A cloud platform-based whole life cycle monitoring system for rice straw decomposition process

Abstract

The application discloses a kind of based on cloud platform's rice straw decomposition process whole life cycle monitoring system, it is related to the field of agricultural informatization, including: field data acquisition terminal is used to collect near-infrared spectroscopy data, image data and operation and environmental data of straw decomposition process;Layered identification and modeling unit is used to identify the initial mass ratio of straw soluble component, cellulose component, hemicellulose component and lignin component and decomposition progress, and establish layered progress bar;Fusion progress calculation unit is used to generate overall decomposition progress bar by weighting each layered progress bar based on initial mass ratio;Progress output unit is used to output layered progress bar and overall decomposition progress bar simultaneously;Pattern mining unit is used to cluster historical layered progress bar, form progress category prototype library, and output potential blocking reason.The application solves the problem that straw decomposition process in field is difficult to monitor in real time and the problem that blocking reason is difficult to identify.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of agricultural informatization, more particularly, to a rice straw decomposition process whole life cycle monitoring system based on a cloud platform. BACKGROUND

[0002] A large amount of straw is produced during rice planting, and if it is directly burned, it will cause serious air pollution and waste of resources, so straw returning to field has become an important utilization method. After straw returning to field, it needs to be gradually decomposed by natural conditions and microorganisms to release nutrients to improve soil structure. However, the biochemical composition of straw is complex, and the decomposition rates of different components are significantly different. Soluble sugars and proteins can be quickly decomposed in the early stage, cellulose and hemicellulose are gradually degraded in the middle stage, and lignin is slowly decomposed in the late stage due to its stable structure, and rice straw contains high silicon and wax, which also hinders the invasion of microbial enzymes. This hierarchical decomposition characteristic causes the overall characteristics of the straw decomposition process to be fast in the early stage, slow in the middle stage, and slow in the late stage.

[0003] In the actual environment of the field, the decomposition progress is not only affected by the component difference, but also closely related to the water layer state, irrigation and drainage operation, secondary crushing condition, inoculant, soil temperature and humidity, and oxidation-reduction potential and other external conditions. Different management measures can significantly change the decomposition rate and stage conversion time. For example, high water layer may inhibit aerobic decomposition, insufficient secondary crushing may reduce the microbial contact area, and insufficient inoculant may delay the decomposition progress of key components. Due to the interweaving of these factors, the decomposition process has high complexity.

[0004] At present, the research on straw decomposition process mostly relies on offline laboratory detection, which is difficult to reflect the real-time progress in the field in time. In actual management, researchers and farmers often cannot obtain intuitive decomposition progress information and cannot clearly identify the potential reasons for decomposition stagnation. This leads to a lack of scientific basis for management measures, which may cause insufficient decomposition, increased greenhouse gas emissions, or rising operating costs. Therefore, there is an urgent need for a method and system that can dynamically monitor the whole life cycle of the decomposition process, identify the hierarchical progress, and excavate the potential blocking reasons, to provide decision support for scientific management. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a rice straw decomposition process whole life cycle monitoring system based on a cloud platform to solve the problems mentioned in the background.

[0006] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0007] A rice straw decomposition process whole life cycle monitoring system based on a cloud platform, comprising:

[0008] Field data acquisition terminals are used to collect near-infrared spectral data, image data, operational and environmental data during the straw decomposition process;

[0009] The layered identification and modeling unit is used to identify the initial mass ratio of soluble components, cellulose components, hemicellulose components and lignin components of straw based on the near-infrared spectral data and the image data, as well as the decomposition progress of each component, and to establish a layered progress bar for the corresponding decomposition process.

[0010] A fusion progress calculation unit is used to weight the layered progress bars according to the initial proportion to generate an overall decomposition progress bar;

[0011] A progress output unit is used to simultaneously output the layered progress bar and the overall corrosion progress bar;

[0012] The pattern mining unit performs cluster analysis on the time-series data of historical hierarchical progress bars in groups, classifying multi-group hierarchical progress bars with similar shapes into a progress category to form a prototype library. When the overall completion rate of a progress category is lower than the average level by more than a preset threshold, the common features of the operation and environmental data within the corresponding progress category are extracted as potential blockage causes and output. In some embodiments, the preset threshold of being lower than the average level by more than two standard deviations can be set to be lower than the mean; or a corresponding preset threshold can be manually selected, such as 20%.

[0013] Furthermore, the pattern mining unit is also used to compare the current layered progress bar with the historical layered progress bars in the prototype library, and when the similarity exceeds a preset threshold, extract the potential blocking reasons corresponding to the historical layered progress bars and perform overlap analysis with the current operation and environment data. When the overlap reaches a preset threshold, the overlapping features are output as potential blocking reasons.

