Aquaculture carbon addition amount inversion control method

Abstract

The invention discloses an aquaculture carbon addition amount inversion control method for solving the problem that the carbon source addition amount of aquaculture water is not accurate, and belongs to the technical field of aquaculture and water quality control. Real-time water quality parameters and ecology are utilized to establish a Gaussian-like fuzzy decision-making system to complete a CO2 emission flux prediction model, prediction and actual measurement data are adopted to construct a itransform carbon addition inversion model, the actual measurement data of the culture water body after carbon addition is adopted to complete the inversion model optimization process, and finally accurate prediction of carbon emission is achieved. The addition amount of an aquaculture carbon source is optimized, and the breeding benefit and the environmental benefit are both improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of aquaculture and water quality control, and particularly relates to a method for inverse control of carbon addition amount in aquaculture. Background Art

[0002] During the current high-density aquaculture process, a large amount of artificial compound feed is fed, and the ammonia nitrogen generated will cause eutrophication of the aquaculture water body, which not only increases the aquaculture burden but also causes serious pollution to the environment. Therefore, water quality regulation by adding carbon sources has become an essential process in aquaculture. When treating sewage, denitrification and nitrogen removal are enhanced by adding carbon sources (such as methanol and sodium acetate), and its core lies in promoting microbial metabolism through exogenous carbon supplementation. However, there are insurmountable technical gaps in this method in the aquaculture scenario: the sewage treatment system has a single goal of water quality purification and does not need to consider the survival problem of organisms in the water body, so high-concentration chemical substances can be directly added; while the aquaculture water body needs to meet the survival needs of aquatic organisms such as aquaculture objects at the same time, and the introduction of any exogenous substances needs to strictly evaluate the ecological risks. Excessive carbon addition may lead to a sudden drop in dissolved oxygen in the water body, explosive proliferation of algae, and even poisoning or death of aquaculture objects. More critically, the residue of chemical carbon sources will directly affect the safety of aquatic products, violating the requirements of modern aquaculture for green production. Compared with the "end treatment" logic of sewage treatment, reasonably regulating the carbon source addition amount and optimizing the water body ecology have become necessary tasks in the aquaculture industry. However, traditional aquaculture carbon source management often relies on the empirical judgment or rough estimation of farmers, and it is difficult to accurately determine the carbon source demand under different water quality conditions, which seriously restricts the sustainable development of aquaculture. Therefore, there is an urgent need for a scientific and accurate method for controlling the carbon source addition amount in aquaculture water bodies to achieve green, healthy, and sustainable aquaculture. Summary of the Invention

[0003] Aiming at the deficiencies in the prior art, the present invention provides a method for inverse control of carbon addition amount in aquaculture, a calculation method and a correction strategy for adding carbon sources to water bodies based on a dual-input Gaussian-like fuzzy prediction system and an itransformer inversion system, so as to effectively control the ammonia nitrogen content in aquaculture water bodies and optimize carbon emissions, and try to avoid problems such as blind addition of carbon sources, lack of process control and feedback caused by relying on manual experience or existing methods, and propose a reasonable addition amount of carbon sources by comprehensively considering the factors of the aquaculture water body environment, maintain a good water body ecology, reduce aquaculture costs, improve aquaculture benefits, and achieve scientific and precise aquaculture processes.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] A method for inverse control of carbon addition amount in aquaculture, comprising the following steps:

[0006] Construct a biological ammonia reduction model, including ammonia nitrogen nitrification reaction, microbial respiration reaction and microbial metabolism reaction; let the proportions of ammonia nitrogen nitrification reaction, microbial respiration reaction and microbial metabolism reaction be a, b and c respectively;

[0007] Construct a class Gaussian MIMO fuzzy prediction model, which takes the amount of ammonium ions to be removed and temperature as the input of the class Gaussian MIMO fuzzy prediction model to predict the fuzzy values of a, b and c; obtain the predicted value of CO2 emissions based on the fuzzy values of a, b and c;

[0008] Construct an i-transformer inversion model to predict the carbon element addition amount according to the predicted value of CO2 emissions, the measured amount of ammonium ions and temperature;

