A method for determining the carbon emission threshold of the whole process of rural human and animal waste utilization

Abstract

The present invention discloses a method for determining the carbon emission threshold of the whole process of human and livestock manure utilization in rural areas, comprising: obtaining carbon emission influencing factors of human, livestock and poultry manure, wherein the carbon emission influencing factors include manure property data, collection method data, storage condition data, treatment process data and field return utilization data; analyzing and processing the carbon emission influencing factors respectively to obtain carbon emission threshold data corresponding to different carbon emission influencing factors; fusing data of different carbon emission influencing factors through an attention mechanism to obtain a final comprehensive carbon emission threshold.

Description

Technical Field

[0001] The present invention belongs to the field of carbon emission information technology, and in particular relates to a method for determining a carbon emission threshold value for the entire process of utilization of human and animal manure in rural areas. Background Art

[0002] In the process of rural human and livestock manure treatment, the determination of carbon emission thresholds faces many technical difficulties. First, the physical and chemical properties of different human and livestock manure vary greatly. Parameters such as organic matter content, moisture, pH value, etc. will affect the subsequent collection, storage, treatment, and utilization, and thus affect carbon emissions. Secondly, different collection methods and frequencies will also cause changes in manure properties, such as deepening of anaerobic fermentation, thereby changing carbon emission fluxes. Furthermore, storage conditions such as temperature, humidity, aerobic / anaerobic environment, etc. will affect the decomposition and transformation of organic matter in manure, resulting in different carbon emission levels. In addition, the choice of manure treatment process routes, such as composting, biogas fermentation, aerobic fermentation, etc., will also directly affect carbon emissions. Finally, when manure is returned to the field for use, factors such as application amount, application time, soil type, etc. will also affect the soil carbon sink function, and thus affect the carbon emission level of the entire system. Therefore, the determination of the carbon emission threshold for the entire process of human and livestock manure utilization in rural areas requires comprehensive consideration of factors in various links such as manure properties, collection methods, storage conditions, treatment processes, and return to field utilization, selection of appropriate carbon emission accounting methods, and a large amount of field monitoring and data accumulation. Only in this way can a scientific and complete threshold determination system be established, providing important theoretical basis and technical support for the coordinated development of manure resource utilization and carbon emission reduction. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a method for determining the carbon emission threshold of the whole process of rural human and animal manure utilization to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above objectives, the present invention provides a method for determining the carbon emission threshold of the whole process of rural human and animal waste utilization, comprising:

[0005] Obtain the factors affecting carbon emissions from human, livestock and poultry manure, including manure properties data, collection method data, storage condition data, treatment process data and field utilization data;

[0006] Analyze and process the carbon emission influencing factors respectively to obtain carbon emission threshold data corresponding to different carbon emission influencing factors;

[0007] The data of different carbon emission influencing factors are fused through the attention mechanism to obtain the final comprehensive carbon emission threshold.

[0008] Optionally, the process of analyzing and processing the feces property data includes:

[0009] Obtaining manure property data, wherein the manure property data includes physical and chemical property parameters of organic matter content, water content and pH value of manure;

[0010] Preprocessing the feces and sewage properties data, wherein the preprocessing process includes missing value processing and outlier processing;

[0011] The pre-treated feces and sewage property data are predicted by a prediction model to obtain carbon emission threshold data corresponding to the feces and sewage property data, wherein the prediction model adopts a machine learning algorithm.

[0012] Optionally, the process of analyzing and processing the collected data includes:

[0013] Obtaining collection method data and setting a preset carbon emission threshold, wherein the collection method data includes a collection method and a collection frequency;

[0014] The collection method data is calculated through an anaerobic fermentation degree assessment model to obtain a quantitative result of the anaerobic fermentation degree, a threshold judgment is performed on the quantitative result of the anaerobic fermentation degree, and a preset carbon emission threshold is dynamically corrected according to the threshold judgment result to obtain a correction value, and the correction value is fused with the preset carbon emission threshold to obtain the corrected carbon emission threshold as the carbon emission threshold data corresponding to the collection method data. The anaerobic fermentation degree assessment model adopts a BP neural network model.

