:

[0016]

[0017] Wherein, n is the number of fish photographed by the camera, is the speed of the i-th fish, is the average speed of all n fish.

[0018] Further, the quantitative relationship between the fish population behavior feedback and the half lethal concentration is calculated by the following formula:

[0019]

[0020] LC 50 is the half lethal concentration of the fish to the ammonia nitrogen in the water body (i.e. the ammonia nitrogen concentration that makes half of the fish die within a specified time), T is the water temperature, DO is the concentration of dissolved oxygen in the aquaculture water body, k is a coefficient related to the selection of the half lethal concentration, when the half lethal concentration is 96h LC 50 , k=1.527; when the half lethal concentration is 48h LC 50 , k=2.915, is the observed amplitude of the tail fin swing of the fish, is the ratio of the length of the tail of the fish to the body length, Γ i is the sensitivity coefficient of the i-th fish to the ammonia nitrogen.

[0021] A circulating water aquaculture water body ammonia nitrogen early warning system based on fish population behavior feedback, comprising:

[0022] A video acquisition module: used for real-time shooting the picture of the aquaculture pond and transmitting to the computer;

[0023] A data acquisition module: used for real-time reading the values of the water temperature and the dissolved oxygen;

[0024] A data processing module: used for calculating the sensitivity coefficient Γ of the fish to the ammonia nitrogen, the coefficient Ω s characterizing the dispersion degree of the fish population, and the quantitative relationship between the fish population behavior feedback and the half lethal concentration of the fish to the ammonia nitrogen;

[0025] An early warning module: used for triggering the alarm to give early warning when the ammonia nitrogen concentration in the aquaculture water body reaches a preset concentration.

[0026] The present application has the following beneficial effects:

[0027] The application provides a circulating water aquaculture water body ammonia nitrogen early warning method and system based on fish population behavior feedback. A deep learning algorithm is used to obtain a foreground target of an aquaculture object, the behavior and growth condition of a fish school are obtained in real time, and the correlation between the behavior and growth condition and the ammonia nitrogen concentration of water quality is combined to make early warning of the ammonia nitrogen concentration of the aquaculture water body based on simulation results, so that the aquaculture personnel can make timely adjustment and equipment control, and suitable water quality conditions are provided for fish growth. Although monitoring various water quality indexes by using a water quality sensor is a good water quality monitoring mode, the sensor used for monitoring ammonia nitrogen often cannot work stably for a long time, and real-time observation data cannot be obtained for a project that needs to be sampled regularly, and the timeliness is greatly reduced. Fish always live in a water environment and are very sensitive to changes in water quality (especially ammonia nitrogen), and their every move can reflect possible changes in the ammonia nitrogen concentration of the water body in real time and long-term pollution. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 This is a simple example of the circulating water aquaculture water body ammonia nitrogen concentration intelligent early warning system based on fish population behavior feedback in the application. DETAILED DESCRIPTION

[0029] The application will be further described below with reference to the drawings.

[0030] REFERENCE Figure 1 This is a simple example of the circulating water aquaculture water body ammonia nitrogen concentration intelligent early warning system based on fish population behavior feedback in the application, which comprises a breeding pond 1, a circulating water treatment system 2, a computer 3, a camera 4, a remote device 5, an alarm 6, a water temperature sensor 7, and a dissolved oxygen sensor 8. The camera 4 is installed above and on the side of the breeding pond 1 and is connected with the computer 3. After the image data related to the fish phenotype and fish behavior obtained by the camera 4 is processed by the computer, the alarm 7 is controlled. The circulating water treatment system is connected outside the breeding pond. The breeding wastewater passes through various water treatment devices of the circulating water treatment system, including a biological filter for treating ammonia nitrogen and facilities for adjusting water temperature and dissolved oxygen and other water quality parameters, so as to improve the water resource utilization rate of the entire circulating water aquaculture and meet the concept of sustainable development.

[0031] The camera is installed above and on the side of the breeding pond and is connected with the input end of the computer, so that the changes in the phenotype and behavior of the fish in the breeding pond are transmitted to the computer in real time through monitoring of video image data. The output end of the computer is connected with a display and an alarm and an alarm indicator light. In addition, the output end of the computer can be connected with a mobile device in a wireless connection mode.

[0032] The device is applied to intelligent early warning of ammonia nitrogen concentration in aquaculture water body based on fish population behavior feedback, and the method comprises the following steps:

[0033] (1) The camera is used to monitor the overall image of the aquaculture pond in real time, and the video picture is transmitted to the computer in real time;

[0034] (2) The deep learning algorithm is used to obtain the foreground target of the aquaculture object, including fish phenotype data and behavior related data, such as fish length, caudal peduncle length, fish tail swing frequency, fish tail swing amplitude, fish gill opening and closing frequency, etc.;

[0035] (3) The computer calculates the sensitivity coefficient Γ of fish to ammonia nitrogen:

[0036]

[0037] Wherein is the fish gill opening and closing frequency of fish when no abnormality occurs, is the fish tail swing frequency of fish when no abnormality occurs, g is the fish gill opening and closing frequency of fish when abnormality occurs, t is the fish tail swing frequency of fish when abnormality occurs, λ is a coefficient related to the change of fish body color, λ∈[0,1], the smaller the influence degree of fish body color under abnormal ammonia nitrogen concentration, the closer λ is to 0, when the body color almost has no change, λ=0, when the fish body surface has significant change, λ is closer to 1.

