Shellfish carbon sequestration speed quantitative estimation method based on neural network model

Abstract

The invention discloses a shellfish carbon sequestration speed quantitative estimation method based on a neural network model, and belongs to the technical field of environmental science and ecology, and the method comprises the steps: obtaining water quality parameter data, building a water quality monitoring data set based on the water quality parameter data, and carrying out the preprocessing; performing data alignment on the shellfish growth data and the preprocessed water quality monitoring data set to obtain a time sequence data set; training a long-short term memory neural network model based on the time sequence data set, obtaining a predicted value of the pH value of the water body in a preset time period through the trained model and monitoring data obtained in real time, and calculating the release rate of carbonate ions, a carbon sequestration speed estimated value is obtained by combining the net absorption amount of carbon dioxide in the shellfish growth process in unit time; and calculating the total carbon sequestration amount of the shellfish in the whole water body based on the carbon sequestration speed estimated value. According to the method, accurate estimation of the carbon sequestration amount of the shellfish culture water body is realized through a method of combining high-precision sensing monitoring and a deep learning model.

Description

Technical Field

[0001] The invention belongs to the technical field of environmental science and ecology, and in particular relates to a method for quantitatively estimating the carbon fixation rate of shellfish based on a neural network model. Background Art

[0002] During the cultivation of shellfish in small water bodies, the pH of the water body will change as the shellfish grow. In order to quantitatively estimate the carbon fixation rate of shellfish, it is necessary to establish a method based on a neural network model to learn and predict the changing characteristics and trends of the pH of the water body. However, in practical applications, there are the following technical problems: First, the change of water pH is not only related to the growth of shellfish, but may also be affected by other factors, such as the activities of other organisms in the water body, changes in ambient temperature, etc. These factors may interfere with the model's learning and prediction of the pH change characteristics. Secondly, different types of shellfish may have different effects on the pH of the water body, and the same type of shellfish may have different effects on the pH of the water body at different growth stages, which requires the model to be able to adapt to different types of shellfish and different growth stages. Thirdly, in practical applications, it may not be possible to obtain water pH change data for a long enough time, which requires the model to be able to learn and predict based on limited data. Finally, since the relationship between the release of carbonate ions and the absorption of carbon dioxide during the growth of shellfish is relatively complex, how to accurately estimate the absorption of carbon dioxide based on the change of water pH is also a difficulty. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a quantitative estimation method for shellfish carbon fixation rate based on a neural network model to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above object, the present invention provides a method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model, comprising:

[0005] Acquire water quality parameter data, and construct a water quality monitoring data set based on the water quality parameter data;

[0006] Preprocessing the water quality monitoring data set, aligning the shellfish growth data with the preprocessed water quality monitoring data set, and obtaining a time series data set;

[0007] Training a long short-term memory neural network model based on the time series data set, and obtaining a predicted value of water pH within a preset time period based on the long short-term memory neural network model and the monitoring data acquired in real time;

[0008] The carbonate ion release rate is obtained based on the predicted water pH value within a preset time period, and the estimated carbon fixation rate is obtained based on the net absorption of carbon dioxide during the growth of shellfish per unit time;

[0009] The total carbon sequestration by shellfish in the entire water column was calculated based on the estimated carbon sequestration rate.

[0010] Optionally, the process of obtaining water quality parameter data includes:

[0011] The area and shape of the small water body to be measured are obtained, a sensor layout scheme is obtained based on the area and shape, and a small water body sensor network is obtained based on the sensor layout scheme; based on a preset sampling interval, water quality parameter data is obtained through the small water body sensor network; wherein the sensor layout scheme includes the number of sensors, layout positions and spacing; the water quality parameter data includes water body pH and environmental factor data, and the environmental factor data includes but is not limited to temperature, dissolved oxygen, conductivity, and turbidity.

[0012] Optionally, the process of constructing a water quality monitoring dataset includes:

[0013] The water quality parameter data are processed by differential coding and symmetric encryption algorithms and then transmitted to a data center; after the data center decompresses and decrypts the water quality parameter data, a median filtering algorithm is used to remove outliers to obtain an initial water quality monitoring data set; based on the measurement data of the water body area and shellfish density, the total number of shellfish in the water body is obtained, the total number of shellfish is associated with the initial water quality parameter data set, and a water quality monitoring data set containing shellfish information is constructed.

[0014] Optionally, the process of obtaining a time series dataset includes:

[0015] The normal range threshold of each indicator in the water quality monitoring data is obtained, and the outliers in the water quality monitoring data set are preliminarily eliminated according to the normal range threshold; the water quality monitoring data set is analyzed by statistical methods, and the noise data threshold is determined according to the distribution characteristics of the statistical quantity. The data exceeding the threshold is determined as noise and the corresponding data is eliminated; the water quality monitoring data set is analyzed by time series analysis method to determine the regularity and trend of the time series data, and the data that does not conform to the regularity and trend is determined as an outlier and the corresponding data is eliminated; the water quality monitoring data after removing the outliers and noise are time-aligned with the shellfish growth data to construct a time series data set; among them, if the time granularity of the water quality monitoring data and the shellfish growth data is inconsistent, interpolation or downsampling is used for data alignment.