[0009] Collect the amount of ammonium ions and temperature in the water body at the initial stage, and use the class Gaussian MIMO fuzzy prediction model to predict the CO2 emissions;

[0010] Interpolate the predicted value of CO2 emissions and divide it into x2 carbon addition stages, use the i-transformer inversion model to predict the carbon element addition amount, and evenly add carbon sources in each stage;

[0011] Measure the amount of ammonium ions and temperature in the water body at the i-th stage, input the i-transformer inversion model to predict the carbon element addition amount at the i-th stage; perform correction with measured data to predict the carbon element addition amount at the i+1-th stage.

[0012] To optimize the above technical solution, the specific measures taken also include:

[0013] Furthermore, the ammonia nitrogen nitrification reaction is expressed by the chemical formula as follows:

[0014]

[0015] In the formula, C x H y O z represents the carbon source molecule, and x, y, z are the numbers of carbon atoms, hydrogen atoms and oxygen atoms in the expression of the added carbon source molecule respectively, represents the ammonium ion;

[0016] The microbial respiration reaction is expressed by the chemical formula as follows:

[0017]

[0018] The microbial metabolism reaction is expressed by the chemical formula as follows:

[0019]

[0020] In the formula, C dH e N f O g represents a protein, where d, e, f, and g are the numbers of carbon atoms, hydrogen atoms, nitrogen atoms, and oxygen atoms in the protein molecule, respectively;

[0021] The ammonium nitrogen nitrification reaction, the microbial respiration reaction, and the microbial metabolism reaction proceed simultaneously in the proportions of a, b, and c. By organizing equations (1), (2), and (3), we obtain

[0022]

[0023] d and f in equation (4) are determined by the internal carbon and nitrogen fixation process of microorganisms.

[0024] Furthermore, the specific process of constructing the class Gaussian MIMO fuzzy prediction model is as follows:

[0025] The inputs of the class Gaussian MIMO fuzzy prediction model include the ammonium ion content and temperature; the outputs of the class Gaussian MIMO fuzzy prediction model include the proportions a, b, and c of the ammonium nitrogen nitrification reaction, the microbial respiration reaction, and the microbial metabolism reaction;

[0026] Fuzzify the inputs and outputs;

[0027] Suppose the temperature is divided into m1 language sets and the ammonium ion content is divided into m2 language sets, obtaining a double-input and triple-output fuzzy prediction system with m1 * m2 rules;

[0028] Perform Mamdani inference and use the centroid method to find the fuzzy values of a, b, and c for the final output.

[0029] Furthermore, the specific method for obtaining the CO2 emission prediction value based on the fuzzy values of a, b, and c is as follows:

[0030] Use the respective scale factors k a , k b and k c to convert the fuzzy values of a, b, and c into actual values. k a represents the scale factor for output a, N dmax represents the actual maximum value of output a, N dmin represents the actual minimum value of output a, n amax represents the maximum value of the output a universe of discourse, n amin represents the minimum value of the output a universe of discourse, k b represents the scale factor for output b, which is used to map the actual range of output b to the corresponding universe of discourse range, B max represents the actual maximum value of output b, B minRepresents the actual minimum value of the output b, n bmax Represents the maximum value of the domain of the output b, n bmin Represents the minimum value of the domain of the output b, k c Represents the scaling factor of the output c, D max Represents the actual maximum value of the output c, D min Represents the actual minimum value of the output c, n cmax Represents the maximum value of the domain of the output c, n cmin Represents the minimum value of the domain of the output c;

[0031] Substitute the actual values of a, b, and c into formula (4) to obtain the predicted carbon dioxide emissions.

[0032] Furthermore, the construction of the i-transformer inversion model is specifically as follows:

[0033] Convert the carbon dioxide emissions into the first time-varying sequence signal x′(t) containing variables a, b, and c through time-reversed interpolation;

[0034] Convert the measured amount of ammonium ions and temperature into the second time-varying sequence signal x″(t) through chronological interpolation;

[0035] Introduce an inverted attention mechanism and a feed-forward network to construct an i-transformer, embed the time points of the time-varying sequence signal into variable tokens, and use the attention mechanism to capture multivariate correlations using variable tokens; at the same time, apply a feed-forward network to each variable token and learn the relationship between variables a, b, c and the change in ammonium ion content.