[0038] The coefficient Ω representing the dispersion degree of the population is calculated s

[0039]

[0040] Wherein n is the number of fish photographed by the camera, the standard deviation of the speed of all fish in the field of view is calculated to obtain the coefficient Ω representing the dispersion degree of the fish population s .

[0041] (4) The water temperature and dissolved oxygen sensor is used to read the values of water temperature and dissolved oxygen in real time, and the quantitative relationship between fish population behavior feedback and median lethal concentration is calculated:

[0042]

[0043] Wherein T is the water temperature, DO is the concentration of dissolved oxygen in the aquaculture water body; k is related to the selection of median lethal concentration, when the median lethal concentration is 96h LC 50 , k=1.527; when the median lethal concentration is 48h LC 50 , k=2.915; is the observed fish tail fin swing amplitude; The ratio of the caudal peduncle length to the body length of the fish respectively. The mapping relationship between the half lethal concentration of ammonia nitrogen of the fish and the phenotype of the fish and the feedback of the fish population behavior is obtained by computer data fitting.

[0044] (5) When the ammonia nitrogen concentration C A exceeds 0.5 96h LC 50 , the alarm continues to sound for 30 seconds, the display light turns red, and the mobile terminal will receive the signal of abnormal ammonia nitrogen concentration. When C A reduces to 0.5 96h LC 50 , the display light turns green, and the signal of abnormal ammonia nitrogen concentration received by the mobile terminal is released.

[0045] The device of the present application adopts a breeding pond 1, a circulating water treatment system 2, a computer 3, a camera 4, a remote device 5, an alarm 6, a water temperature sensor 7, a dissolved oxygen sensor 8, etc. to constitute a breeding water ammonia nitrogen concentration early warning device. According to the change characteristics of the ammonia nitrogen concentration with the fish population behavior, the ammonia nitrogen content condition can be timely reflected without affecting the normal growth and development of the fish.

[0046] The above disclosure is only a specific embodiment of the present patent, but the present patent is not limited to this. For those skilled in the art, the deformation made without departing from the present application should be considered as belonging to the protection scope of the present application.

Claims

1. A method for ammonia nitrogen early warning of a recirculating aquaculture water body based on fish population behavior feedback, characterized in that, Specifically comprising: Real-time pictures of the culture pond are taken by a camera and transmitted to a computer, and the foreground target of the cultured object is obtained through a deep learning algorithm; The sensitivity coefficient Γ of fish to ammonia nitrogen is calculated by using the foreground target of the cultured object, and the formula is: ， wherein, is the frequency of gill opening and closing of the fish when no abnormality occurs, is the frequency of tail wagging of the fish when no abnormality occurs, is the frequency of gill opening and closing of the fish when an abnormality occurs, is the frequency of wagging of the fish when an abnormality occurs, and λ is a coefficient related to the change in the body color of the fish. By calculating the standard deviation of the speed of all fish in the field of view, a coefficient representing the degree of dispersion of the fish group is obtained , the formula is: ， wherein n is the number of fish photographed by the camera, is the speed of the ith fish, is the average speed of all n fish; The values of water temperature and dissolved oxygen are read in real time, and the ammonia-nitrogen concentration is calculated according to the sensitive coefficient Γ of fish to ammonia-nitrogen and the coefficient characterizing the dispersion degree of fish population The quantitative relationship between the behavior feedback of fish population and the median lethal concentration of ammonia-nitrogen for fish is calculated, and the formula is: ， wherein, is the half lethal concentration of ammonia nitrogen for the fish, i.e. the concentration of ammonia nitrogen at which half of the fish die within a specified time, T is the water temperature, DO is the concentration of dissolved oxygen in the aquaculture water body, k is a coefficient related to the selection of the half lethal concentration, e is the base of the natural logarithm, is the observed amplitude of the caudal fin oscillation of the fish, is the ratio of the length of the caudal peduncle to the length of the body of the fish, is the sensitivity coefficient of the i-th fish to ammonia nitrogen, and when the concentration of ammonia nitrogen in the aquaculture water body exceeds the preset concentration, an alarm is triggered to give a warning.

2. The method according to claim 1, wherein the method is characterized by, The foreground target of the cultured object includes the body length of fish, the caudal peduncle length of fish, the swing frequency of fish tail, the swing amplitude of fish tail, and the opening and closing frequency of fish gill.

3. A circulating water aquaculture water body ammonia nitrogen early warning system based on fish population behavior feedback, characterized in that, The system for implementing the method of any one of claims 1-2 comprises: A video acquisition module for real-time shooting of the picture of the culture pond and transmission to the computer; A data acquisition module for real-time reading of the values of water temperature and dissolved oxygen; Data processing module: for calculating the sensitivity coefficient Γ of fish to ammonia nitrogen, the coefficient representing the dispersion degree of fish population and the quantitative relationship between the fish population behavior feedback and the median lethal concentration of fish to ammonia nitrogen; An early warning module for triggering an alarm to give an early warning when the ammonia nitrogen concentration in the cultured water body reaches a preset concentration.

Images