[0016] Optionally, the time series data set is processed by a minimum-maximum normalization method, and after all eigenvalues ​​are mapped to the range of 0-1, they are divided into a training set and a test set in a ratio of 8:2; a long short-term memory neural network model is trained based on the training set and the test set; wherein the long short-term memory neural network model includes an input layer, a hidden layer and an output layer, and the hidden layer includes 64 LSTM units.

[0017] Optionally, the process of obtaining an estimate of the carbon sequestration rate includes:

[0018] Based on the carbonic acid balance equation and the predicted value of water pH within a preset time period, the release rate of carbonate ions is calculated; the dissolution equilibrium equation of carbon dioxide in water is obtained, and combined with the respiration intensity and shell growth rate data of shellfish, the net absorption of carbon dioxide during the growth of shellfish per unit time is calculated; based on the carbonate ion release rate and the net absorption of carbon dioxide, an estimated carbon fixation rate is obtained.

[0019] Optionally, a procedure to obtain an estimate of the carbon sequestration rate based on the carbonate ion release rate and the net carbon dioxide uptake includes:

[0020] The carbonate ion release rate and the net carbon dioxide absorption were fitted using the least squares method; a carbon cycle balance model was obtained, and the fitted carbonate ion release rate and net carbon dioxide absorption functions were substituted into the carbon cycle balance model to obtain a carbon balance equation, and the carbon balance equation was solved to obtain an estimated carbon fixation rate; through sensitivity analysis, the impact of changes in carbonate ion release rate and net carbon dioxide absorption on the estimated carbon fixation rate was analyzed to determine the uncertainty range of the estimated carbon fixation rate.

[0021] Optionally, water body area measurements and shellfish density records are obtained, and an estimate of the total carbon fixation by shellfish in the entire water body is calculated based on the water body area measurements, shellfish density records and carbon fixation rate estimates.

[0022] Optionally, it also includes: determining whether the estimated total carbon fixation of shellfish exceeds a preset threshold, and marking it if it exceeds the threshold; performing statistical analysis on the estimated total carbon fixation of shellfish calculated each time to obtain the average carbon fixation, maximum carbon fixation and minimum carbon fixation within a time period; comparing the results of the statistical analysis with historical data to determine whether the current total carbon fixation is within a normal range; and issuing an early warning if it exceeds the normal range.

[0023] Compared with the prior art, the present invention has the following advantages and technical effects:

[0024] The present invention continuously monitors the changes in water quality parameters by arranging a high-density sensor network in a small water body, and constructs a time series data set in combination with shellfish density data. The long short-term memory neural network model is used to learn the relationship between the pH change of the water body and the growth of shellfish and environmental factors, and predict the future trend of the pH change of the water body. According to the prediction results, the carbon fixation rate of shellfish is calculated in combination with the carbonic acid balance equation, the carbon dioxide dissolution balance equation, and the respiration and shell growth rate of shellfish. Finally, based on the water body area and shellfish density, the total carbon fixation of shellfish in the entire water body is estimated. The present invention realizes the accurate estimation of the carbon fixation of shellfish aquaculture water by combining high-precision sensor monitoring and deep learning models, which provides reliable data support for evaluating the impact of shellfish aquaculture on the carbon cycle of water bodies, and is of great significance for studying the carbon sink effect of aquaculture. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0026] Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0028] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] Embodiment 1

[0030] like Figure 1 As shown, this embodiment provides a method for quantitatively estimating the carbon fixation rate of shellfish based on a neural network model, comprising:

[0031] Acquire water quality parameter data, and construct a water quality monitoring data set based on the water quality parameter data;

[0032] Furthermore, the process of obtaining water quality parameter data includes:

[0033] The area and shape of the small water body to be measured are obtained, a sensor layout scheme is obtained based on the area and shape, and a small water body sensor network is obtained based on the sensor layout scheme; based on a preset sampling interval, water quality parameter data is obtained through the small water body sensor network; wherein the sensor layout scheme includes the number of sensors, layout positions and spacing; the water quality parameter data includes water body pH and environmental factor data, and the environmental factor data includes but is not limited to temperature, dissolved oxygen, conductivity, and turbidity.

[0034] Furthermore, the process of constructing a water quality monitoring dataset includes:

[0035] The water quality parameter data are processed by differential coding and symmetric encryption algorithms and then transmitted to a data center; after the data center decompresses and decrypts the water quality parameter data, a median filtering algorithm is used to remove outliers to obtain an initial water quality monitoring data set; based on the measurement data of the water body area and shellfish density, the total number of shellfish in the water body is obtained, the total number of shellfish is associated with the initial water quality parameter data set, and a water quality monitoring data set containing shellfish information is constructed.

