Carbon beneficiation method for oyster culture

Abstract

The invention relates to an oyster culture carbon popularity method, which is specifically applied to the field of mariculture, and aims to realize quantification and management of carbon popularity value by accurately calculating and predicting carbon fixed quantity in an oyster culture process, and obviously improve the accuracy of oyster growth prediction by adopting an improved Gompertz model and an LSTM neural network. Compared with a traditional method, the growth prediction error of the method is reduced by about 30%. Multi-dimensional environmental parameter monitoring and analysis including water temperature, salinity, pH value, dissolved oxygen and the like are introduced, and the influence of environmental factors on oyster growth is comprehensively evaluated. Therefore, the cultivation management is more accurate, and the survival rate and the growth speed of oysters are improved. Accurate calculation of the carbon fixation efficiency in different growth stages is realized through a dynamic stage model; experimental data show that the carbon fixing amount measured and calculated by the method is 20-30% higher than that measured and calculated by a traditional method; based on real-time data and a prediction model, the system can dynamically adjust a breeding strategy.

Description

Technical Field

[0001] The present invention relates to a carbon inclusive method for oyster aquaculture, which is specifically applied to the field of marine aquaculture and aims to achieve quantification and management of carbon inclusive value by accurately calculating and predicting the carbon fixation amount in the oyster aquaculture process. Background Art

[0002] In recent years, as the global climate change problem becomes increasingly serious, marine carbon sinks have attracted widespread attention as an important means to address climate change. As an important component of marine carbon sinks, oyster farming not only has significant ecological value, but also can bring considerable economic benefits to coastal areas. However, traditional oyster farming methods still have many problems in terms of carbon sink efficiency and economic benefits.

[0003] At present, the research on carbon sequestration in oyster aquaculture mainly focuses on the measurement and evaluation of carbon fixation. For example, some scholars have conducted a life cycle assessment of carbon fixation in oyster aquaculture along the coast of China. Although this study provides valuable data, it mainly focuses on static assessment and lacks consideration of dynamic changes in the aquaculture process.

[0004] There are several major problems with traditional oyster aquaculture carbon sequestration methods:

[0005] 1. Insufficient accuracy of growth prediction models: Commonly used growth curve models (such as the Gompertz model) fail to fully consider the dynamic impact of environmental factors, resulting in a large deviation between the prediction results and the actual situation.

[0006] 2. Incomplete analysis of environmental factors: Most studies only consider basic parameters such as water temperature and salinity, ignoring factors such as dissolved oxygen and pH that have important effects on oyster growth and carbon fixation.

[0007] 3. Inaccurate calculation of carbon fixation efficiency: Existing methods often use a fixed carbon fixation coefficient, which fails to reflect the differences in carbon fixation capacity of oysters at different growth stages.

[0008] 4. Lack of real-time optimization mechanism: Traditional farming methods make it difficult to adjust farming strategies in a timely manner according to environmental changes and oyster growth conditions, resulting in inefficient resource utilization.

[0009] 5. The assessment of carbon benefit value is not comprehensive: Most studies only focus on carbon fixation, ignoring the comprehensive value of oyster farming in pollutant reduction, ecosystem services, etc. Summary of the invention

[0010] The present invention proposes an innovative carbon inclusive method for oyster aquaculture. The method aims to solve the following technical problems: how to improve the accuracy of oyster growth prediction, how to comprehensively consider the impact of environmental factors on oyster growth and carbon fixation, how to accurately calculate the carbon fixation efficiency at different growth stages, how to achieve dynamic optimization of aquaculture strategies, and how to comprehensively evaluate the carbon inclusive value of oyster aquaculture.

[0011] The present invention provides a carbon inclusive method for oyster aquaculture, comprising the following steps:

[0012] (1) Design experiments based on actual needs to obtain multiple sets of raw data on seawater quality, carbon emissions, and carbon fixation efficiency during oyster aquaculture, clean and standardize the obtained raw data, and obtain a standardized data set;

[0013] (2) The oyster growth curve was fitted by the improved Gompertz model. After the fitting results were obtained, the model parameters were optimized and the fitting was verified based on the different growth states of the actual oyster growth curve.

[0014] (3) Construct a multivariate nonlinear regression model, fit the key parameters in the oyster growth curve according to the influencing factors - experimental related environmental parameters, obtain the nonlinear regression function describing the key parameters and environmental parameters, and construct an oyster farming data set;

[0015] (4) Construct a dynamic stage model and calculate the stage carbon fixation efficiency: Based on the weight in the initial and final stages, calculate the weight changes in the early, middle and late stages of oyster growth, and calculate the carbon fixed by a single oyster in each stage; introduce a dynamic stage model as a calculation method for stage carbon fixation efficiency, and calculate the carbon sink per unit carbon emission through the stage carbon fixation efficiency;

[0016] (5) The carbon sequestration value was calculated in parallel using an improved total carbon fixation calculation method: based on the stage-by-stage carbon fixation efficiency of the dynamic stage model and the proportion of atmospheric pollutant carbon fixed by oysters at each growth stage, the carbon fixation amount, actual carbon sink value, and total carbon fixation value were calculated respectively;

[0017] (6) LSTM neural network and Isolation Forest algorithm are used to detect and adjust outliers in aquaculture data; LSTM neural network fitting is used to detect outliers in aquaculture data and feed them back to the model for reconstruction, and finally outliers are adjusted through model fitting;

[0018] (7) Generate aquaculture transformation plans based on spatial differences and carbon fixation differences, analyze the carbon fixation capacity of aquaculture plots in time and space, change the aquaculture scale of suitable aquaculture sites and optimize the spatial layout according to historical management plans; evaluate the performance of the prediction model based on the actual measurement values ​​of historical experiments, verify the validity of the prediction results, and continuously optimize the experimental model; optimize aquaculture strategies based on predicted marine environmental factors and carbon fixation capacity, guide the establishment of a carbon inclusive value system in different sea areas, establish models of the most suitable regions, suitable growth cycles, and suitable aquaculture areas, and provide data-based suggestions through the model to achieve the design of the most suitable oyster aquaculture area.

[0019] Specifically, when fitting the dynamic growth curves of oysters at different growth stages, a dynamic growth curve fitting process combining standard growth curve fitting, probability density function fitting, and growth index prediction was used to obtain the time required for a single oyster to reach sexual maturity from the beginning of cultivation, and the amount of calcium and carbon absorbed by a single oyster at different growth stages; a multi-factor sensitivity analysis was performed on the key parameters affecting oyster growth; among them, the probability density function fitting adopted the normal distribution:

[0020]

[0021] Among them, x is the observed value, μ is the mean value, and σ is the standard deviation; multi-factor sensitivity analysis uses analysis of variance (ANOVA):

[0022]

[0023] Among them, F is the F statistic, MSB is the between-group mean square, and MSW is the within-group mean square.