[0015] Optionally, the process of analyzing and processing the storage condition data includes:

[0016] Acquiring storage condition data, wherein the storage condition data includes temperature and humidity data of the manure storage environment;

[0017] The storage condition data are processed by a deep learning model to obtain decomposition conversion rate data, and the decomposition conversion rate data are calculated by a carbon balance equation to obtain carbon emission threshold data corresponding to the storage condition data.

[0018] Optionally, the process of analyzing and processing the processing data includes:

[0019] Acquire treatment process data, wherein the treatment process data includes carbon emission detection data under different treatment processes, wherein the treatment processes include composting process, biogas fermentation process and aerobic fermentation process;

[0020] The processing data are clustered by clustering algorithm to obtain clustering results of different processing processes;

[0021] The mean and standard deviation of carbon emissions are calculated for the clustering results of different treatment processes respectively, and the carbon emission threshold data corresponding to the treatment process data are obtained based on the calculation results of the mean and standard deviation.

[0022] Optionally, the process of analyzing and processing the land use data includes:

[0023] Obtaining data on returning manure to the field, including the application amount, application time and soil type data of returning manure to the field;

[0024] The data of returning farmland to land is predicted through machine learning algorithms to obtain the predicted value of carbon sink potential;

[0025] The negative value of the predicted carbon sink potential is used as the carbon emission threshold data for farmland return utilization data.

[0026] Optionally, different weight coefficients are determined for different carbon emission influencing factors through an attention mechanism, and weighted sum calculations are performed according to the weight coefficients to obtain corresponding weight coefficients.

[0027] Optionally, a weighted average method is used to fuse the carbon emission threshold data.

[0028] Optionally, after obtaining the final comprehensive carbon emission threshold, the following is also included:

[0029] Set up IoT sensor nodes, calculate real-time carbon emissions based on the IoT sensor nodes, make judgments based on the carbon emissions calculation results and the comprehensive carbon emissions threshold, and obtain early warning results.

[0030] Optionally, real-time carbon emissions calculation using deep learning models.

[0031] Compared with the prior art, the present invention has the following advantages and technical effects:

[0032] The present invention discloses a precise control method for carbon emissions from rural human and livestock manure and sewage systems. The present invention establishes a prediction model between manure and sewage properties and carbon emissions through a machine learning algorithm, and judges the degree of anaerobic fermentation in combination with manure and sewage collection methods and frequency data to correct the carbon emission threshold; collects manure and sewage storage environment data, analyzes the decomposition and conversion rate of organic matter under different storage conditions, and establishes a correlation model between storage conditions and carbon emissions; for different manure and sewage treatment process routes, a clustering algorithm is used to determine the carbon emission threshold intervals of various treatment processes; obtains manure and sewage return to the field utilization data, establishes a correlation model between application parameters and soil carbon sequestration capacity, and dynamically adjusts the carbon emission threshold of the entire system. The present invention comprehensively considers the carbon emission influencing factors of various links such as manure and sewage properties, collection methods, storage conditions, treatment processes, and return to the field utilization, constructs a carbon emission assessment framework for rural human and livestock manure and sewage systems, objectively determines the weight coefficients of each factor, and calculates the comprehensive carbon emission threshold of the entire system. In actual applications, data from each link are collected in real time through IoT sensors, and the established carbon emission prediction model is used to dynamically evaluate the real-time carbon emission level. If the warning threshold is exceeded, the warning mechanism is triggered, and the corresponding manure management strategy is adjusted to achieve accurate control of carbon emissions from rural human and livestock manure systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] 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:

[0034] Figure 1 This is a flow chart of a method for determining a carbon emission threshold value for the entire process of utilizing human and animal waste according to an embodiment of the present invention;

[0035] Figure 2 A schematic diagram of a method for determining a carbon emission threshold value for the entire process of utilizing human and animal waste according to an embodiment of the present invention;

[0036] Figure 3 This is another schematic diagram of the method for determining the carbon emission threshold of the entire process of utilization of human and animal waste according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] 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.

[0038] 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.