[0036] Specifically, according to the area and shape of the small water body, the layout of the sensor, including the number, location and spacing of the sensors, is determined to ensure that the sensor network can cover the entire water body and the distance between the sensors meets the requirements of high-density deployment. The sensor collects water quality parameter data, including pH, temperature, dissolved oxygen, conductivity, turbidity, etc., and continuously collects data according to the preset high-frequency sampling interval to obtain time series data reflecting water quality changes. The sensor transmits the collected water quality parameter data to the data center in real time through the wireless communication module. Data compression and encryption algorithms are used during data transmission to reduce transmission delay and ensure data security. The data center receives the water quality parameter data transmitted by the sensor, decompresses and decrypts the data, and performs data cleaning and preprocessing to remove noise and outliers to obtain a high-quality water quality monitoring data set. According to the measurement data of the water body area and shellfish density, the total number of shellfish in the water body is determined, and the total number of shellfish is associated with the water quality parameter data to construct a water quality monitoring data set containing shellfish information.

[0037] Exemplarily, the design of a water quality monitoring system for a small water body needs to consider multiple aspects. The first is the sensor layout scheme, which can be arranged in a grid layout or a concentric circle layout according to the shape of the water body. For example, for a circular pond with an area of ​​1 hectare, 15 sensors can be arranged in three concentric circles, with 3 sensors arranged at an inner circle radius of 10 meters, 5 sensors arranged at a middle circle radius of 20 meters, and 7 sensors arranged at an outer circle radius of 30 meters. In this way, full coverage of the entire water body can be achieved, while the sensor spacing is kept between 10-20 meters, meeting the requirements of high-density deployment. The water quality parameters collected by the sensor include pH value, water temperature, dissolved oxygen, conductivity, and turbidity. The sampling frequency can be set to once every 5 minutes to capture short-term fluctuations in water quality. For example, the pH value should be kept between 6.5-8.5 under normal circumstances. If the pH value drops sharply from 7.2 to 6.0 in a short period of time, it may indicate a water acidification problem. Data transmission uses wireless communication technology, such as LoRa or NB-IoT, which can achieve long-distance and low-power transmission. Data compression can use algorithms such as differential coding to compress the original data to about 50% of the original. Encryption can use AES-128 bit encryption to ensure data transmission security. After the data center receives the data, it first decompresses and decrypts it. Then, outliers are removed by methods such as median filtering, such as removing values ​​that deviate too much from three consecutive data points. For example, if the dissolved oxygen concentration suddenly jumps from 8mg / L to 20mg / L and then returns to 7.5mg / L, the 20mg / L data point is likely to be an outlier.

[0038] Preprocessing the water quality monitoring data set, aligning the shellfish growth data with the preprocessed water quality monitoring data set, and obtaining a time series data set;

[0039] Furthermore, the process of obtaining a time series dataset includes:

[0040] The normal range threshold of each indicator in the water quality monitoring data is obtained, and the outliers in the water quality monitoring data set are preliminarily eliminated according to the normal range threshold; the water quality monitoring data set is analyzed by statistical methods, and the noise data threshold is determined according to the distribution characteristics of the statistical quantity. The data exceeding the threshold is determined as noise and the corresponding data is eliminated; the water quality monitoring data set is analyzed by time series analysis method to determine the regularity and trend of the time series data, and the data that does not conform to the regularity and trend is determined as an outlier and the corresponding data is eliminated; the water quality monitoring data after removing the outliers and noise are time-aligned with the shellfish growth data to construct a time series data set; among them, if the time granularity of the water quality monitoring data and the shellfish growth data is inconsistent, interpolation or downsampling is used for data alignment.

[0041] Specifically, the original water quality monitoring data and shellfish growth data are obtained, and the normal range thresholds of each monitoring indicator are set for the water quality monitoring data. The monitoring data that exceeds the threshold range is determined as an outlier. Statistical methods are used to analyze the water quality monitoring time series data, including calculating statistical quantities such as mean, variance, and standard deviation. The noise data threshold is determined according to the distribution characteristics of the statistics, and the data that exceeds the threshold is determined as noise. The water quality monitoring data is analyzed by time series analysis methods, including stationarity tests, autocorrelation analysis, etc., to determine the regularity and trend of the time series data, and the data that does not conform to the regularity and trend is determined as an outlier. The water quality monitoring data after removing the outliers and noise are time-aligned with the shellfish growth data to construct a time series data set. If the time granularity of the water quality monitoring data and the shellfish growth data is inconsistent, interpolation or downsampling methods are used for data alignment.