[0036] Furthermore, the fuzzification of the input and output is specifically as follows:

[0037] Input ammonium ion NH4 + The content range is: Nmin - Nmax (mg / L), N max Represents the actual input The maximum value of the content, N min Represents the actual input The minimum value of the content; the input temperature range is: Tmin - Tmax (°C), T max Represents the maximum value of the actual input temperature, T min Represents the minimum value of the actual input temperature; the output a range is: Ndmin - Ndmax (%), N dmax Represents the actual maximum value of the output a, N dmin Represents the actual minimum value of the output a, the output b range is: Bmin - Bmax (%), B max Represents the actual maximum value of the output b, B minRepresents the actual minimum value of output b. The range of output c is: Dmin - Dmax (%), D max Represents the actual maximum value of output c, D min Represents the actual minimum value of output c;

[0038] The range of the actual input quantity is mapped to the corresponding universe of discourse range by using the scale factors of each input quantity, and the range of the actual output quantity is mapped to the corresponding universe of discourse range by using the scale factors of each output quantity.

[0039] Furthermore, the rule base is specifically as follows:

[0040] If (T i , N j ) = Q ij , then Ndi = a i , Bi = b i , Di = c i ;

[0041] In the formula, T i represents the i-th linguistic value in the m1 linguistic sets divided by temperature, N j represents the j-th linguistic value in the m2 linguistic sets divided by the ammonium ion content, Q ij represents the input combination state composed of the i-th linguistic value T i of temperature and the j-th linguistic value N j of the ammonium ion content. Ndi represents the fuzzy linguistic quantity of the ammonia nitrogen nitrification reaction at the i-th input, a i represents the fuzzy decision output of the ammonia nitrogen nitrification reaction ratio at the i-th input, Bi represents the fuzzy linguistic quantity of the respiration reaction at the i-th input, b i represents the fuzzy decision output of the respiration reaction ratio at the i-th input, Di represents the fuzzy linguistic quantity of the metabolic reaction at the i-th input, c i represents the fuzzy decision output of the metabolic reaction ratio at the i-th input.

[0042] Furthermore, the scale factors of each input quantity are specifically as follows:

[0043]

[0044]

[0045] In the formula, k N represents the scale factor of the content, which is used to map the range of the actually input content to the corresponding universe of discourse range, N max represents the maximum value (mg / L) of the actually input content, N min represents the actually input The minimum value of the content (mg / L), n 1max denote The maximum value of the content domain, n 1min denote The minimum value of the content domain, k T Denote the scale factor of temperature, which is used to map the actual input temperature range to the corresponding domain range, T max Denote the maximum value of the actual input temperature (°C), T min Denote the minimum value of the actual input temperature (°C), n 2max Denote the maximum value of the temperature domain, n 2min Denote the minimum value of the temperature domain

[0046] The scale factors of the above-mentioned output quantities are specifically as follows:

[0047]

[0048]

[0049]

[0050] k a Denote the scale factor of output a, N dmax Denote the actual maximum value of output a, N dmin Denote the actual minimum value of output a, n amax Denote the maximum value of the output a domain, n amin Denote the minimum value of the output a domain, k b Denote the scale factor of output b, which is used to map the actual range of output b to the corresponding domain range, B max Denote the actual maximum value of output b, B min Denote the actual minimum value of output b, n bmax Denote the maximum value of the output b domain, n bmin Denote the minimum value of the output b domain, k c Denote the scale factor of output c, D max Denote the actual maximum value of output c, D min Denote the actual minimum value of output c, n cmax Denote the maximum value of the output c domain, n cmin Denote the minimum value of the output c domain.