[0024] Specifically, the gradient descent method and the stochastic gradient descent method are used to converge the errors of the predicted results of calcium and carbon absorbed by oysters at different growth stages, and the calcium and carbon absorbed by each oyster at different growth stages are obtained. Based on the weight in the initial stage and the final stage, the weight changes in the early stage, middle stage and late stage of oyster growth are determined respectively, and the polluted carbon fixed in each stage is obtained; wherein, the update formula of the gradient descent method is:

[0025]

[0026] Among them, θ is the model parameter, α is the learning rate, is the gradient of the loss function with respect to the parameters.

[0027] Specifically, parallel computing was used to adjust the growth status of different oysters with dynamic adjustment factors, optimize the dynamic learning ability of the model, and obtain the carbon fixation rate and total carbon fixation in each stage of the growth cycle; the dynamic adjustment factor calculation formula is:

[0028]

[0029] Among them, AF is the dynamic adjustment factor, β is the adjustment coefficient, Observed_Growth is the observed growth, and Predicted_Growth is the predicted growth.

[0030] Specifically, the parallel computing method is used to obtain the changes in carbon dioxide emissions caused by climate change in the next 50 years. The climate factors caused by climate change are studied using climate models and adaptive climate models as inputs to achieve the coupling of different models such as climatology models and emission models. Based on the existing models, the climate change and greenhouse gas emissions in the next 50 years are simulated. The coupled climate model is used to develop future climate scenarios and analyze the impact risks of greenhouse gas emissions on future climate. Among them, the basic equation of the coupled climate model is:

[0031]

[0032] Where T is the global average temperature, t is the time, Q is the solar constant, α is the Earth's albedo, ∈ is the Earth's effective emissivity, σ is the Stefan-Boltzmann constant, and F is the radiative forcing caused by greenhouse gases.

[0033] Specifically, the calculation process is iterated multiple times and optimized in combination with the LSTM neural network model. The spatial environmental characteristics and time change information are introduced. The absorption and conversion of different pollutants in different sea areas in the marine environment are fully considered. The LSTM neural network model is used for deep learning, and an environmental prediction model based on the marine space domain is constructed for data mining. Among them, the prediction model of the LSTM network is:

[0034] y(t)=f(x(t),x(t-1),...,x(tn)),

[0035] Among them, y(t) is the predicted value at time t, x(t) is the input feature at time t, f is the nonlinear function defined by the LSTM network, and n is the historical time step considered.

[0036] Specifically, the random forest model is used to describe the dynamic changes of different elements in the marine environment caused by different characteristics of aquaculture methods. The law model of carbon changes is described through the changes in the growth rate and physiological behavior of marine organisms, and an environmental model of growth curve characteristics is constructed; the Isolation Forest model is used to extract features of the growth curve, and a deep neural network is used for learning; the environmental data is used as the basis for deep learning, feature extraction and prediction of the deep neural network to realize the construction of a prediction model of marine environmental elements of the growth curve; the basic formula of the random forest model is:

[0037]

[0038] Among them, f(x) is the prediction result of random forest, B is the number of decision trees, fb(x) is the prediction result of the b-th decision tree; the Isolation Forest anomaly detection score is:

[0039]

[0040] Among them, s(x,n) is the anomaly score, E(h(x)) is the average path length of sample x, and c(n) is the average path length when the sample set size is n.

[0041] Specifically, the actual carbon fixation value and the future carbon fixation value obtained through carbon emission prediction are used to obtain the true carbon fixation value, and the corresponding compensation ratio is given based on the prediction of the global carbon sink. The standard growth curve and probability density function are introduced to fit the growth process, and the Gompertz model is preferably used as the growth curve function to realize the division of individual development parameters and stages in the calculation. The relevant parameters are input into the random forest model to complete the optimization of the calculation process of each parameter and obtain the staged carbon fixation efficiency.

[0042] Specifically, the carbon inclusive value includes the pollutant reduction value, carbon sink amount and the value of achieving regional governance. The carbon sink estimation of oyster aquaculture is based on the dynamic fitting of the growth stage. The carbon emission value calculation includes the current value, future value, prevention value, emission reduction value and carbon emission management cost of atmospheric pollutants. The oyster aquaculture carbon inclusive method is divided into an atmospheric pollutant indicator data management system, a carbon emission and carbon sink accounting and prediction system and a carbon inclusive value accounting system. The atmospheric pollutant indicator data management system includes atmospheric pollutant monitoring equipment and an atmospheric pollutant computing cloud and storage equipment, which include program codes capable of collecting, calculating and summarizing atmospheric pollutant data.

[0043] Specifically, the atmospheric pollutant monitoring equipment monitors the changing patterns of atmospheric pollutants and carbon inclusive value, as well as the carbon sink amount and carbon sink benefit based on the atmospheric pollutants and carbon inclusive value accounting method; the monitoring based on the changing patterns of atmospheric pollutants and carbon inclusive value, as well as the carbon sink amount and carbon sink benefit includes: a) monitoring and calculating the carbon inclusive value accounting method; b) monitoring marine ecological and environmental changes; through the analysis of the data obtained from the monitoring and the carbon inclusive value accounting method, a model of the most suitable area and suitable aquaculture area is established, and data-based suggestions are given through the model to increase the suitable aquaculture scale in the suitable aquaculture area to achieve the most suitable oyster aquaculture area. domain design; (2) the carbon emission and carbon sink accounting and prediction system is built with an aquaculture carbon sink algorithm for iterative algorithm optimization. The calculation process outputs the carbon fixation amount at each stage and the overall carbon fixation amount, and the output system compares the actual aquaculture with the calculated carbon fixation amount at each stage and the overall carbon fixation amount based on the data to obtain the carbon benefit value; (3) the carbon benefit value accounting system collects historical data, obtains seasonal seawater movement forecast information for each stage in the future based on predicted climate change data and pollution conditions, and calculates the carbon sink amount based on the proportion of different types of pollutants and the carbon sequestration capacity and carbon sink benefits of different types of pollutants.

[0044] The present invention achieves the following beneficial effects through innovative data collection, model building and system integration:

[0045] 1. Improve prediction accuracy: The improved Gompertz model and LSTM neural network are used to significantly improve the accuracy of oyster growth prediction. Experimental results show that the growth prediction error of this method is reduced by about 30% compared with the traditional method.

[0046] 2. Comprehensive environmental analysis: Introducing multi-dimensional environmental parameter monitoring and analysis, including water temperature, salinity, pH value, dissolved oxygen, etc., to comprehensively evaluate the impact of environmental factors on oyster growth. This makes aquaculture management more precise and improves the survival rate and growth rate of oysters.