[0039] like Figure 1-3In this embodiment, a method for determining a carbon emission threshold for the entire process of rural human and animal waste utilization may specifically include:

[0040] Step S101, based on the manure property data of human, livestock and poultry manure, including physical and chemical property parameters such as organic matter content, moisture content and pH value, a machine learning algorithm is used to establish a prediction model between manure properties and carbon emissions, and by inputting manure property parameters, an estimated carbon emission threshold range is obtained.

[0041] Obtain human, livestock and poultry feces and sewage samples, determine the physical and chemical property parameters of organic matter content, moisture content and pH value of the human, livestock and poultry feces and sewage samples, and establish a feces and sewage property data set; use data preprocessing technology to process missing values ​​and outliers in the feces and sewage property data set to obtain a preprocessed feces and sewage property data set; divide the preprocessed feces and sewage property data set into a training set and a test set, the training set is used for model training, and the test set is used for model evaluation; select at least one machine learning algorithm based on the training set to construct a prediction model for feces and sewage properties and carbon emissions; use the training set to The prediction model is trained using test set data, and the performance of the prediction model is optimized by adjusting the hyperparameters of the prediction model to obtain a trained prediction model; the test set data is input into the trained prediction model to evaluate the prediction performance of the prediction model, and if the prediction performance meets the preset requirements, the prediction model is determined to be the final prediction model; human, livestock and poultry manure samples to be predicted are obtained, manure trait data of the human, livestock and poultry manure samples to be predicted are measured, and the manure trait data are input into the final prediction model to obtain the estimated carbon emission threshold of the human, livestock and poultry manure samples to be predicted.

[0042] Specifically, when obtaining human, livestock and poultry feces samples, 200 different human, livestock and poultry feces samples were collected, each weighing 500 grams. The physical and chemical properties of the human, livestock and poultry feces samples, such as organic matter content, moisture content and pH value, were determined, wherein the organic matter content was determined by potassium dichromate volumetric method, the moisture content was determined by drying method, and the pH value was determined by pH meter. The measurement results were entered into an Excel table to establish a feces property data set. For the feces property data set, data preprocessing technology was used, the maximum likelihood estimation method was used to interpolate missing values, and the box plot and Z-score method were used to identify and process outliers to obtain the cleaned feces property data set. The cleaned feces property data set was divided into a training set and a test set in a ratio of 8:2. The training set contained 160 sample data and the test set contained 40 sample data. According to the training set, the support vector machine regression (SVR) algorithm and the random forest regression (RFR) algorithm were selected to construct a prediction model for feces properties and carbon emissions. The grid search and cross-validation methods were used to optimize the hyperparameters of the SVR and RFR models, such as the kernel function type, penalty coefficient C and kernel function parameter γ of SVR, the number of decision trees, the maximum number of features and the maximum tree depth of RFR, etc., and the prediction model with the best performance was obtained through multiple rounds of iterative training. The test set data was input into the trained prediction model to evaluate the prediction performance indicators such as the root mean square error (RMSE), mean absolute error (MAE) and determination coefficient (R²). The results showed that the RMSE of the SVR model was 85, MAE was 63, and R² was 92, and the RMSE of the RFR model was 92, MAE was 71, and R² was 90, all of which met the preset requirements. Therefore, the SVR model was determined as the final prediction model. The human and livestock manure samples to be predicted were obtained, and the organic matter content was 45%, the water content was 73%, and the pH value was 2. These physical and chemical property parameters were input into the final SVR prediction model, and the estimated carbon emission threshold range of the manure sample was 350~580kg carbon dioxide equivalent / ton of manure.

[0043] Step S102, obtaining the manure collection method and collection frequency data, determining the degree of anaerobic fermentation of manure, and if the fermentation degree exceeds a preset threshold, correcting the carbon emission threshold according to the fermentation degree to obtain a more accurate threshold range.