[0051] The beneficial effects of the present invention are as follows: The present invention realizes the effective control of ammonia nitrogen content in the aquaculture water body and optimizes the carbon emission, and tries to avoid the problems of blind addition of carbon source, lack of process control and feedback caused by relying on artificial experience or existing methods. The reasonable addition amount of carbon source is proposed by comprehensively considering the factors of the aquaculture water body environment, maintaining a good ecological environment of the water body, reducing the aquaculture cost and improving the aquaculture benefit. Description of the Drawings

[0052] Figure 1 This is the overall flowchart of the method for inverting and controlling the carbon addition amount in aquaculture proposed by the present invention.

[0053] Figure 2 This is the schematic diagram of the i-transformer inversion model. Specific Embodiments

[0054] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0055] Embodiment 1

[0056] The present invention proposes a method for inverting and controlling the carbon addition amount in aquaculture. The overall process of this method is as Figure 1 shown and includes the following steps:

[0057] Construct a biological ammonia reduction model, including ammonia nitrogen nitrification reaction, microbial respiration reaction, and microbial metabolism reaction; let the proportions of ammonia nitrogen nitrification reaction, microbial respiration reaction, and microbial metabolism reaction be a, b, and c respectively.

[0058] The ammonia nitrogen nitrification reaction is expressed by the chemical formula as follows:

[0059]

[0060] In the formula, C x H y O z represents the carbon source molecule, and x, y, and z are the numbers of carbon atoms, hydrogen atoms, and oxygen atoms in the expression of the added carbon source molecule respectively, represents the ammonium ion;

[0061] The microbial respiration reaction is expressed by the chemical formula as follows:

[0062]

[0063] The microbial metabolism reaction is expressed by the chemical formula as follows:

[0064]

[0065] In the formula, C d H e N f O gDenote the protein, where d, e, f, and g are the numbers of carbon atoms, hydrogen atoms, nitrogen atoms, and oxygen atoms in the protein molecule, respectively;

[0066] The ammonia nitrogen nitrification reaction, the microbial respiration reaction, and the microbial metabolism reaction proceed simultaneously in the proportions of a, b, and c. By organizing Equation (1), Equation (2), and Equation (3), we get:

[0067]

[0068] That is to say, under the condition of sufficient dissolved oxygen in the water body and sufficient carbon source, for every 1 unit removed it will produce unit of CO2.

[0069] d and f in Equation (4) are determined by the internal carbon and nitrogen fixation process of microorganisms. Taking the most important organic matter in bacteria, propargyl amino acid C5H7NO2, as an example, then d = 5 and f = 1; substituting it in, we get:

[0070] For every 1 unit removed it will produce unit of CO2. Thus, the relationship between the ammonia nitrogen content and CO2 emissions is established.

[0071] Construct a class-Gaussian MIMO fuzzy prediction model, which uses the amount of ammonium ions to be removed and the temperature as the inputs of the class-Gaussian MIMO fuzzy prediction model to predict the fuzzy values of a, b, and c; based on the fuzzy values of a, b, and c, obtain the predicted value of CO2 emissions;

[0072] Specifically, constructing the class-Gaussian MIMO fuzzy prediction model is as follows:

[0073] The inputs of the class-Gaussian MIMO fuzzy prediction model include the ammonium ion content and the temperature; the outputs of the class-Gaussian MIMO fuzzy prediction model include the proportions a, b, and c of the ammonia nitrogen nitrification reaction, the microbial respiration reaction, and the microbial metabolism reaction; fuzzify the inputs and outputs, specifically:

[0074] The range of the input ammonium ion NH4 + quantity is: Nmin - Nmax (mg / L), where N max represents the maximum value of the actual input content, and N min represents the minimum value of the actual input content; the input temperature range is: Tmin - Tmax (°C), where T max represents the maximum value of the actual input temperature, and T min represents the minimum value of the actual input temperature; the range of the output a is: Ndmin - Ndmax (%), where N dmax represents the actual maximum value of the output a, and Ndmin Indicates the actual minimum value of output a. The range of output b is: Bmin - Bmax (%), B max Indicates the actual maximum value of output b, B min Indicates the actual minimum value of output b. The range of output c is: Dmin - Dmax (%), D max Indicates the actual maximum value of output c, D min Indicates the actual minimum value of output c;

[0075] The amount and temperature of ammonium ions to be removed need to be mapped from the actual range to the universe of discourse range first, and then input into the class Gaussian MIMO fuzzy prediction model using the scale factor mapping.