[0047] 3. Accurate carbon fixation calculation: Through the dynamic stage model, the carbon fixation efficiency at different growth stages is accurately calculated. Experimental data show that the carbon fixation calculated by this method is 20-30% higher than that of traditional methods.

[0048] 4. Real-time optimization strategy: Based on real-time data and prediction models, the system can dynamically adjust the breeding strategy, such as adjusting the breeding density, feeding amount, etc. This not only improves the efficiency of resource utilization, but also reduces the risk of breeding.

[0049] 5. Comprehensive value assessment: This method takes into account multiple factors such as carbon fixation, pollutant reduction and ecosystem services, and achieves a comprehensive assessment of the carbon benefit value of oyster aquaculture. This provides a scientific basis for the formulation of relevant policies and incentives.

[0050] 6. Improved economic benefits: By optimizing aquaculture management, this method significantly improves the yield and quality of oysters. Experimental results show that the profit margin of farms using this method is 15-25% higher than that of traditional methods.

[0051] 7. Strong scalability: This method designs a modular system architecture, which can be locally adjusted according to the characteristics of different regions, and has good adaptability and promotion value.

[0052] In summary, the carbon inclusive method for oyster aquaculture proposed in this invention not only solves the technical problems existing in traditional methods, but also achieves remarkable results in improving carbon sink efficiency, optimizing aquaculture management, and improving economic benefits. This provides innovative technical support for promoting marine carbon sink management and sustainable aquaculture, and has important practical significance and social value. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is the overall logic block diagram of the present invention.

[0054] Figure 2 It is a logic block diagram of the dynamic phase model of the present invention. DETAILED DESCRIPTION

[0055] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects thereof are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0056] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0057] Please refer to Figure 1-2 The present invention relates to a carbon inclusive method for oyster aquaculture. The method comprises the following steps:

[0058] Step S1: Collecting raw data related to breeding.

[0059] First, the data acquisition module 1 of the present invention designs a corresponding experimental scheme according to actual aquaculture needs. Through various sensors and monitoring equipment arranged in the aquaculture area, multiple groups of original data such as seawater quality, carbon emissions and carbon fixation efficiency during oyster aquaculture are collected. These data include but are not limited to water temperature, salinity, pH value, dissolved oxygen, chlorophyll a concentration, total suspended particulate matter concentration, etc.

[0060] Preferably, in one embodiment of the present invention, the collection frequency is once an hour to ensure the timeliness and continuity of the data. The collected data is stored in the local database of the data collection module 1 after preliminary processing.

[0061] Next, the data acquisition module 1 cleans and standardizes the collected raw data. During the data cleaning process, first remove obvious outliers, such as water temperature data that exceeds the reasonable range (such as water temperature below 0°C or above 40°C). Then, use the moving average method to smooth the data to eliminate the impact of short-term fluctuations. The standardization process uses the Z-score method, that is:

[0062] Z=(X-μ) / σ,

[0063] Among them, X is the original data, μ is the mean value of the data, and σ is the standard deviation. In this way, data of different dimensions can be converted to the same scale, which is convenient for subsequent analysis and modeling.

[0064] Step S2: Fitting the oyster growth curve by the improved Gompertz model.

[0065] After completing the data preprocessing, the growth curve fitting module 2 starts working. This module uses the improved Gompertz model to fit the growth curve of oysters. The mathematical expression of the traditional Gompertz model is:

[0066]

[0067] Where W(t) represents the weight of the oyster at time t, A is the theoretical maximum weight of the oyster, k is the growth rate parameter, t is time, and t i is the inflection point time. In order to improve the fitting accuracy of the model, the present invention improves the traditional Gompertz model. Specifically, the environmental factor correction term E(t) is introduced, and the corrected model expression is:

[0068]

[0069] Where E(t) = 1 + α (T(t) - T opt )+β·(S(t)-S opt ),

[0070] T(t) and S(t) represent the water temperature and salinity at time t, respectively. opt and S opt represent the optimum water temperature and salinity respectively, and α and β are correction coefficients.

[0071] Preferably, in one embodiment of the present invention, T opt The value is 22℃, S optThe value is 28%o, which is the optimal condition for oyster growth based on a large amount of experimental data. The value ranges of α and β are [-0.05, 0.05] and [-0.02, 0.02] respectively, and the specific values ​​are optimized and determined by the least squares method.

[0072] Step S3: Construct a multivariate nonlinear regression model.

[0073] After the growth curve fitting is completed, the multivariate regression module 3 starts working. This module constructs a multivariate nonlinear regression model to describe the relationship between the key parameters in the oyster growth curve and the environmental parameters. The general form of the model is:

[0074]

[0075] Among them, y is the dependent variable (such as growth rate k), x1, x2, x3 are independent variables (such as water temperature, salinity, pH value, etc.), β0, β1, β2, β3 are regression coefficients, and ε is the random error term.

[0076] In order to determine the best regression model, the present invention adopts stepwise regression method. First, a full model including all possible influencing factors is established, and then insignificant variables are gradually deleted according to the significance level of each variable (usually p<0.05) until all retained variables are significant.

[0077] Preferably, in one embodiment of the present invention, the regression model finally determined is:

[0078] k=0.015+0.002·T-0.00005·T 2 +0.001·S-0.00002·S 2 -0.05

[0079] pH + 0.003 pH 2 , where T is water temperature (℃), S is salinity (%o), and pH is the pH value of seawater. The determination coefficient of this model is R 2 It reaches 0.85, indicating that the model has a good fitting effect.

[0080] Step S4: construct a dynamic stage model and calculate the stage carbon fixation efficiency.

[0081] After obtaining the oyster growth curve and the environmental factors influencing model, the dynamic stage model module 4 starts working. The main task of this module is to construct a dynamic stage model and calculate the stage carbon fixation efficiency. First, according to the growth characteristics of oysters, the entire growth cycle is divided into three stages: larval stage (0-2 months), growth stage (2-6 months) and maturity stage (after 6 months). Then, the weight change in each stage is calculated:

[0082] ΔW i =W(t i )-W(t i-1 ),

[0083] Where, ΔW i is the weight gain in the ith stage, W(t i ) and W(t i -1) are the body weights at the end and beginning of the stage respectively. Next, calculate the carbon fixation efficiency of each stage:

[0084]

[0085] Among them, E i is the carbon fixation efficiency of the i-th stage, C i and C i -1 respectively represent the cumulative carbon fixation at the end and beginning of the stage, t i and t i -1 represents the end and start time of this phase respectively.

[0086] Preferably, in one embodiment of the present invention, based on a large amount of experimental data, the average carbon fixation efficiency of oysters at different growth stages is as follows:

[0087] Larval stage: 0.02 g C / (ind·day);

[0088] Growth period: 0.05 g C / (ind·day);

[0089] Maturity: 0.03 g C / (ind·day);

[0090] These values ​​can be used as the initial parameters of the model and then dynamically adjusted according to actual monitoring data.