[0044] Obtain data related to the manure collection method and collection frequency, input the data into a pre-constructed manure anaerobic fermentation degree assessment model, and obtain a quantitative result of the manure anaerobic fermentation degree; compare the quantitative result with a preset threshold value, and if the fermentation degree exceeds the preset threshold value, trigger the carbon emission threshold correction process; according to the difference between the manure anaerobic fermentation degree and the preset threshold value, use a piecewise function or a classifier algorithm to dynamically correct the preset carbon emission threshold value to obtain a carbon emission threshold correction value that matches the actual fermentation degree; fuse the carbon emission threshold correction value with the original carbon emission threshold value, and determine the corrected carbon emission threshold value by weighted average or nonlinear combination;

[0045] The post-analysis related content also includes: based on the revised carbon emission threshold value, using the confidence interval estimation method to generate a carbon emission estimation interval centered on the revised threshold as an accurate carbon emission threshold range; inputting the manure collection method, collection frequency, anaerobic fermentation degree data and the revised carbon emission threshold range into the carbon emission accounting model to obtain the estimated carbon emission value under the manure treatment condition; comparing the carbon emission estimation value with the revised carbon emission threshold range to determine whether the manure treatment process meets the carbon emission compliance requirements. If not, triggering an early warning and adjusting the manure treatment process parameters until the carbon emissions fall within the threshold range.

[0046] Specifically, through the survey of 100 farms, it was found that the manure collection methods are mainly divided into two types: manual manure cleaning and automatic manure scraping, and the collection frequency is divided into three situations: once a day, twice a day, and three times a day. The collection method and frequency data were input into the pre-built anaerobic fermentation degree assessment model of manure based on BP neural network, and the anaerobic fermentation degree of manure of each farm was quantitatively scored. The results showed that the anaerobic fermentation degree of manure in 30 farms exceeded the preset threshold of 6, and the carbon emission threshold correction process needed to be triggered. For these 30 farms, a piecewise function algorithm was used to set the carbon emission threshold correction coefficient to 2, 5 and 8 according to the difference in the fermentation degree exceeding the threshold, corresponding to the situation where the fermentation degree exceeded the threshold of 1, 2 and 3, and the carbon emission threshold was dynamically corrected accordingly. Then, the weighted average method was used to merge the correction coefficient with the original threshold to obtain the corrected carbon emission threshold value.

[0047] Then, the confidence interval estimation method was used, with the corrected threshold as the mean and the confidence level set to 95%, to calculate the upper and lower limits of the carbon emission estimation interval. Finally, the manure collection method, collection frequency, anaerobic fermentation degree data, and the corrected carbon emission threshold interval were input into the carbon emission accounting model. The estimated carbon emissions of the manure treatment process of each farm were calculated through material balance and emission factor method, and compared with the carbon emission threshold interval to determine whether the carbon emission compliance requirements were met. The results showed that the estimated carbon emissions of 5 farms exceeded the upper limit of the threshold interval, triggering an early warning. By optimizing the manure collection method and frequency, reducing the exposure time of manure, and improving the efficiency of manure treatment, the manure treatment process parameters were adjusted to make the carbon emissions fall back into the threshold range and meet the compliance requirements.

[0048] Step S103, collect data such as temperature and humidity of the manure storage environment, analyze the decomposition and conversion rate of organic matter under different storage conditions through machine learning algorithms, establish a correlation model between storage conditions and carbon emissions, and determine the corresponding carbon emission level according to actual storage conditions.

[0049] Acquire temperature and humidity data of the manure storage environment, and use the data as input features of the model; acquire decomposition and conversion rate data of organic matter under different storage conditions, train the rate data with a support vector machine algorithm, and obtain a rate prediction model; calculate carbon emissions through a carbon balance equation based on the decomposition and conversion rate of organic matter obtained by the rate prediction model; construct a correlation model between storage condition characteristics and the carbon emissions, and train the correlation model with a decision tree algorithm; acquire temperature and humidity data of the actual storage environment, input the temperature and humidity data into the correlation model, and obtain the carbon emission level corresponding to the temperature and humidity data; determine whether the carbon emission level exceeds a preset threshold, and if so, trigger an early warning to indicate the need to adjust the storage conditions; based on the early warning result, control the temperature and humidity adjustment equipment of the storage environment to adjust the temperature and humidity of the storage environment to a reasonable range determined by the correlation model.