[0042] Exemplary, specifically, the processing of water quality monitoring and shellfish growth data is a complex task that requires multiple steps and techniques to complete. First, it is necessary to set the normal range thresholds for each monitoring indicator. For example, for pH, 6.5-8.5 may be set as the normal range; for dissolved oxygen, 5-9 mg / L may be set as the normal range. Data outside these ranges will be marked as outliers. Next, statistical methods are used to analyze the water quality monitoring time series data. For example, the calculated mean of pH may be 7.2 with a standard deviation of 0.3. The mean ± 3 times the standard deviation can be set as the noise data threshold, that is, 6.3-8.1. Data falling outside this range may be regarded as noise. Time series analysis methods help understand the regularity and trend of the data. For example, through autocorrelation analysis, it may be found that the dissolved oxygen concentration shows a periodic change every 24 hours, which is related to the diurnal temperature change. Data that does not conform to this periodicity may be judged as an outlier. Data alignment is a key step to ensure that water quality monitoring data is consistent with shellfish growth data in time. Assuming that water quality data are collected hourly and shellfish growth data are measured daily, it may be necessary to average the water quality data over a 24-hour period to align with the shellfish growth data.

[0043] Training a long short-term memory neural network model based on the time series data set, and obtaining a predicted value of water pH within a preset time period based on the long short-term memory neural network model and the monitoring data acquired in real time;

[0044] Furthermore, the time series data set is processed by the minimum-maximum normalization method, and after all eigenvalues ​​are mapped to the range of 0-1, they are divided into a training set and a test set in a ratio of 8:2; a long short-term memory neural network model is trained based on the training set and the test set; wherein the long short-term memory neural network model includes an input layer, a hidden layer and an output layer, and the hidden layer includes 64 LSTM units.

[0045] Specifically, a time series data set is constructed based on historical data of water pH, shellfish growth and environmental factors as training data for the long short-term memory neural network model. The long short-term memory neural network model is used to construct a deep learning model suitable for time series data by setting the input layer, hidden layer and output layer. The constructed time series data set is divided into a training set and a test set, and the data is normalized to improve the training effect and generalization ability of the model. By setting appropriate hyperparameters, such as the number of hidden layers, the number of hidden units, and the learning rate, the long short-term memory neural network model is trained to learn the relationship between changes in water pH and shellfish growth and environmental factors. During the training process, the weights and biases of the model are continuously adjusted through the back propagation algorithm and the gradient descent optimization method to minimize the prediction error and improve the fitting ability of the model. The trained long short-term memory neural network model is used to predict the test set data and evaluate the performance of the model. The prediction accuracy and generalization ability of the model are judged by calculating indicators such as the mean square error and the mean absolute error. Based on the prediction results of the model, the impact of changes in water pH on shellfish growth is analyzed, and the relationship between environmental factors and water pH and shellfish growth is explored to provide scientific decision-making support and management basis for the aquaculture industry.

[0046] For example, taking a shellfish farming area in a bay as an example, water pH, shellfish growth and environmental factor data for the past five years are collected. Water pH data include daily pH values, shellfish growth data include weekly average shellfish weight, and environmental factor data include daily water temperature, dissolved oxygen, etc. These data are sorted in chronological order to form a time series data set containing multiple features. The long short-term memory neural network model is a special recurrent neural network suitable for processing and predicting time series data. The model structure includes an input layer, a hidden layer, and an output layer. The input layer receives time series data, the hidden layer contains LSTM units, which can capture long-term dependencies, and the output layer produces prediction results. For example, the input layer can receive water pH and environmental factor data for the past 30 days, the hidden layer contains 64 LSTM units, and the output layer predicts the growth of shellfish in the next 7 days. Data set partitioning and normalization are key steps to improve the model training effect. The constructed time series data set is divided into a training set and a test set in a ratio of 8:2. For normalization, the minimum-maximum normalization method can be used to map all feature values ​​to the range of 0-1. For example, normalizing the pH value from the range of 6.5-8.5 to 0-1 helps eliminate the dimensional differences between different features and improve the training efficiency and generalization ability of the model. Hyperparameter settings have an important impact on model performance. The optimal hyperparameter combination can be found through methods such as grid search or random search. For example, try different numbers of hidden layers (1-3 layers), numbers of hidden units (32, 64, 128) and learning rates (0.001, 0.01, 0.1), and select the hyperparameter combination that performs best on the validation set. In the model training process, the back propagation algorithm and gradient descent optimization method are the core mechanisms. By calculating the error between the predicted value and the true value, the error gradient is back propagated to update the model parameters. The Adam optimizer can be selected, which can adaptively adjust the learning rate and accelerate convergence. Set an appropriate batch size (such as 32 or 64) and the number of training rounds (such as 100 rounds), and evaluate the performance of the model on the validation set after each round of training to avoid overfitting. Model evaluation is an important part of judging the prediction effect. The performance of the model on the test set was evaluated using indicators such as mean square error (MSE) and mean absolute error (MAE).

[0047] Specifically, obtain a pre-trained neural network model and its model parameters for subsequent water quality prediction and analysis. Collect various monitoring indicator data of the target water body in real time through water quality monitoring equipment as input data of the model. Preprocess the collected water quality monitoring data, including data cleaning and normalization, to meet the input format requirements of the model. Input the preprocessed water quality monitoring data into the neural network model, and start the model for predictive analysis and calculation. Extract the predicted value of the pH of the water body in the future from the output results of the model. According to the changes in the predicted values, judge the future trend of the pH of the water body, such as an upward trend, a downward trend or a fluctuating trend. Visualize the prediction results and change trends of the pH of the water body so that relevant personnel can analyze and make decisions.