[0091] Step S5: Calculate the carbon sink value in parallel using the improved total carbon fixation calculation method.

[0092] Based on the dynamic phased model constructed above, the carbon sink value calculation module 5 starts to work. This module uses an improved total carbon fixation calculation method to calculate the carbon sink value in parallel.

[0093] First, calculate the total carbon fixation:

[0094] TC=∑(E i ·P i ·T i ),

[0095] Among them, TC is the total carbon fixation, E i is the carbon fixation efficiency of the i-th stage, P i is the proportion of atmospheric pollutant carbon fixed in the i-th stage, Ti is the duration of the i-th phase. Then, the carbon sink value is calculated:

[0096] CV=TC·P·(1+R) n ,

[0097] Among them, CV is the carbon sink value, P is the carbon price (yuan / ton), R is the annual discount rate, and n is the calculation period.

[0098] Preferably, in one embodiment of the present invention, the carbon price P is 60 yuan / ton (refer to the average price of China's carbon market), the annual discount rate R is 4%, and the calculation period n is 20 years. These parameters can be adjusted according to actual conditions.

[0099] Step S6: Use LSTM neural network and Isolation Forest algorithm to detect and adjust outliers in the breeding data.

[0100] In order to improve data quality and model accuracy, the outlier detection module 6 uses LSTM neural network and IsolationForest algorithm to detect and adjust outliers in the breeding data.

[0101] First, use the LSTM neural network to model time series data. The core formula of LSTM includes:

[0102] f t =σ(W f ·[h t-1 , x t ]+b f ),

[0103] i t =σ(W i ·[h t-1 , x t ]+b i ),

[0104]

[0105]

[0106] o t =o(W o ·[h t-1 , x t ]+b o ),

[0107] h t =o t *tanh(C t ),

[0108] Among them, ft For the forget gate, i t is the input gate, is the candidate state, C t is the cell state, o t is the output gate, h t is the hidden state, σ is the sigmoid function, W and b are the weight and bias parameters.

[0109] Then, the trained LSTM model is used to make predictions on the data, and data points whose predicted values ​​differ from the actual values ​​by more than 3 standard deviations are marked as potential outliers.

[0110] Next, the Isolation Forest algorithm is used to further verify these potential outliers. The formula for calculating the anomaly detection score of Isolation Forest is:

[0111]

[0112] Among them, s(x,n) is the anomaly score, E(h(x)) is the average path length of sample x, and c(n) is the average path length when the sample set size is n. Finally, the data points confirmed as outliers are adjusted. The adjustment method uses local linear interpolation, that is:

[0113]

[0114] Among them, x′ i is the adjusted value, x i-1 and x i+1 are the normal values ​​before and after the abnormal value, t i ,t i-1 and t i+1 is the corresponding time point.

[0115] Step S7: Generate aquaculture transformation plans based on spatial differences and carbon fixation differences.

[0116] Finally, the aquaculture transformation plan generation module 7 generates a suitable aquaculture transformation plan based on the previous analysis results.

[0117] First, the aquaculture plots were evaluated based on spatial differences and carbon fixation differences. The evaluation indicators include water quality suitability index (WQI), carbon fixation potential index (CPI) and economic benefit index (EEI). The calculation formulas for these indices are:

[0118]

[0119] Among them, w i and q iare the weights and standardized values ​​of water quality parameters, TC is the total carbon fixation of the plot, and TC avg is the average carbon fixation of all plots, P is the expected benefit, and C is the cost.

[0120] Then, based on these indicators, a multi-objective optimization algorithm (such as NSGA-II) is used to generate a Pareto optimal solution set as candidate aquaculture transformation schemes.

[0121] Preferably, in one embodiment of the present invention, the aquaculture transformation plan includes the following aspects: adjustment of aquaculture density, optimization of aquaculture cycle, feeding strategy and water quality improvement measures. For example, based on water quality conditions and carbon fixation potential, the optimal aquaculture density is usually 30-50 / m 2 Based on the growth curve and carbon fixation efficiency, the optimal harvest time is generally 8-12 months.

[0122] Step S8: Historical data management and model optimization.

[0123] In order to continuously improve the prediction accuracy, the present invention gradually adds historical data to the data set, and combines the historical management plan to build a historical data management plan based on months. At the same time, the prediction model is adjusted according to the actual experimental conditions, and the meteorological conditions of the predicted ocean are monitored.

[0124] In addition, the present invention also measures the aquaculture platform and collects comprehensive data on marine life in key areas. Through these data, the carbon sink benefits of individual oysters are analyzed and calculated, and the carbon sink benefits of individual platforms are adjusted and calculated based on the changes in the aquaculture scale of individual aquaculture platforms. Finally, the carbon sink benefits of the entire sea area are predicted according to actual needs, and a corresponding aquaculture transformation plan is generated.

[0125] Step S9: Calculation of carbon credit value.

[0126] The present invention comprehensively calculates the value of carbon benefits, including the value of pollutant reduction, carbon sink and the value of regional governance. Based on the dynamic fitting of the growth stage, the carbon sink estimation of oyster aquaculture, the carbon emission value calculation includes the current value, future value, prevention value, emission reduction value and carbon emission management cost of atmospheric pollutants.

[0127] Preferably, in one embodiment of the present invention, the carbon credit value calculation formula is:

[0128] CV=Σ(V i *W i )+CS*P+RGV,

[0129] Among them, CV is the total carbon inclusive value, V i is the value of each item (such as current value, future value, etc.), Wi is the corresponding weight, CS is the carbon sink, P is the carbon price, and RGV is the regional governance value.

[0130] Step S10: System Integration and Implementation

[0131] The carbon inclusive method for oyster aquaculture of the present invention is divided into three main systems: an atmospheric pollutant index data management system, a carbon emission and carbon sink accounting and prediction system, and a carbon inclusive value accounting system.

[0132] The air pollutant index data management system includes air pollutant monitoring equipment, air pollutant computing cloud and storage equipment. The system can collect, calculate and summarize air pollutant data to provide basic data support for subsequent analysis.

[0133] The carbon emission and carbon sink accounting and prediction system has a farming carbon sink algorithm built in it for iterative algorithm optimization. The system calculates and outputs the carbon fixation at each stage and the overall carbon fixation, and compares the calculation results based on actual farming data to derive a more accurate carbon benefit value.

[0134] The carbon credit value accounting system is responsible for collecting historical data and obtaining seasonal seawater movement forecast information for future stages based on predicted climate change data and pollution conditions. The system considers the proportion of different types of pollutants, combines the carbon sequestration capacity and carbon sink benefits of different types of pollutants, and ultimately calculates the carbon sink amount.