[0050] Specifically, by laying out temperature and humidity sensors in the farm, the temperature and humidity data of the manure storage environment are collected in real time, and the collected data are uploaded to the cloud server through the wireless communication module. Under laboratory conditions, different storage temperatures (15℃, 25℃, 35℃) and humidity (40%, 60%, 80%) were simulated to determine the decomposition and conversion rate of organic matter in manure. Each condition was repeated 3 times, and a total of 27 sets of rate data were obtained. The temperature, humidity and corresponding organic matter decomposition rate data were randomly divided into training sets and test sets, and the support vector machine algorithm was used for training. The model parameters were optimized by grid search and cross-validation to obtain the rate prediction model, and the average absolute error on the test set was 28. According to the organic matter decomposition and conversion rate obtained by the rate prediction model, the carbon emissions were calculated by the carbon balance equation. The calculation formula is: carbon emissions = organic matter decomposition × carbon conversion coefficient, where the carbon conversion coefficient is 58. The calculated carbon emission data and the corresponding temperature and humidity data are used to construct an association model, and the decision tree algorithm is used for training. Through the visualization of the decision tree structure, it is found that temperature and humidity are the main factors affecting carbon emissions, and there is an interactive effect. By inputting the temperature and humidity data of the actual storage environment into the trained association model, the carbon emission level under this condition can be predicted. The carbon emission threshold is set at 0.5kg / (m3·d), and the predicted carbon emission level is compared with the threshold. If the threshold is exceeded, an early warning is triggered, and according to the branching rules of the decision tree model, it is determined that the temperature needs to be adjusted to 20~30℃ and the humidity needs to be adjusted to a reasonable range of 50%~70%. The temperature and humidity adjustment equipment is connected through an automatic control system, and the temperature and humidity parameters of the storage environment are adjusted in real time according to the early warning results to maintain it in the optimal range determined by the model, thereby effectively controlling the carbon emission level of the manure storage process and achieving emission reduction and efficiency improvement.

[0051] Step S104, for different manure treatment process routes, such as composting, biogas fermentation, aerobic fermentation, etc., obtain the carbon emission monitoring data of each process, use a clustering algorithm to classify, and determine the carbon emission threshold range of each treatment process.

[0052] Obtain carbon emission monitoring data for different manure treatment process routes, including composting, biogas fermentation and aerobic fermentation;

[0053] Preprocessing the carbon emission monitoring data to remove abnormal values ​​and noise data therein to obtain preprocessed carbon emission monitoring data;

[0054] Extracting characteristic indicators of carbon emission amount, emission time, and emission frequency according to the pre-processed carbon emission monitoring data, and constructing a carbon emission characteristic vector;

[0055] The K-means clustering algorithm is used to perform cluster analysis on the carbon emission feature vector to obtain clustering results of different manure treatment processes;

[0056] For each type of manure treatment process, the mean and standard deviation of its carbon emissions are calculated to determine the carbon emission threshold range for this type of process.

[0057] The follow-up also includes relevant process evaluation content: obtaining carbon emission monitoring data of the new manure and sewage treatment process and extracting its carbon emission characteristic vector; comparing the carbon emission characteristic vector of the new manure and sewage treatment process with the carbon emission threshold interval of the existing process category to determine the process category to which it belongs; based on the degree of deviation between the carbon emission characteristics of the new manure and sewage treatment process and the carbon emission threshold interval of the process category to which it belongs, the carbon emission performance of the new process is evaluated to obtain an evaluation result.

[0058] Specifically, infrared CO2 sensors and methane sensors were installed at the discharge ports of different manure treatment processes in the farm to collect carbon emission concentration data of composting process, biogas fermentation process and aerobic fermentation process in real time, and the data were recorded every 5 minutes for continuous monitoring for 1 month. The collected data were cleaned by a data cleaning algorithm to remove outliers and random fluctuations that exceeded the measurement range, and the pre-processed carbon emission time series data were obtained. Based on the pre-processed data, 12 characteristic indicators such as the total daily carbon emission, emission peak time and peak duration of each process were extracted to construct a 60-dimensional carbon emission feature vector. The K-means clustering algorithm was used to cluster the feature vectors, and the optimal cluster number k=3 was determined by the elbow rule, corresponding to the three manure treatment processes. The mean and standard deviation of carbon emissions of each type of manure treatment process were calculated, and the carbon emission threshold interval of the composting process was [2,8]kg / (m3·d), the threshold interval of the biogas fermentation process was [5,3]kg / (m3·d), and the threshold interval of the aerobic fermentation process was [5,1]kg / (m3·d).