[0048] Exemplarily, based on the raw data obtained from water quality monitoring, data cleaning, normalization and other methods are used for preprocessing to obtain monitoring data that conforms to the input format of the neural network model. By loading the pre-trained neural network model, the pre-processed water quality monitoring data is input into the model, and the model is started for predictive analysis and calculation. During the calculation process of the model, historical water quality data and current input data are used to continuously optimize model parameters and improve prediction accuracy through forward propagation and backward propagation of the neural network. If the prediction accuracy of the model reaches the preset threshold, the predicted value of water pH in the model output result is extracted; otherwise, the model continues to be optimized.

[0049] The carbonate ion release rate is obtained based on the predicted water pH value within a preset time period, and the estimated carbon fixation rate is obtained based on the net absorption of carbon dioxide during the growth of shellfish per unit time;

[0050] Further, the process of obtaining an estimate of the carbon sequestration rate includes:

[0051] Based on the carbonic acid balance equation and the predicted value of water pH within a preset time period, the release rate of carbonate ions is calculated; the dissolution equilibrium equation of carbon dioxide in water is obtained, and combined with the respiration intensity and shell growth rate data of shellfish, the net absorption of carbon dioxide during the growth of shellfish per unit time is calculated; based on the carbonate ion release rate and the net absorption of carbon dioxide, an estimated carbon fixation rate is obtained.

[0052] Specifically, first of all, the carbonic acid balance equation is a chemical balance equation that describes the mutual conversion between carbonic acid, bicarbonate and carbonate ions in water. For example, assuming that the pH of the water body is predicted to be decreasing at a certain moment, that is, the concentration of hydrogen ions is increasing, according to the principle of carbonic acid balance, carbonate ions will combine with hydrogen ions to generate bicarbonate, and then generate carbonic acid, which is finally released in the form of carbon dioxide. This process can calculate the release rate of carbonate ions. For example, when the pH value drops from 8.2 to 8.1, the carbonate ion release rate per liter of seawater can be estimated by the balance equation to be about 0.1 mmol per hour. The respiration of shellfish releases carbon dioxide, and the shell growth process requires the absorption of calcium carbonate, which involves the absorption of carbon dioxide. By obtaining the dissolution balance equation of carbon dioxide in water, the solubility of carbon dioxide in water under specific conditions can be known. For example, in seawater at 15 degrees Celsius and a salinity of 35, the solubility of carbon dioxide is about 1.5 mmol per liter. Combining the respiration intensity of shellfish, for example, a certain oyster releases 0.05 mmol of carbon dioxide per hour through respiration, and its shell growth rate, for example, the shell weight increases by 0.1 grams per day, corresponding to the absorption of calcium carbonate of about 0.001 mmol, the net absorption of carbon dioxide during the growth of shellfish per unit time can be calculated. For example, taking into account respiration and shell growth, the net absorption of carbon dioxide by the oyster is about 0.02 mmol per day. The carbon cycle balance model is a dynamic model that comprehensively considers the release of carbonate ions and the absorption of carbon dioxide by shellfish. For example, in the above example, if the daily release of carbonate ions in each liter of seawater is 2.4 mmol, and the net absorption of carbon dioxide by oysters per liter of seawater is 0.02 mmol per day, then it can be preliminarily estimated that the carbon fixation rate of shellfish farming in a unit volume of water is 0.02 mmol per liter per day, because the amount of carbon dioxide absorbed by shellfish can offset part of the release of carbonate ions caused by the decrease in pH. Machine learning algorithms can use historical data to train prediction models. For example, a long short-term memory network (LSTM) model can be used to input water quality data (such as pH, temperature, salinity, etc.) and shellfish growth data (such as shell length, weight, etc.) from the past week, and the training model can predict the trend of water pH changes and shellfish carbon fixation rate in the next 24 hours. For example, the model predicts that the pH value will drop from 8.1 to 8.05 in the next 24 hours, and the shellfish carbon fixation rate will drop to 0.015 millimoles per liter per day. The preset threshold is the basis for judging the risk of abnormal water quality. For example, the preset threshold of the shellfish carbon fixation rate is set to 0.01 millimoles per liter per day. If the predicted value is lower than this threshold, an early warning is triggered. For example, the predicted carbon fixation rate in the above example is 0.015 millimoles per liter per day. Although it has decreased, it is still higher than the threshold and will not trigger an early warning. However, if the predicted value drops to 0.008 millimoles per liter per day, it is lower than the threshold and the early warning mechanism is triggered.