[0135] Through the coordinated work of these three systems, the present invention realizes the comprehensive management and precise calculation of carbon benefits in oyster aquaculture, providing strong support for the sustainable development of marine aquaculture.

[0136] The present invention adopts an innovative method to fit the dynamic growth curve of oysters at different growth stages, combining standard growth curve fitting, probability density function fitting and growth index prediction. This comprehensive method can more accurately describe the growth process of oysters.

[0137] First, the standard growth curve fitting adopts the improved Gompertz model, as described in step S2. Then, the probability density function fitting is introduced to describe the distribution characteristics of the growth parameters. Specifically, the normal distribution function is used:

[0138]

[0139] Where x is the observed value (e.g. oyster weight), μ is the mean, and σ is the standard deviation. This better captures the randomness and variability in the growth process.

[0140] Then, through this fitting process, we can obtain the time required for a single oyster from the beginning of cultivation to reach sexual maturity, as well as the amount of calcium and carbon absorbed at different growth stages. For example, in one embodiment, we found that it takes an average of 8 months for oysters to grow from larvae to sexual maturity, and the calcium absorption rate is fastest during the growth period (2-6 months), which is about 0.2g / day, while the carbon absorption is relatively stable throughout the growth cycle, averaging 0.05g / day.

[0141] Finally, multi-factor sensitivity analysis of key parameters affecting oyster growth was performed using analysis of variance (ANOVA):

[0142] F = MSB / MSW,

[0143] Among them, F is the F statistic, MSB is the mean square between groups, and MSW is the mean square within groups. Through this analysis, we can determine which environmental factors have the greatest impact on oyster growth, thereby providing guidance for aquaculture management. For example, we may find that the F values ​​of water temperature and salinity are 15.3 and 12.7, respectively, which are much higher than other factors, indicating that these two factors are the most critical environmental parameters affecting oyster growth.

[0144] In order to further improve the prediction accuracy, the present invention uses the gradient descent method and the stochastic gradient descent method to optimize the prediction results of calcium and carbon absorbed by oysters at different growth stages. The update formula of the gradient descent method is:

[0145]

[0146] Among them, θ is the model parameter, α is the learning rate, is the gradient of the loss function with respect to the parameters. In practical applications, we usually choose a smaller initial learning rate, such as 0.01, and then use a learning rate decay strategy, that is, after a certain number of iterations, multiply the learning rate by a coefficient less than 1 (such as 0.95). This allows you to quickly approach the optimal solution in the early stages of training, and adjust the parameters more finely in the later stages.

[0147] This method allows us to obtain accurate estimates of calcium and carbon absorption by each oyster at different growth stages. For example, in one embodiment, we found that calcium absorption by oysters in the larval stage (0-2 months), growth stage (2-6 months) and mature stage (6 months and above) was approximately 0.1 g / day, 0.2 g / day and 0.15 g / day, respectively, and carbon absorption was approximately 0.02 g / day, 0.05 g / day and 0.03 g / day, respectively.

[0148] Based on the weight at the beginning and the end, we can determine the weight changes in the early, middle and late stages of oyster growth. For example, assuming that the initial weight of an oyster is 0.5g and the final weight is 50g, we can divide its growth process into three stages: 0.5-5g, 5-25g and 25-50g. By calculating the weight gain in each stage and combining it with the carbon absorption rate obtained above, we can derive the amount of polluted carbon fixed in each stage.

[0149] The present invention adopts parallel computing and introduces dynamic adjustment factors to optimize the growth status of different oysters. The calculation formula of the dynamic adjustment factor is:

[0150]

[0151] Among them, AF is the dynamic adjustment factor, β is the adjustment coefficient, Observed_Growth is the observed growth, and Predicted_Growth is the predicted growth. In practical applications, the value of β is usually between 0.1-0.5, and can be adjusted according to actual conditions. For example, if the observed growth is 20% higher than the predicted value, and β is 0.3, then the adjustment factor AF is 1.06, indicating that the output of the prediction model needs to be increased by 6%.

[0152] This dynamic adjustment mechanism can significantly improve the adaptability and prediction accuracy of the model. In this way, we can obtain the carbon fixation rate and total carbon fixation of each stage in the growth cycle. For example, in one embodiment, we may find that the carbon fixation rate of oysters is the highest during the growth period (2-6 months), which is about 0.05g C / (ind·day), while in the larval stage and mature stage, it is 0.02 and 0.03g C / (ind·day), respectively. The total carbon fixation for the entire growth cycle (assuming 10 months) may reach about 12g C / ind.

[0153] The present invention uses parallel computing to simulate the changes in carbon dioxide emissions caused by climate change in the next 50 years. We use a coupled climate model to develop future climate scenarios, and its basic equation is:

[0154]

[0155] Where T is the global average temperature, t is the time, and Q is the solar constant (about 1366W / m 2 ), α is the Earth’s albedo (average of about 0.3), ∈ is the Earth’s effective emissivity (about 0.612), and σ is the Stefan-Boltzman constant (5.67×10 -8 W / (m 2 ·K 4)), F is the radiative forcing caused by greenhouse gases.

[0156] In the model, we considered a variety of greenhouse gas emission scenarios (such as RCP2.6, RCP4.5, RCP8.5, etc.), and combined these scenarios with marine ecosystem models to predict future changes in the marine environment. For example, under the RCP8.5 scenario, we may predict that by 2070, the global average temperature will rise by about 3°C, the sea level will rise by about 0.5m, and the pH value of seawater will drop by about 0.3. These changes will directly affect the growth environment and carbon fixation capacity of oysters.

[0157] This approach allows us to analyze the risk of greenhouse gas emissions affecting future climate change and adjust aquaculture strategies accordingly. For example, we may recommend planning the relocation or transformation of aquaculture sites in advance in areas where sea level rise is predicted to be faster; in areas where severe seawater acidification is predicted, we may need to develop more acid-resistant oyster varieties or take mitigation measures.

[0158] The present invention uses multiple iterations of the calculation process and combines it with the LSTM neural network model for optimization, introduces spatial environmental characteristics and time change information, and fully considers the absorption and conversion of different pollutants in different sea areas in the marine environment. The prediction model of the LSTM neural network is:

[0159] y(t)=f(x(t),x(t-1),…,x(tn)),

[0160] Among them, y(t) is the predicted value at time t, x(t) is the input feature at time t, f is the nonlinear function defined by the LSTM network, and n is the historical time step considered. In practical applications, we usually choose n=24 or n=48, that is, consider the historical data of the previous 1-2 days to predict future environmental changes.