[0059] The new manure treatment process was monitored continuously for 7 days, its carbon emission feature vector was extracted, and its Euclidean distance with the three cluster centers was calculated. The category with the smallest distance was the category to which the new process belonged, and the judgment result was biogas fermentation process. Further calculation of the new process carbon emission average was 4.5kg / (m3·d), which was within the threshold range of biogas fermentation process, with a deviation of 14%. The evaluation result showed that the new process met the carbon emission performance requirements of biogas fermentation process.

[0060] Step S105, obtain data such as the application amount, application time, and soil type of manure returned to the field, establish a correlation model between application parameters and soil carbon sequestration capacity through a machine learning algorithm, judge the soil carbon sequestration potential based on the actual application situation, add the soil carbon sequestration potential as a carbon emission threshold factor to the overall system, and dynamically adjust the carbon emission threshold of the entire system.

[0061] Obtaining application parameter data of the amount, application time and soil type of manure return to the field, and using them as input features of the machine learning algorithm; using support vector machine, random forest and neural network machine learning algorithms to establish a nonlinear association model between the application parameters and soil carbon sequestration capacity; predicting the carbon sequestration potential of the soil under different application parameters based on the established association model, and obtaining a predicted value of the carbon sequestration potential;

[0062] The application parameter data of the actual process of returning manure to the field is obtained, and the data is input into the association model to determine the soil carbon sink potential under the current application conditions; the predicted soil carbon sink potential is compared with a preset carbon sink potential threshold, and if it is lower than the threshold, the carbon emission threshold of the entire system is dynamically adjusted; according to the adjusted carbon emission threshold, the application parameters of the process of returning manure to the field are optimized, and the application parameters include application amount and application time, which are used to update and improve the machine learning association model to achieve adaptive optimization of the model.

[0063] Specifically, the land use type data of 500 mu of farmland around a farm were obtained through drone aerial photography technology, including 300 mu of corn planting area, 150 mu of wheat planting area, and the rest were other crops. At the same time, soil sensors were used to obtain soil organic matter content, total nitrogen content, pH value and other physical and chemical properties, as well as the application amount and application time data of farm manure returning to the field. The acquired data were randomly divided into training sets and test sets, and three machine learning algorithms, support vector machine, random forest and BP neural network, were used to establish a nonlinear association model between application parameters and soil carbon sequestration capacity. The model hyperparameters were optimized by grid search and cross-validation methods, and the optimal parameters of the support vector machine model were Gaussian kernel function and penalty coefficient C=10, the optimal parameters of the random forest model were 100 decision trees and 8 maximum tree depth, and the optimal parameters of the BP neural network model were 32 hidden layer nodes and 01 learning rate. The optimized model was applied to the test set to evaluate the model prediction performance. The average prediction accuracy of the three models reached more than 85%. Based on the trained machine learning model, the carbon sequestration potential after returning manure to the fields in corn-growing soil, wheat-growing soil and other crop soils was predicted, and the predicted carbon sequestration potential of corn fields was 2tC / (hm2·d), wheat fields was 9tC / (hm2·d), and other crop fields was 7tC / (hm2·d). The real-time collected data was added to the original training set, and the machine learning model was updated and adaptively optimized online using the incremental learning method to further improve the prediction accuracy and generalization ability of the model.

[0064] Step S106, comprehensively considers the carbon emission influencing factors of each link such as manure properties, collection methods, storage conditions, treatment processes, and return to farmland utilization, constructs a carbon emission assessment framework for the rural human and livestock manure system, determines the weight coefficient of each factor, and calculates the comprehensive carbon emission threshold of the entire system.

[0065] Obtain data and information on the characteristics, collection methods, storage conditions, treatment processes, and field utilization of manure and sewage in each link of the rural human and livestock manure and sewage system as factors affecting carbon emissions;

[0066] The carbon emission influencing factor data is cleaned, converted and standardized by using data preprocessing technology to obtain the preprocessed carbon emission influencing factor data, and the carbon emission amount data corresponding to the preprocessed carbon emission influencing factor data is obtained through the above-mentioned different carbon emission threshold data acquisition process;

[0067] The attention mechanism is used to determine the weight coefficients of factors affecting carbon emissions. Based on the carbon emissions and weight coefficients of each link, data fusion technology is used to calculate the comprehensive carbon emission threshold of the entire system to complete the carbon emission assessment of rural human and livestock manure and sewage systems.