[0053] Further, the process of obtaining an estimate of the carbon fixation rate based on the carbonate ion release rate and the net carbon dioxide absorption includes:

[0054] The carbonate ion release rate and the net carbon dioxide absorption were fitted using the least squares method; a carbon cycle balance model was obtained, and the fitted carbonate ion release rate and net carbon dioxide absorption functions were substituted into the carbon cycle balance model to obtain a carbon balance equation, and the carbon balance equation was solved to obtain an estimated carbon fixation rate; through sensitivity analysis, the impact of changes in carbonate ion release rate and net carbon dioxide absorption on the estimated carbon fixation rate was analyzed to determine the uncertainty range of the estimated carbon fixation rate.

[0055] Specifically, a carbon cycle balance model is established based on the carbonate ion release rate and the net carbon dioxide absorption data. The carbonate ion release and carbon dioxide absorption processes are considered in the model, and the carbon fixation rate is solved by the balance equation. The carbonate ion release rate and the net carbon dioxide absorption data are fitted by the least squares method to obtain the fitting curve and fitting parameters as the input of the carbon cycle balance model. The fitted carbonate ion release rate and net carbon dioxide absorption function are substituted into the carbon cycle balance model, the carbon balance equation is listed, and the estimated carbon fixation rate is solved. Through sensitivity analysis, the impact of changes in carbonate ion release rate and net carbon dioxide absorption on the estimated carbon fixation rate is analyzed to determine the uncertainty range of the estimated carbon fixation rate. The Monte Carlo simulation method can also be used to calculate the probability distribution of the estimated carbon fixation rate based on the probability distribution of carbonate ion release rate and net carbon dioxide absorption, and obtain the confidence interval of the carbon fixation rate. According to the estimated carbon fixation rate and confidence interval, the carbon fixation capacity of shellfish can be judged, the importance of shellfish in the carbon cycle can be determined, and a basis can be provided for formulating carbon sink protection measures.

[0056] For example, establishing a carbon cycle balance model is a key step in estimating the carbon fixation rate of shellfish. The model takes into account the two main processes of carbonate ion release and carbon dioxide absorption. Taking oyster farming as an example, by measuring the change of carbonate ion concentration in the water body over time, the carbonate ion release rate data can be obtained. At the same time, the carbon dioxide absorption during oyster respiration and shell growth is monitored to obtain the net carbon dioxide absorption data. By fitting these data using the least squares method, the functional relationship between the carbonate ion release rate and the net carbon dioxide absorption over time can be obtained. For example, the carbonate ion release rate may show a linear growth trend, while the net carbon dioxide absorption may show an exponential growth. Substitute these fitting functions into the carbon cycle balance model, list the carbon balance equation, and solve the equation to obtain the estimated carbon fixation rate of oysters. In order to evaluate the reliability of the estimated carbon fixation rate, a sensitivity analysis is required. By changing the input values ​​of carbonate ion release rate and net carbon dioxide absorption, the changes in the estimated carbon fixation rate are observed. For example, when the carbonate ion release rate increases by 10%, the carbon fixation rate may increase by 5%, while when the net carbon dioxide absorption increases by 10%, the carbon fixation rate may increase by 8%. This indicates that the carbon fixation rate is more sensitive to changes in the net carbon dioxide absorption. Monte Carlo simulation methods can be used to further quantify the uncertainty of the carbon fixation rate estimate. Assuming that the carbonate ion release rate and the net carbon dioxide absorption follow a normal distribution, the probability distribution of the carbon fixation rate can be obtained through multiple random sampling and calculations. For example, the simulation results may show that the carbon fixation rate of oysters is estimated to be 0.5 grams of carbon per day per individual, and the 95% confidence interval is 0.45-0.55 grams of carbon per day per individual. Based on this estimate, the importance of oysters in the carbon cycle can be evaluated. If a standard farm can produce 1 million oysters per year, the annual carbon fixation can reach 182.5 tons of carbon. This data can provide an important basis for formulating carbon sink protection measures, such as expanding the scale of oyster farming or optimizing farming technology to improve carbon fixation efficiency. In order to improve the accuracy of the model, it is very important to use a cross-validation method. The collected data are randomly divided into a training set (such as 70%) and a test set (such as 30%). The carbon cycle balance model is established with the training set data, and then the accuracy of the model prediction results is verified with the test set data. By repeating this process many times, the model parameters can be continuously optimized and the reliability of carbon fixation rate estimation can be improved. For example, the average absolute error of the initial model on the test set may be 0.08 grams of carbon / day / individual. By adjusting the model structure, such as introducing environmental factors such as water temperature and salinity, the error may be reduced to 0.05 grams of carbon / day / individual. This optimization not only improves the estimation accuracy, but also reveals the key factors affecting the carbon fixation capacity of shellfish. The establishment and optimization of the carbon cycle balance model is crucial for accurately assessing the carbon fixation capacity of shellfish. By combining methods such as least squares fitting, sensitivity analysis, Monte Carlo simulation and cross-validation, reliable estimates of carbon fixation rate and their uncertainty ranges can be obtained.These results not only help us understand the role of shellfish in the carbon cycle, but also provide a scientific basis for the development of carbon sink protection strategies. At the same time, the continuous optimization process of the model can also reveal the key factors affecting carbon fixation efficiency, and point out the direction for improving aquaculture technology and enhancing carbon sink capacity.