[0161] The advantage of LSTM networks is that they can capture long-term dependencies, which is particularly important for simulating seasonal changes and long-term trends in the marine environment. For example, in one embodiment, we may find that the daily variation pattern of water temperature is significantly different in spring and autumn, and the LSTM network can automatically learn and utilize this seasonal feature to improve prediction accuracy.

[0162] Through this method, we built an environmental prediction model based on the ocean spatial domain and conducted data mining. For example, we may find that water quality parameters (such as dissolved oxygen and pH) in certain sea areas are strongly correlated with the tidal cycle. This information can be used to optimize aquaculture management strategies, such as adjusting the feeding time or water change frequency.

[0163] The present invention uses a random forest model to describe the dynamic changes of different elements in the marine environment caused by different characteristics of aquaculture methods. The basic formula of the random forest model is:

[0164]

[0165] Where f(x) is the prediction result of the random forest, B is the number of decision trees, and fb(x) is the prediction result of the bth decision tree. In practical applications, we usually choose 100-500 decision trees to balance computational efficiency and model performance.

[0166] The advantage of random forest models is that they can handle high-dimensional data and nonlinear relationships, which is particularly important for modeling complex marine ecosystems. For example, we might use a random forest model to predict the effects of different combinations of stocking density, feeding strategies, and water quality parameters on oyster growth rate and carbon fixation efficiency.

[0167] In addition, the present invention also uses the Isolation Forest model to extract features from the growth curve. Through this method, we can identify abnormal growth patterns, such as growth stagnation or sudden acceleration, which may indicate environmental problems or disease outbreaks.

[0168] Finally, we used the environmental data as the basis for deep learning, feature extraction and prediction of deep neural networks to establish a prediction model of marine environmental factors for growth curves. This comprehensive approach can more comprehensively capture the complex relationship between environmental factors, culture methods and oyster growth.

[0169] The present invention uses the actual carbon fixation value and the future carbon fixation value obtained by carbon emission prediction to calculate the real carbon fixation value. Specifically, we use the following formula:

[0170] RCF=ACF*(1+GR) n ,

[0171] Where RCF is the real carbon fixation value, ACF is the actual measured current carbon fixation value, GR is the predicted annual growth rate, and n is the number of predicted years. For example, if the current measured carbon fixation value is 10kg / m 2 / year, the predicted annual growth rate is 2%, considering the situation in the next 20 years, then the actual carbon fixation value is about 14.86kg / m 2 / year.

[0172] Based on the prediction of global carbon sinks, we give the corresponding compensation ratio. The calculation formula of the compensation ratio is:

[0173] CR=(RCF-GCF) / GCF*100%,

[0174] Among them, CR is the compensation ratio, RCF is the real carbon fixation value, and GCF is the global average carbon fixation value. This method can incentivize farmers whose carbon fixation efficiency is higher than the global average and promote the development of the entire industry in a more environmentally friendly direction.

[0175] In addition, the present invention introduces a standard growth curve and a probability density function to fit the growth process, and preferably uses a Gompertz model as a growth curve function. In the calculation, the individual development parameters and stages are divided, and the relevant parameters are input into the random forest model to optimize the calculation process of each parameter, and finally obtain the staged carbon fixation efficiency.

[0176] The present invention comprehensively calculates the value of carbon benefits, including the value of pollutant reduction, carbon sink and the value of regional governance. Based on the dynamic fitting of the growth stage, the carbon sink estimation of oyster aquaculture, the carbon emission value calculation includes the current value, future value, prevention value, emission reduction value and carbon emission management cost of atmospheric pollutants.

[0177] Specifically, we use the following formula to calculate the total carbon inclusive value:

[0178] TCV=PRV+CSV+RGV,

[0179] Among them, TCV is the total carbon benefit value, PRV is the pollutant reduction value, CSV is the carbon sink value, and RGV is the regional governance value.

[0180] The calculation of the Pollutant Reduction Value (PRV) takes into account a variety of pollutants, such as nitrogen, phosphorus, etc.:

[0181] PRV=∑(Q i ·P i ),

[0182] Among them, Q i is the reduction amount of the i-th pollutant, P i The carbon sink value (CSV) is calculated based on the carbon fixation amount and carbon price:

[0183] CSV=CF·CP·(1+DR) n ,

[0184] Among them, CF is the carbon fixation, CP is the carbon price, DR is the discount rate, and n is the calculation period.

[0185] This comprehensive value accounting method can more accurately reflect the ecological and economic value of oyster farming, and provide an important basis for policy formulation and industrial planning.

[0186] The carbon inclusive method for oyster aquaculture of the present invention is divided into three main systems: an atmospheric pollutant index data management system, a carbon emission and carbon sink accounting and prediction system, and a carbon inclusive value accounting system.

[0187] The air pollutant index data management system includes air pollutant monitoring equipment, air pollutant computing cloud, and storage equipment. The system adopts a distributed architecture and can collect, process, and store a large amount of environmental monitoring data in real time. For example, we may deploy multiple water quality sensors in the breeding area to collect data every 5 minutes. These data are transmitted to the computing cloud via wireless network for real-time processing and analysis.

[0188] The carbon emission and carbon sink accounting and prediction system has aquaculture carbon sink algorithm built in it for iterative algorithm optimization. The system uses machine learning methods, such as gradient boosting decision tree (GBDT) or deep neural network, to predict future carbon emissions and carbon sinks. The system will regularly (such as weekly) use the latest monitoring data to update the model to ensure the accuracy and timeliness of the prediction results.

[0189] The carbon inclusive value accounting system is responsible for collecting historical data and obtaining seasonal seawater movement forecast information for future stages based on predicted climate change data and pollution conditions. The system uses time series analysis methods, such as ARIMA or Prophet models, to predict seasonal changes in seawater movement. These forecast results can be used to optimize aquaculture strategies, such as adjusting aquaculture density or selecting the best harvest time.

[0190] Through the coordinated work of these three systems, the present invention realizes the comprehensive management and precise calculation of carbon benefits in oyster aquaculture, and provides strong technical support for the sustainable development of marine aquaculture.

[0191] In summary, the carbon inclusive method for oyster aquaculture of the present invention achieves accurate quantification and prediction of carbon fixation and environmental impacts during oyster aquaculture through innovative data collection, model construction and system integration. This not only provides farmers with tools to optimize management strategies, but also provides a scientific basis for the government to formulate relevant policies, which is of great significance for promoting the sustainable development of marine aquaculture and responding to climate change.

[0192] In order to verify the superiority of the present invention, we conducted a series of experiments to compare the embodiments of the present invention with the comparative examples of the traditional method. The following is a detailed experimental process and result analysis.

[0193] Example 1: We selected an oyster culture area of ​​5 hectares in a sea area in Xiamen, Fujian Province, and applied the oyster culture carbon benefit method of the present invention to conduct a 12-month experiment. We used the improved Gompertz model to fit the growth curve, used the LSTM neural network to predict environmental factors, and used the dynamic stage model to calculate the carbon fixation efficiency. The culture density was set to 40 / m 2 , regularly monitor water quality and record growth conditions.