[0076]

[0077]

[0078] In the formula, k N Indicates The scale factor of the content, which is used to map the actual input Content range to the corresponding universe of discourse range, N max Indicates the actual input The maximum value (mg / L) of the content, N min Indicates the actual input The minimum value (mg / L) of the content, n 1max Indicates The maximum value of the content universe of discourse, n 1min Indicates The minimum value of the content universe of discourse, k T Indicates the scale factor of temperature, which is used to map the actual input temperature range to the corresponding universe of discourse range, T max Indicates the maximum value (°C) of the actual input temperature, T min Indicates the minimum value (°C) of the actual input temperature, n 2max Indicates the maximum value of the temperature universe of discourse, n 2min Indicates the minimum value of the temperature universe of discourse.

[0079] Use the scale factors of each output quantity to map the range of the actual output quantity to the corresponding universe of discourse range. The scale factors of each output quantity are specifically:

[0080]

[0081]

[0082]

[0083] k a Indicates the scale factor of output a, Ndmax Denotes the actual maximum value of output a, N dmin Denotes the actual minimum value of output a, n amax Denotes the maximum value of the universe of discourse of output a, n amin Denotes the minimum value of the universe of discourse of output a, k b Denotes the scale factor of output b, used to map the actual range of output b to the corresponding universe of discourse range, B max Denotes the actual maximum value of output b, B min Denotes the actual minimum value of output b, n bmax Denotes the maximum value of the universe of discourse of output b, n bmin Denotes the minimum value of the universe of discourse of output b, k c Denotes the scale factor of output c, D max Denotes the actual maximum value of output c, D min Denotes the actual minimum value of output c, n cmax Denotes the maximum value of the universe of discourse of output c, n cmin Denotes the minimum value of the universe of discourse of output c.

[0084] Suppose the temperature is divided into m1 language sets and the ammonium ion content is divided into m2 language sets. Based on the language sets, a rule base is established to obtain a double-input triple-output fuzzy prediction system with m1*m2 rules. The rules are expressed as follows:

[0085] If (T i , N j ) = Q ij , then Ndi = a i , Bi = b i , Di = c i ;

[0086] In the formula, T i Denotes the i-th language value in the m1 language sets divided by temperature, N j Denotes the j-th language value in the m2 language sets divided by ammonium ion content, Q ij Denotes the input combination state composed of the i-th language value T of temperature i and the j-th language value N of ammonium ion content j . Ndi denotes the fuzzy language quantity of ammonia nitrogen nitrification reaction at the i-th input, a i Denotes the fuzzy decision output of the ammonia nitrogen nitrification reaction ratio at the i-th input, Bi denotes the fuzzy language quantity of the respiratory reaction at the i-th input, b i Denotes the fuzzy decision output of the respiratory reaction ratio at the i-th input, Di denotes the fuzzy language quantity of the metabolic reaction at the i-th input, c i Denotes the fuzzy decision output of the metabolic reaction ratio at the i-th input.

[0087] Membership functions of the dual-input triple-output fuzzy prediction system. Taking the fuzzy subset of ammonia nitrogen content as an example,

[0088]

[0089] where, represents the correlation degree of ammonia nitrogen content with respect to its fuzzy language subset, N m1_1 represents the actual value of the input ammonia nitrogen content, represents the central value of the fuzzy language subset m1_1 of ammonia nitrogen content, g Nm1_1 , s Nm1_1 are all constants greater than 0.

[0090] Perform Mamdani inference and use the centroid method to find the fuzzy values of a, b, and c for the final output.