[0014] Furthermore, the pattern mining unit is also used to perform a preliminary similarity comparison between the current layered progress bar and the historical layered progress bars in the prototype library, and select multiple matching historical layered progress bars with the highest average similarity based on the comparison results, and select the progress bar with the fastest completion of the overall decay progress bar among the matching historical layered progress bars as a reference to perform corresponding operations and adjust the environmental data.

[0015] Furthermore, the operational and environmental data includes any one or more of the following: water layer status, irrigation and drainage operation records, secondary fragmentation parameters, type and dosage of microbial agents, soil temperature, soil moisture, pH value, and redox potential.

[0016] Furthermore, the hierarchical identification and modeling unit is constructed using a supervised learning-based training model. The training model takes near-infrared spectral data and image data as input, and uses the actual contents of soluble components, cellulose components, hemicellulose components and lignin components obtained through chemical analysis as supervision labels. It is trained by minimizing the loss function between the predicted content and the chemically measured content.

[0017] Furthermore, the field data acquisition terminal is also used to collect carbon dioxide flux data and methane flux data;

[0018] The system also includes an anchor calibration unit, used to calibrate the progress bar of the soluble component to a completion rate of no less than 90% and the progress bar of the cellulose component to a completion rate of 20% to 30% when an early peak is detected in the carbon dioxide flux data; and to calibrate the progress bar of the hemicellulose component to a completion rate of 70% to 80% when a late peak is detected in the methane flux data.

[0019] Furthermore, the completion rate of the overall corrosion progress bar P Determined by the following formula:

[0020] ;

[0021] in, For the first i The completion status of each layered progress bar. For the first i The initial mass percentage of each component satisfies , n =4.

[0022] Furthermore, when the pattern mining unit extracts common features of operation and environmental data within a certain progress category, it performs feature extraction on each set of operation and environmental data within the progress category to form multiple sets of feature vectors, aggregates and statistically analyzes the feature vectors, and identifies features that appear frequently in the samples of this category but appear infrequently in the non-blocking category through a feature selection algorithm. These features are then used as common features of the progress category and output as potential blocking causes.

[0023] Furthermore, the feature extraction includes calculating the mean, variance, range, and trend slope for time-series environmental data, and calculating the operation frequency, operation interval, and parameter mean for operational data.

[0024] Furthermore, the feature selection algorithm can be any of the following: a feature ranking method based on information gain, a correlation screening method based on mutual information, or a sparse feature selection method based on LASSO regression.

[0025] The advantage of this invention over existing technologies lies in its proposed cloud-based system for monitoring the entire lifecycle of rice straw decomposition. This system can collect near-infrared spectral data, image data, and operational and environmental data in the field. Through hierarchical identification and modeling, the soluble, cellulose, hemicellulose, and lignin components of the straw are modeled as independent hierarchical progress bars. These bars are then weighted according to their initial mass percentages to form an overall decomposition progress bar, which can be visually displayed on the cloud platform interface. More specifically, this invention uses a hierarchical identification and modeling unit to establish independent progress bars for different components and generates an overall decomposition progress bar based on the initial mass percentage of each component. Since the decomposition rates of different components naturally differ—soluble components decompose rapidly in the early stages, cellulose and hemicellulose decompose gradually in the middle stages, while lignin remains in a slow degradation state for a long period—the overall progress bar inevitably exhibits a non-linear characteristic of rapid increase in the early stages, slowing down in the middle stages, and gradually decreasing in the later stages. This invention, through this hierarchical modeling and weighted fusion mechanism, visually maps the true biochemical laws of straw decomposition onto the progress bar, thereby achieving a visual display of the non-linear decomposition progress.

[0026] Building upon this foundation, this invention further introduces a pattern mining unit to perform cluster analysis on historical stratified progress bars, establishing a prototype library of progress categories. When the overall decomposition progress of a certain category is significantly slower, the system can automatically extract common features from the operational and environmental data within that category and output them as potential causes of blockage. By comparing the similarity between the current progress bar and the historical prototype library, the system can identify blockage patterns that highly overlap with the current scenario, thereby revealing possible causes of inefficient decomposition. This mechanism avoids relying solely on gas flux or human experience for judgment, improving the scientific rigor and reliability of problem localization. Furthermore, this invention quantifies operational and environmental data through feature extraction and selection methods, ensuring that the identification of potential blockage causes has a data-driven basis. This not only distinguishes between normal and blockage scenarios but also identifies key influencing factors. With the assistance of an anchor point calibration unit, the system uses the flux peaks of carbon dioxide and methane to correct different stratified progress bars, further improving monitoring accuracy. The overall solution realizes a complete chain from stratified modeling, progress fusion, blockage identification to cause mining, providing strong support for the scientific management of straw decomposition. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the overall system structure of the present invention;