[0068] Specifically, drone aerial photography technology is used to obtain image data of the collection, storage, treatment, and return of human and animal manure from 500 households around a village. At the same time, IoT sensors are used to collect parameters such as moisture content, pH value, and organic matter content of manure. The acquired image data, parameter data, and other data are cleaned to remove outliers and missing values. The attention mechanism is used as a related algorithm for data fusion. By inputting the above-mentioned influencing factors into the attention mechanism, the corresponding comprehensive carbon emission threshold is output; in the attention mechanism, each input influencing factor will be converted into a query, key, and value, and these converted values ​​are completed through a linear transformation layer. The attention weight is determined by calculating the similarity between the query and all keys through the dot product in the attention mechanism. After the attention weight is determined, the weighted sum in the attention mechanism is used for calculation. The values ​​of the input influencing factors are weighted and calculated according to the attention weight to fuse the carbon emission thresholds of each link. The comprehensive carbon emission threshold of the rural human and livestock manure system is obtained to be 0.6kgCO2 / kg. As a deep learning model, the attention mechanism can be trained and optimized according to historical data, which will not be elaborated here.

[0069] Step S107, in actual application, the Internet of Things sensors are used to collect data on manure characteristics, environmental parameters, etc. in each link in real time, and transmit them to the cloud platform for big data analysis. The established carbon emission prediction model is used to dynamically evaluate the real-time carbon emission level. If it exceeds the warning threshold, the warning mechanism is triggered and the corresponding manure management strategy is adjusted to achieve precise control of carbon emissions from rural human and livestock manure systems.

[0070] Obtain data collected by IoT sensors at various stages of the rural human and animal waste system, including waste property data and environmental parameter data;

[0071] Transmitting the data to a cloud platform via a wireless communication network;

[0072] Preprocessing the data received by the cloud platform to obtain a standardized data set;

[0073] Using big data analysis technology to perform feature extraction and data mining on the standardized data set to obtain key factors affecting carbon emission levels and their weight coefficients;

[0074] Input the key factors and their weight coefficients into a pre-built carbon emission prediction model, and use a machine learning algorithm to dynamically predict the carbon emission level at the current moment;

[0075] The predicted carbon emission level is compared with a preset warning threshold. If the level exceeds the warning threshold, it is determined to be abnormal carbon emission and a warning mechanism is triggered.

[0076] According to the warning level determined by the warning mechanism, a corresponding manure and sewage management strategy is obtained from the preset control strategy;

[0077] The manure and sewage management strategy is sent to the Internet of Things control terminal, and the corresponding operating instructions are executed through the automated control equipment to dynamically adjust the rural human and livestock manure and sewage system to achieve precise control of carbon emissions.

[0078] Specifically, 100 IoT sensor nodes were installed in a village to collect data on properties such as moisture content, pH value, organic matter content, and ambient temperature and humidity of manure in real time. Each node collected data every 10 minutes and transmitted it to the cloud platform through the Zigbee wireless communication protocol. The cloud platform uses the Hadoop distributed file system to store and manage the massive data received, and uses the Spark big data processing framework to clean and standardize the raw data to obtain a standardized data set containing 5,000 samples and 20 feature variables. Then, the random forest algorithm was used to evaluate the feature importance of the standardized data set, and 10 key factors such as moisture content of manure, pH value, and ambient temperature were screened out. The logistic regression algorithm was used to establish a carbon emission prediction model, and the optimal parameters of the model were determined through 5-fold cross-validation, with an average prediction accuracy of 85%. The manure characteristics and environmental parameter data collected at the current moment are input into the prediction model, and the carbon emission level is obtained to be 0.627kgCO2 / kg, which exceeds the preset warning threshold (0.6kgCO2 / kg). The system automatically triggers a yellow warning and obtains the corresponding manure management measures from the preset control strategy library, including increasing the frequency of manure turning and adjusting the temperature of the fermentation tank, and sends them to the IoT control terminal. After receiving the control instructions, the control terminal performs the corresponding operations through the automated equipment, such as starting the manure turning device and adjusting the fermentation tank temperature control valve, etc., to achieve real-time dynamic regulation of carbon emissions in the rural human and livestock manure system, reduce greenhouse gas emissions, and improve the level of manure resource utilization.