[0091] Based on the fuzzy values of a, b, and c, the predicted value of CO2 emissions is specifically:

[0092] Use the respective scale factors k a , k b and k c to convert the fuzzy values of a, b, and c into actual values. k a represents the scale factor for output a, N dmax represents the actual maximum value of output a, N dmin represents the actual minimum value of output a, n amax represents the maximum value of the universe of discourse of output a, n amin represents the minimum value of the universe of discourse of output a, k b represents the scale factor for output b, which is used to map the actual range of output b to the corresponding universe of discourse range, B max represents the actual maximum value of output b, B min represents the actual minimum value of output b, n bmax represents the maximum value of the universe of discourse of output b, n bmin represents the minimum value of the universe of discourse of output b, k c represents the scale factor for output c, D max represents the actual maximum value of output c, D min represents the actual minimum value of output c, n cmax represents the maximum value of the universe of discourse of output c, n cmin represents the minimum value of the universe of discourse of output c;

[0093] Substitute the actual values of a, b, and c into formula (4) to obtain the predicted carbon dioxide emissions.

[0094] Construct an i-transformer inversion model for predicting the amount of carbon element added based on the predicted value of CO2 emissions, the measured amount of ammonium ions, and temperature; the principle of the i-transformer inversion model is as Figure 2 shown, specifically as follows:

[0095] Convert the carbon dioxide emissions into the first time-varying sequence signal x′(t) containing variables a, b, and c through time-reversed interpolation;

[0096] Convert the measured amount of ammonium ions and temperature into the second time-varying sequence signal x″(t) through chronological interpolation;

[0097] Introduce an inverted attention mechanism and a feed-forward network to construct an i-transformer, embed the time points of the time-varying sequence signal into variable tokens, and use the attention mechanism to capture the multi-variable correlation with variable tokens; at the same time, apply a feed-forward network to each variable token and learn the relationship between variables a, b, and c and the change in ammonium ion content.

[0098] Collect the amount of ammonium ions and temperature in the water body at the initial stage, and use the class Gaussian MIMO fuzzy prediction model to predict CO2 emissions;

[0099] After interpolating the predicted value of CO2 emissions, divide it into x2 carbon addition stages, and use the i-transformer inversion model to predict the amount of carbon element added, and evenly add carbon sources in each stage;

[0100] Measure the amount of ammonium ions and temperature in the water body at the i-th stage, input them into the i-transformer inversion model to predict the amount of carbon element added at the i-th stage; perform correction on the measured data to predict the amount of carbon element added at the i + 1-th stage.

[0101] Application cases

[0102] 1. Input-output fuzzification

[0103] Input NH4+ amount (mg / L) range: Nmin - Nmax (0.2 - 0.6 mg / L)

[0104] Input temperature (°C) range: Tmin - Tmax (10 - 40 °C)

[0105] Output a: Ndmin - Ndmax (20 - 50%)

[0106] Output b: Bmin - Bmax (40 - 90%)

[0107] Output c: Dmin - Dmax (30 - 90%)

[0108] 2. Scale factor

[0109] In this embodiment

[0110] In this embodiment

[0111] In this embodiment

[0112] In this embodiment

[0113] In this embodiment

[0114] 3. Rules

[0115] Suppose the temperature is divided into 5 language sets and the ammonia nitrogen content is divided into 5 language sets, then a double-input triple-output fuzzy prediction system with 5 * 5 rules will be obtained, and its rules are expressed as follows: The rule table is established as follows:

[0116] In the table, T represents the set of language variables for the temperature input divided by the set, and N represents the set of language variables for the ammonia nitrogen divided by the input set.

[0117]

[0118]

[0119] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this application can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0120] The above is only the preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the concept of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the technical field, several improvements and refinements made without departing from the principle of the present invention should be regarded as within the protection scope of the present invention.