[0194] Comparative Example 1: In a similar sea area about 2 kilometers away from Example 1, we selected a breeding area of ​​comparable size and used the traditional fixed parameter Gompertz model and simple linear regression to perform growth prediction and environmental factor analysis. The breeding density was the same as in Example 1, but no dynamic adjustment was performed.

[0195] Comparative Example 2: In another adjacent sea area, we selected an oyster culture area that uses conventional culture methods. This culture area does not conduct precise environmental monitoring and model prediction, and is managed only based on experience. The culture density is 35 / m 2 .

[0196] Comparative Example 3: In order to further verify the adaptability of the present invention, we conducted a control experiment in a sea area in Zhoushan Islands, Zhejiang Province. The hydrological conditions in this sea area are different from those in Fujian Sea Area. We still apply the method of the present invention, but do not make local adjustments. The breeding density is set at 45 / m 2 .

[0197] Testing standards and methods:

[0198] 1. Oyster growth rate: 100 oysters were randomly selected each month to measure shell length and weight, and the average monthly growth rate was calculated.

[0199] 2. Survival rate: Count the number of deaths each month and calculate the cumulative survival rate.

[0200] 3. Water quality indicators: Measure water temperature, salinity, pH value, dissolved oxygen and other parameters every week.

[0201] 4. Carbon fixation: estimated by oyster shell weight and soft tissue weight, and measured by monthly sampling.

[0202] 5. Economic benefits: record breeding costs and product sales revenue, and calculate profit margins.

[0203] The following are the test results of each group after 12 months:

[0204] index Example 1 Comparative Example 1 Comparative Example 2 Comparative Example 3 Average shell length (cm) 8.5 7.8 7.2 8.1 Average weight (g) 85 75 68 80 Survival rate (%) 92 85 80 88 <![CDATA[Carbon fixation amount (kg / m 2 )]]> 2.8 2.2 1.9 2.5 Profit margin (%) 35 28 22 31

[0205] From the above results, it can be seen that Example 1 of the present invention has obvious advantages in all indicators. Compared with Comparative Example 1, the average shell length of oysters in Example 1 increased by 9%, and the average weight increased by 13.3%, which is mainly due to the improved Gompertz model we adopted can more accurately describe the growth curve of oysters, thereby optimizing the aquaculture management strategy.

[0206] In terms of survival rate, Example 1 is 7 and 12 percentage points higher than Comparative Examples 1 and 2, respectively. This significant improvement is mainly due to our LSTM neural network environmental factor prediction system, which can warn of adverse environmental changes in advance, allowing farmers to take timely countermeasures.

[0207] Carbon fixation is one of the core indicators of the present invention. The carbon fixation of Example 1 is 27.3% higher than that of Comparative Example 1 and 47.4% higher than that of Comparative Example 2. This result fully demonstrates the excellent performance of our dynamic stage model in improving carbon fixation efficiency. By accurately calculating the carbon fixation efficiency of each growth stage, we can optimize the breeding strategy in a targeted manner and maximize the carbon fixation potential.

[0208] In terms of economic benefits, the profit margin of Example 1 is 7 and 13 percentage points higher than that of Comparative Examples 1 and 2, respectively. This not only reflects a higher yield and survival rate, but also shows that the method of the present invention can effectively reduce breeding costs and improve resource utilization efficiency.

[0209] It is worth noting that although comparative example 3 applies the method of the present invention, due to the lack of optimization adjustment for the local environment, its performance is slightly inferior to that of example 1, but it is still significantly better than comparative examples 1 and 2. This result shows that the present invention has good adaptability and scalability, but when applied in different sea areas, localized adjustment can further improve the effect.

[0210] In summary, the carbon inclusive method for oyster aquaculture of the present invention has shown significant advantages in improving oyster growth rate, survival rate, carbon fixation efficiency and economic benefits. This method based on precise modeling and dynamic adjustment can not only maximize the carbon sink value of oyster aquaculture, but also improve the overall benefits of the aquaculture industry, providing an innovative solution for marine carbon sink management and sustainable aquaculture.

[0211] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A carbon inclusive method for oyster aquaculture, characterized in that: The following steps are involved: (1) Design experiments based on actual needs to obtain multiple sets of raw data on seawater quality, carbon emissions, and carbon fixation efficiency during oyster farming, clean and standardize the raw data, and obtain a standardized data set; (2) The oyster growth curve was fitted by the improved Gompertz model. After the fitting results were obtained, the model parameters were optimized and the fitting was verified based on the different growth states of the actual oyster growth curve. (3) Construct a multivariate nonlinear regression model, fit the key parameters in the oyster growth curve according to the influencing factors-experimental related environmental parameters, obtain the nonlinear regression function describing the key parameters and environmental parameters, and construct an oyster farming data set; (4) Construct a dynamic stage model and calculate the stage carbon fixation efficiency: Based on the weight in the initial and final stages, calculate the weight changes in the early, middle and late stages of oyster growth, and calculate the carbon fixed by a single oyster in each stage; introduce a dynamic stage model as a calculation method for stage carbon fixation efficiency, and calculate the carbon sink per unit carbon emission through the stage carbon fixation efficiency; (5) The carbon sequestration value was calculated in parallel using an improved total carbon fixation calculation method: based on the stage-by-stage carbon fixation efficiency of the dynamic stage model and the proportion of atmospheric pollutant carbon fixed by oysters at each growth stage, the carbon fixation amount, actual carbon sink value, and total carbon fixation value were calculated respectively; (6) LSTM neural network and Isolation Forest algorithm are used to detect and adjust outliers in aquaculture data; LSTM neural network fitting is used to detect outliers in aquaculture data and feed them back to the model for reconstruction, and finally outliers are adjusted through model fitting; (7) Generate aquaculture transformation plans based on spatial differences and carbon fixation differences, analyze the carbon fixation capacity of aquaculture plots in time and space, change the aquaculture scale of suitable aquaculture sites and optimize the spatial layout according to historical management plans; evaluate the performance of the prediction model based on the actual measurement values ​​of historical experiments, verify the validity of the prediction results, and continuously optimize the experimental model; optimize aquaculture strategies based on predicted marine environmental factors and carbon fixation capacity, guide the establishment of a carbon inclusive value system in different sea areas, establish models of the most suitable regions, suitable growth cycles, and suitable aquaculture areas, and provide data-based suggestions through the model to achieve the design of the most suitable oyster aquaculture area.