[0079] It should be noted that the data in the above technical solution of the present invention is only an exemplary display of relevant data to facilitate the understanding of relevant personnel, and does not fall into the method that must be implemented. Relevant personnel can adjust the above data according to actual needs and the specific content of the collected data.

[0080] 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 determining the carbon emission threshold of the whole process of rural human and animal waste utilization, characterized in that: include: Obtain the factors affecting carbon emissions from human, livestock and poultry manure, including manure properties data, collection method data, storage condition data, treatment process data and field utilization data; Analyze and process the carbon emission influencing factors respectively to obtain carbon emission threshold data corresponding to different carbon emission influencing factors; The data of different carbon emission influencing factors are fused through the attention mechanism to obtain the final comprehensive carbon emission threshold; The process of analyzing and processing the feces property data includes: Obtaining manure property data, wherein the manure property data includes physical and chemical property parameters of organic matter content, water content and pH value of manure; Preprocessing the feces and sewage properties data, wherein the preprocessing process includes missing value processing and outlier processing; Predicting the feces and sewage property data after pretreatment by using a prediction model to obtain carbon emission threshold data corresponding to the feces and sewage property data, wherein the prediction model adopts a machine learning algorithm; The process of analyzing and processing the data collected includes: Obtaining collection method data and setting a preset carbon emission threshold, wherein the collection method data includes a collection method and a collection frequency; The collection method data is calculated by an anaerobic fermentation degree assessment model to obtain a quantitative result of the anaerobic fermentation degree, a threshold judgment is performed on the quantitative result of the anaerobic fermentation degree, and a preset carbon emission threshold is dynamically corrected according to the threshold judgment result to obtain a correction value, and the correction value is fused and calculated with the preset carbon emission threshold to obtain the corrected carbon emission threshold as the carbon emission threshold data corresponding to the collection method data. The anaerobic fermentation degree assessment model adopts a BP neural network model; The process of analyzing and processing the storage condition data includes: Acquiring storage condition data, wherein the storage condition data includes temperature and humidity data of the manure storage environment; The storage condition data is processed by a deep learning model to obtain decomposition conversion rate data, and the decomposition conversion rate data is calculated by a carbon balance equation to obtain carbon emission threshold data corresponding to the storage condition data; The process of analyzing and processing the processing data includes: Acquire treatment process data, wherein the treatment process data includes carbon emission detection data under different treatment processes, wherein the treatment processes include composting process, biogas fermentation process and aerobic fermentation process; The processing data are clustered by clustering algorithm to obtain clustering results of different processing processes; The mean and standard deviation of carbon emissions are calculated for the clustering results of different treatment processes respectively, and the carbon emission threshold data corresponding to the treatment process data is obtained according to the calculation results of the mean and standard deviation; The process of analyzing and processing the land use data includes: Obtaining data on returning manure to the field, including the application amount, application time and soil type data of returning manure to the field; The data of returning farmland to land is predicted through machine learning algorithms to obtain the predicted value of carbon sink potential; The negative value of the predicted carbon sink potential is used as the carbon emission threshold data for farmland return utilization data.

2. The method according to claim 1, characterized in that: Different weight coefficients are determined for different carbon emission influencing factors through the attention mechanism, and weighted and calculated according to the weight coefficients to obtain the corresponding weight coefficients.

3. The method according to claim 1, characterized in that The weighted average method is used to fuse the carbon emission threshold data.

4. The method according to claim 1, characterized in that After obtaining the final comprehensive carbon emission threshold, it also includes: Set up IoT sensor nodes, calculate real-time carbon emissions based on the IoT sensor nodes, make judgments based on the carbon emissions calculation results and the comprehensive carbon emissions threshold, and obtain early warning results.

5. The method according to claim 1, characterized in that: Real-time carbon emissions calculation using deep learning models.

Images