Claims

1. An inversion control method for the carbon addition amount in aquaculture, characterized in that, Including the following steps: Construct a biological ammonia reduction model, including ammonia nitrogen nitrification reaction, microbial respiration reaction and microbial metabolism reaction; Let the proportions of ammonia nitrogen nitrification reaction, microbial respiration reaction and microbial metabolism reaction be a, b and c respectively; Construct a class-Gaussian MIMO fuzzy prediction model, which uses the amount of ammonium ions to be removed and temperature as the input of the class-Gaussian MIMO fuzzy prediction model to predict the fuzzy values of a, b and c; Based on the fuzzy values of a, b and c, obtain the predicted value of CO2 emissions; Construct an i-transformer inversion model, which is used to predict the carbon element addition amount according to the predicted value of CO2 emissions, the measured amount of ammonium ions and temperature; Collect the amount of ammonium ions and temperature in the water body at the initial stage, and use the class-Gaussian MIMO fuzzy prediction model to predict the CO2 emissions; Interpolate the predicted value of CO2 emissions and divide it into x2 carbon addition stages, use the i-transformer inversion model to predict the carbon element addition amount, and uniformly add carbon sources in each stage; Measure the amount of ammonium ions and temperature in the water body at the i-th stage, input them into the i-transformer inversion model to predict the carbon element addition amount at the i-th stage; Perform correction with measured data to predict the carbon element addition amount at the i+1-th stage.

2. The method for inverting and controlling the carbon addition amount in aquaculture according to claim 1, wherein, The ammonia nitrogen nitrification reaction is expressed by the chemical formula as follows: In the formula, C x H y O z represents a carbon source molecule, where x, y, and z are the numbers of carbon atoms, hydrogen atoms, and oxygen atoms in the expression of the added carbon source molecule respectively, represents an ammonium ion; The microbial respiration reaction is expressed by the chemical formula as follows: The microbial metabolism reaction is expressed by the chemical formula as follows: In the formula, C d H e N f O g represents a protein, and d, e, f, and g are the numbers of carbon atoms, hydrogen atoms, nitrogen atoms, and oxygen atoms in the protein molecule, respectively; The ammonia nitrogen nitrification reaction, microbial respiration reaction and microbial metabolism reaction proceed simultaneously according to the proportions of a, b and c. Rearrange Equation (1), Equation (2) and Equation (3) to obtain d and f in Equation (4) are determined by the internal carbon and nitrogen fixation process of microorganisms.

3. The method for inverting and controlling the carbon addition amount in aquaculture according to claim 1, wherein The specific process of constructing the class-Gaussian MIMO fuzzy prediction model is as follows: The input of the class-Gaussian MIMO fuzzy prediction model includes ammonium ion content and temperature; The output of the class-Gaussian MIMO fuzzy prediction model includes the proportions a, b and c of ammonia nitrogen nitrification reaction, microbial respiration reaction and microbial metabolism reaction; Fuzzify the input and output; Suppose the temperature is divided into m1 language sets and the ammonium ion content is divided into m2 language sets. Based on the language sets, establish a rule base to obtain a double-input triple-output fuzzy prediction system with m1*m2 rules; Perform Mamdani reasoning, and use the centroid method to find the fuzzy values of a, b, and c of the final output.

4. The method for inverting and controlling the carbon addition amount in aquaculture according to claim 2, wherein The specific method of obtaining the predicted value of CO2 emissions based on the fuzzy values of a, b and c is as follows: Adopt each scale factor k a 、k b and k c to convert the fuzzy values of a, b, and c into actual values. k a represents the scale factor for output a, N dmax represents the actual maximum value of output a, N dmin represents the actual minimum value of output a, n amax represents the maximum value of the domain of output a, n amin represents the minimum value of the domain of output a, k b represents the scale factor for output b, which is used to map the actual range of output b to the corresponding domain range, B max represents the actual maximum value of output b, B min represents the actual minimum value of output b, n bmax represents the maximum value of the domain of output b, n bmin represents the minimum value of the domain of output b, k c represents the scale factor for output c, D max represents the actual maximum value of output c, D min represents the actual minimum value of output c, n cmax represents the maximum value of the domain of output c, n cmin represents the minimum value of the domain of output c; Substitute the actual values of a, b and c into Equation (4) to obtain the predicted carbon dioxide emissions.

5. The method for inverting and controlling the carbon addition amount in aquaculture according to claim 1, wherein The specific method of constructing the i-transformer inversion model is as follows: Convert the carbon dioxide emissions into the first time-varying sequence signal x￠(t) containing variables a, b, c by time-reversed interpolation; Convert the measured amount of ammonium ions and temperature into the second time-varying sequence signal x￠￠(t) by time-order interpolation; An i-transformer is constructed by introducing an inverted attention mechanism and a feed-forward network. The time points of the time-varying sequence signal are embedded into the variable tokens, and the attention mechanism uses the variable tokens to capture the multivariate correlations. At the same time, a feed-forward network is applied to each variable token to learn the relationship between variables a, b, c and the change in ammonium ion content.