2. The carbon inclusive method for oyster aquaculture according to claim 1, characterized in that: When fitting the dynamic growth curves of oysters at different growth stages, a dynamic growth curve fitting process combining standard growth curve fitting, probability density function fitting, and growth index prediction was used to obtain the time required for a single oyster to reach sexual maturity from the beginning of cultivation, and the amount of calcium and carbon absorbed by a single oyster at different growth stages; a multi-factor sensitivity analysis was performed on the key parameters affecting oyster growth; among them, the probability density function fitting adopted normal distribution: Among them, x is the observed value, μ is the mean value, and σ is the standard deviation; multi-factor sensitivity analysis uses variance analysis ANOVA: Among them, F is the F statistic, MSB is the between-group mean square, and MSW is the within-group mean square.

3. The carbon inclusive method for oyster aquaculture according to claim 2, characterized in that: The gradient descent method and the stochastic gradient descent method are used to converge the errors of the predicted results of calcium and carbon absorbed by oysters at different growth stages, and the calcium and carbon absorbed by each oyster at different growth stages are obtained. Based on the weight in the initial and final stages, the weight changes in the early, middle and late stages of oyster growth are determined, and the polluted carbon fixed in each stage is obtained; the update formula of the gradient descent method is: Among them, θ is the model parameter, α is the learning rate, is the gradient of the loss function with respect to the parameters.

4. The carbon inclusive method for oyster aquaculture according to claim 1, characterized in that: Using parallel computing, dynamic adjustment factors are used to adjust the growth status of different oysters, optimize the dynamic learning ability of the model, and obtain the carbon fixation rate and total carbon fixation in each stage of the growth cycle; the dynamic adjustment factor calculation formula is: Among them, AF is the dynamic adjustment factor, β is the adjustment coefficient, Observed_Growth is the observed growth, and Predicted_Growth is the predicted growth.

5. The carbon inclusive method for oyster aquaculture according to claim 4, characterized in that: Using parallel computing, we can obtain the changes in carbon dioxide emissions caused by climate change in the next 50 years. We use climate models and adaptive climate models to study the climate factors caused by climate change as input, realize the coupling of different models such as climatology models and emission models, simulate the climate change and greenhouse gas emissions in the next 50 years based on the existing models, use coupled climate models to develop future climate scenarios, and analyze the impact of greenhouse gas emissions on future climate. The basic equation of the coupled climate model is: Where T is the global average temperature, t is the time, Q is the solar constant, α is the Earth's albedo, ∈ is the Earth's effective emissivity, σ is the Stefan-Boltzmann constant, and F is the radiative forcing caused by greenhouse gases.

6. The carbon inclusive method for oyster aquaculture according to claim 1, characterized in that: The calculation process is iterated multiple times and optimized in combination with the LSTM neural network model. The spatial environmental characteristics and time change information are introduced. The absorption and conversion of different pollutants in different sea areas in the marine environment are fully considered. The LSTM neural network model is used for deep learning to build an environmental prediction model based on the marine space domain for data mining. Among them, the prediction model of the LSTM network is: y(t)=f(x(t),x(t-1),...,x(tn)), Among them, y(t) is the predicted value at time t, x(t) is the input feature at time t, f is the nonlinear function defined by the LSTM network, and n is the historical time step considered.

7. The carbon inclusive method for oyster aquaculture according to claim 1, characterized in that: The random forest model is used to describe the dynamic changes of different elements in the marine environment caused by different characteristics of aquaculture methods. The law model of carbon changes is described through the changes in the growth rate and physiological behavior of marine organisms, and an environmental model of growth curve characteristics is constructed; the Isolation Forest model is used to extract features of the growth curve, and a deep neural network is used for learning; the environmental data is used as the basis for deep learning, feature extraction and prediction of the deep neural network to realize the construction of a prediction model of marine environmental elements for the growth curve; the basic formula of the random forest model is: Among them, f(x) is the prediction result of random forest, B is the number of decision trees, fb(x) is the prediction result of the b-th decision tree; the Isolation Forest anomaly detection score is: Among them, s(x,n) is the anomaly score, E(h(x)) is the average path length of sample x, and c(n) is the average path length when the sample set size is n.

8. The carbon inclusive method for oyster aquaculture according to claim 1, characterized in that: The actual carbon fixation value and the future carbon fixation value obtained through carbon emission prediction are used to obtain the true carbon fixation value, and the corresponding compensation ratio is given based on the prediction of the global carbon sink. The standard growth curve and probability density function are introduced to fit the growth process, and the Gompertz model is selected as the growth curve function to realize the division of individual development parameters and stages in the calculation. The relevant parameters are input into the random forest model to complete the optimization of the calculation process of each parameter and obtain the staged carbon fixation efficiency.

9. The carbon inclusive method for oyster aquaculture according to claim 1, characterized in that: The carbon benefit value includes pollutant reduction value, carbon sink amount and regional governance value. The carbon sink estimation of oyster farming is based on dynamic fitting of the growth stage. The carbon emission value calculation includes the current value, future value, prevention value, emission reduction value and carbon emission management cost of atmospheric pollutants. The oyster farming carbon benefit method is divided into an atmospheric pollutant indicator data management system, a carbon emission and carbon sink accounting and prediction system and a carbon benefit value accounting system. The atmospheric pollutant indicator data management system includes atmospheric pollutant monitoring equipment and an atmospheric pollutant computing cloud and storage equipment, which include program codes capable of collecting, calculating and summarizing atmospheric pollutant data.

10. The carbon inclusive method for oyster aquaculture according to claim 9, characterized in that: The air pollutant monitoring equipment monitors the changing rules of air pollutants and carbon inclusive value, as well as the carbon sink amount and carbon sink benefit based on the air pollutant and carbon inclusive value accounting method; The monitoring based on the changing rules of atmospheric pollutants and carbon credit value and carbon sink and carbon credit benefits includes: a) monitoring and calculating carbon credit accounting methods; b) monitoring marine ecological and environmental changes; establishing a model of the most suitable area and suitable aquaculture area through analysis of the data obtained from monitoring and the carbon credit value accounting method, giving data-based suggestions through the model, increasing the appropriate aquaculture scale in the suitable aquaculture area to achieve the design of the most suitable oyster aquaculture area; (2) the carbon emission and carbon sink accounting and prediction system is constructed with an aquaculture carbon sink algorithm for iterative algorithm optimization, the calculation process outputs the carbon fixation amount and the overall carbon fixation amount at each stage, and the output system compares the actual aquaculture with the calculated carbon fixation amount and the overall carbon fixation amount at each stage based on the data to obtain the carbon credit value; (3) the carbon credit value accounting system collects historical data, obtains seasonal seawater movement forecast information for each stage in the future based on the predicted climate change data and pollution conditions, and calculates the carbon sink amount based on the proportion of different types of pollutants and the carbon sequestration capacity and carbon sink benefits of different types of pollutants.

Images