[0057] The total carbon sequestration by shellfish in the entire water column was calculated based on the estimated carbon sequestration rate.

[0058] Furthermore, the measured values ​​of water body area and shellfish density are obtained, and the estimated value of the total carbon fixation of shellfish in the entire water body is calculated based on the measured values ​​of water body area, the recorded values ​​of shellfish density and the estimated value of carbon fixation rate.

[0059] Furthermore, it also includes: judging whether the estimated total carbon fixation of shellfish exceeds a preset threshold, and marking it if it exceeds the threshold; performing statistical analysis on the estimated total carbon fixation of shellfish calculated each time to obtain the average carbon fixation, maximum carbon fixation and minimum carbon fixation within a time period; comparing the results of the statistical analysis with historical data to judge whether the current total carbon fixation is within a normal range; and issuing an early warning if it exceeds the normal range.

[0060] Specifically, the estimated carbon fixation rate of shellfish, the measured value of water area and the recorded value of shellfish density are obtained, and the three variables are input into the preset calculation model; the model calculates the estimated total carbon fixation of shellfish in the entire water body based on the input carbon fixation rate, water area and shellfish density; it is determined whether the calculated estimated total carbon fixation value exceeds the preset threshold, and if so, the value is marked as an outlier; the calculated estimated total carbon fixation value is statistically analyzed to obtain the average carbon fixation amount, maximum carbon fixation amount and minimum carbon fixation amount within a time period; the results obtained from the statistical analysis are compared with historical data to determine whether the current total carbon fixation amount is within the normal range; based on the comparison results, a decision tree algorithm is used to determine the extent of the impact of current shellfish farming on the carbon cycle of the water body; the calculated estimated total carbon fixation amount and the determined extent of impact and other data are stored in a database to provide support for subsequent data analysis and decision-making.

[0061] For example, the estimated carbon fixation rate of shellfish is a key indicator for assessing the carbon sink function of aquaculture. For example, the average carbon fixation rate of oysters in a certain bay aquaculture area is 2.5 grams per day per oyster. The water area measurement reflects the scale of aquaculture and can be obtained through satellite remote sensing or field measurements. Assume that the area of ​​the aquaculture area in the bay is 500 hectares. The recorded value of shellfish density represents the number of shellfish per unit area and is determined by sampling survey. For example, the average density in the area is 5 per square meter. Entering these three variables into the preset calculation model, the total carbon fixation estimate of shellfish in the entire water body can be obtained. The model may use a simple linear relationship or a complex nonlinear function. In this example, the model is assumed to be: Total carbon fixation = carbon fixation rate × water area × shellfish density. The calculation shows that the daily carbon fixation in the aquaculture area of ​​the bay is about 62.5 tons. In order to judge abnormal situations, a reasonable threshold needs to be set. If the preset threshold is 80 tons / day, the calculation result does not exceed the threshold and will not be marked as an abnormal value. This threshold setting helps to detect abnormal situations in the aquaculture process in a timely manner, such as large-scale death of shellfish or stunted growth. Statistical analysis of the calculation results can be used to obtain the average, maximum and minimum carbon fixation over a period of time. Assuming continuous monitoring for 30 days, the average carbon fixation is 60 tons / day, the maximum is 70 tons / day, and the minimum is 50 tons / day. These statistical data can reflect the stability and fluctuation range of the carbon fixation capacity of the aquaculture area. By comparing the statistical results with historical data, it can be determined whether the current total carbon fixation is within the normal range. If historical data show that the average carbon fixation of the bay during the same period is 58-65 tons / day, the monitoring results of this time are within the normal range. This comparative analysis helps to discover long-term trend changes and provide a basis for aquaculture management. The decision tree algorithm is used to determine the impact of shellfish aquaculture on the carbon cycle of water bodies. The decision tree can comprehensively consider multiple factors, such as carbon fixation, water quality parameters, phytoplankton biomass, etc. By setting different decision nodes and thresholds, the impact can be divided into three levels: low, medium and high. For example, when the carbon fixation is within the normal range and the water quality is good, it is judged as low impact; when the carbon fixation is close to the upper limit and the risk of eutrophication of the water body increases, it is judged as medium impact. Finally, the total carbon sequestration estimate and impact data are stored in a database. These data can be used for long-term trend analysis, seasonal change research, and aquaculture strategy optimization. For example, by analyzing the carbon sequestration efficiency in different seasons, the aquaculture density and feeding amount can be adjusted to maximize the carbon sink benefit. At the same time, these data also provide a scientific basis for the formulation of carbon trading and ecological compensation policies, and promote the sustainable development of aquaculture.