6. The method for inverting and controlling the carbon addition amount in aquaculture according to claim 3, wherein The specific method for fuzzifying the input and output is as follows: Input ammonium ion NH4 + The content range is: Nmin - Nmax (mg / L), N max Indicates the actual input The maximum value of the content, N min Indicates the actual input The minimum value; the input temperature range is: Tmin - Tmax (°C), T max Indicates the maximum value of the actual input temperature, T min Indicates the minimum value of the actual input temperature; the range of output a is: Ndmin - Ndmax (%), N dmax Indicates the actual maximum value of output a, N dmin Indicates the actual minimum value of output a, the range of output b is: Bmin - Bmax (%), B max Indicates the actual maximum value of output b, B min Indicates the actual minimum value of output b, the range of output c is: Dmin - Dmax (%), D max Indicates the actual maximum value of output c, D min Indicates the actual minimum value of output c; The range of the actual input quantity is mapped to the corresponding universe of discourse range by using the scale factor of each input quantity, and the range of the actual output quantity is mapped to the corresponding universe of discourse range by using the scale factor of each output quantity.

7. The method for inverting and controlling the carbon addition amount in aquaculture according to claim 3, wherein The specific rule base is as follows: If (T i , N j ) = Q ij , then Ndi = a i , Bi = b i , Di = c i ; Where, T i represents the i-th linguistic value in the m1 linguistic sets divided by temperature, N j represents the j-th linguistic value in the m2 linguistic sets divided by the ammonium ion content, Q ij represents the input combination state composed of the i-th linguistic value T of temperature i and the j-th linguistic value N of the ammonium ion content j . Ndi represents the fuzzy linguistic quantity of the ammonia nitrogen nitrification reaction at the i-th input, a i represents the fuzzy decision output of the ammonia nitrogen nitrification reaction ratio at the i-th input, Bi represents the fuzzy linguistic quantity of the respiratory reaction at the i-th input, b i represents the fuzzy decision output of the respiratory reaction ratio at the i-th input, Di represents the fuzzy linguistic quantity of the metabolic reaction at the i-th input, c i represents the fuzzy decision output of the metabolic reaction ratio at the i-th input.

8. The method for inverting and controlling the carbon addition amount in aquaculture according to claim 6, characterized in that, The specific scale factors of each input quantity are as follows: Where k N represents the proportionality factor of the content, which is used to map the actually input content range to the corresponding universe range, N max represents the actually input maximum value (mg / L) of the content, N min represents the actually input minimum value (mg / L) of the content, n 1max represents maximum value of the content universe, n 1min represents minimum value of the content universe, k T represents the proportionality factor of the temperature, which is used to map the actually input temperature range to the corresponding universe range, T max represents the maximum value (°C) of the actually input temperature, T min represents the minimum value (°C) of the actually input temperature, n 2max represents the maximum value of the temperature universe, n 2min represents the minimum value of the temperature universe The specific scale factors of each output quantity are as follows: k a Represents the scale factor of output a, N dmax Represents the actual maximum value of output a, N dmin Represents the actual minimum value of output a, n amax Represents the maximum value of the universe of discourse of output a, n amin Represents the minimum value of the universe of discourse of output a, k b Represents the scale factor of output b, used to map the actual range of output b to the corresponding universe of discourse range, B max Represents the actual maximum value of output b, B min Represents the actual minimum value of output b, n bmax Represents the maximum value of the universe of discourse of output b, n bmin Represents the minimum value of the universe of discourse of output b, k c Represents the scale factor of output c, D max Represents the actual maximum value of output c, D min Represents the actual minimum value of output c, n cmax Represents the maximum value of the universe of discourse of output c, n cmin Represents the minimum value of the universe of discourse of output c.

Images