[0062] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model, characterized in that: The following steps are involved: Acquire water quality parameter data, and construct a water quality monitoring data set based on the water quality parameter data; Preprocessing the water quality monitoring data set, aligning the shellfish growth data with the preprocessed water quality monitoring data set, and obtaining a time series data set; Training a long short-term memory neural network model based on the time series data set, and obtaining a predicted value of water pH within a preset time period based on the long short-term memory neural network model and the monitoring data acquired in real time; The carbonate ion release rate is obtained based on the predicted water pH value within a preset time period, and the estimated carbon fixation rate is obtained based on the net absorption of carbon dioxide during the growth of shellfish per unit time; The total carbon sequestration by shellfish in the entire water column was calculated based on the estimated carbon sequestration rate.

2. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 1, characterized in that: The process of obtaining water quality parameter data includes: The area and shape of the small water body to be measured are obtained, a sensor layout scheme is obtained based on the area and shape, and a small water body sensor network is obtained based on the sensor layout scheme; based on a preset sampling interval, water quality parameter data is obtained through the small water body sensor network; wherein the sensor layout scheme includes the number of sensors, layout positions and spacing; the water quality parameter data includes water body pH and environmental factor data, and the environmental factor data includes but is not limited to temperature, dissolved oxygen, conductivity, and turbidity.

3. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 1, characterized in that: The process of constructing a water quality monitoring dataset includes: The water quality parameter data are processed by differential coding and symmetric encryption algorithms and then transmitted to a data center; after the data center decompresses and decrypts the water quality parameter data, a median filtering algorithm is used to remove outliers to obtain an initial water quality monitoring data set; based on the measurement data of the water body area and shellfish density, the total number of shellfish in the water body is obtained, the total number of shellfish is associated with the initial water quality parameter data set, and a water quality monitoring data set containing shellfish information is constructed.

4. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 1, characterized in that: The process of obtaining a time series dataset includes: The normal range threshold of each indicator in the water quality monitoring data is obtained, and the outliers in the water quality monitoring data set are preliminarily eliminated according to the normal range threshold; the water quality monitoring data set is analyzed by statistical methods, and the noise data threshold is determined according to the distribution characteristics of the statistical quantity. The data exceeding the threshold is determined as noise and the corresponding data is eliminated; the water quality monitoring data set is analyzed by time series analysis method to determine the regularity and trend of the time series data, and the data that does not conform to the regularity and trend is determined as an outlier and the corresponding data is eliminated; the water quality monitoring data after removing the outliers and noise are time-aligned with the shellfish growth data to construct a time series data set; among them, if the time granularity of the water quality monitoring data and the shellfish growth data is inconsistent, interpolation or downsampling is used for data alignment.

5. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 1, characterized in that: The time series data set is processed by the minimum-maximum normalization method. After all eigenvalues ​​are mapped to the range of 0-1, they are divided into a training set and a test set in a ratio of 8:2; a long short-term memory neural network model is trained based on the training set and the test set; wherein the long short-term memory neural network model includes an input layer, a hidden layer and an output layer, and the hidden layer includes 64 LSTM units.

6. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 1, characterized in that: The process for obtaining an estimate of carbon sequestration rate involves: Based on the carbonic acid balance equation and the predicted value of water pH within a preset time period, the release rate of carbonate ions is calculated; the dissolution equilibrium equation of carbon dioxide in water is obtained, and combined with the respiration intensity and shell growth rate data of shellfish, the net absorption of carbon dioxide during the growth of shellfish per unit time is calculated; based on the carbonate ion release rate and the net absorption of carbon dioxide, an estimated carbon fixation rate is obtained.

7. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 6, characterized in that: The process of obtaining an estimate of the carbon sequestration rate based on the carbonate ion release rate and the net carbon dioxide uptake involves: The carbonate ion release rate and the net carbon dioxide absorption were fitted using the least squares method; a carbon cycle balance model was obtained, and the fitted carbonate ion release rate and net carbon dioxide absorption functions were substituted into the carbon cycle balance model to obtain a carbon balance equation, and the carbon balance equation was solved to obtain an estimated carbon fixation rate; through sensitivity analysis, the impact of changes in carbonate ion release rate and net carbon dioxide absorption on the estimated carbon fixation rate was analyzed to determine the uncertainty range of the estimated carbon fixation rate.

8. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 1, characterized in that: Water body area measurements and shellfish density records are obtained and an estimate of the total carbon sequestration by shellfish in the entire water body is calculated based on the water body area measurements and shellfish density records and the estimated carbon sequestration rate.

9. The method for quantitatively estimating carbon fixation rate of shellfish based on a neural network model according to claim 1, characterized in that: Also includes: Determine whether the estimated total carbon sequestration of shellfish exceeds a preset threshold, and if so, mark the shellfish; Perform statistical analysis on the estimated total carbon fixation of shellfish calculated each time to obtain the average, maximum and minimum carbon fixation within a time period; compare the results of the statistical analysis with historical data to determine whether the current total carbon fixation is within the normal range; If it exceeds the normal range, an early warning will be issued.

Images