[0028] Figure 2 This is a schematic diagram of the structure of the layered identification and modeling unit in this invention;

[0029] Figure 3 This is a schematic diagram of the pattern mining unit in this invention;

[0030] Figure 4 This is a schematic diagram of the anchor point calibration unit in this invention;

[0031] Figure 5 This is a schematic diagram of the progress bar of the present invention. Detailed Implementation

[0032] The specific embodiments of the present invention will now be described with reference to the accompanying drawings.

[0033] Figure 1 shows an overall schematic diagram of the system of the present invention. The system of the present invention aims to realize the full life cycle monitoring of the rice straw decomposition process through the cloud platform, covering multiple links such as data acquisition, component identification, progress modeling, pattern mining and calibration adjustment.

[0034] In a specific embodiment, the field data acquisition terminal is responsible for collecting various types of data in real time during the rice straw decomposition process, including near-infrared spectral data, image data, operational and environmental data, as well as carbon dioxide flux data and methane flux data.

[0035] Near-infrared spectroscopy data is used to analyze the content of soluble components, cellulose, hemicellulose, and lignin in straw. In a specific embodiment, a portable near-infrared spectrometer can be used for data collection, operating in the wavelength range of 900 nm to 1700 nm, covering the characteristic absorption peaks of the main components of straw. For example, soluble components such as sugars show significant absorption in the 1100 nm to 1300 nm range, cellulose and hemicellulose exhibit characteristic peaks in the 1400 nm to 1600 nm range, and lignin shows strong absorption near 1650 nm. The spectrometer is equipped with a high-resolution grating with a resolution better than 5 nm to ensure data accuracy. In a specific embodiment, the collection frequency can be set to once daily, conducted in the early morning when sunlight is stable to reduce ambient light interference. Each collection covers multiple sampling points in the field, distributed in a grid with a grid spacing of 2 meters to ensure coverage of the entire decomposition area.

[0036] Image data is used to assist in identifying physical morphological changes in straw, such as color, texture, and degree of fragmentation. The acquisition device employs a high-resolution RGB camera equipped with a wide-angle lens to cover a large field area. In a specific embodiment, image acquisition is synchronized with spectral data collection, with images taken once daily. Image data is stored in JPEG format, maintaining the original resolution to preserve detail. The image data is subsequently used to extract surface features of the straw; for example, a color change from green to yellowish-brown reflects the progress of decomposition, and a texture change from regular to loose reflects the degree of fragmentation.

[0037] Operational and environmental data include water layer status, irrigation and drainage operation records, secondary fragmentation parameters, inoculant type and dosage, soil temperature, soil moisture, pH value, and redox potential. This data collection was accomplished through a combination of sensor readings and manual recording. The specific implementation is as follows:

[0038] The water level was measured using an ultrasonic water level sensor. The sensor had an accuracy of ±1 mm and a measurement range of 0 to 30 cm. Data was recorded once per hour.

[0039] Irrigation and drainage operation records are manually entered by field managers through a cloud platform, recording the time, duration, and volume of each irrigation or drainage operation in cubic meters.

[0040] For secondary fragmentation parameters, the operating parameters of the mechanical fragmentation equipment are recorded, including fragmentation particle size, number of fragmentation cycles, and fragmentation coverage area, and are automatically uploaded to the cloud platform through the equipment's built-in log system.

[0041] For the type and dosage of microbial agents, the names of the microbial agents, such as Bacillus subtilis and Trichoderma, are recorded manually, and the dosage is uploaded through the cloud platform interface.

[0042] Soil temperature and humidity were recorded using embedded sensors. The temperature measurement range was -10℃ to 50℃ with an accuracy of ±0.5℃; the humidity measurement range was 0% to 100% with an accuracy of ±2%. The sensors were buried at a depth of 10 cm and recorded data once per hour.

[0043] pH and redox potential were measured using a portable soil analyzer. The pH measurement range was 3.5 to 9.0 with an accuracy of ±0.1; the redox potential measurement range was -500 mV to 500 mV with an accuracy of ±5 mV. Measurements were taken once daily.

[0044] Carbon dioxide and methane flux data reflect microbial activity and gas emissions during straw decomposition. In specific embodiments, a portable gas flux meter is used, equipped with a non-dispersive infrared sensor for carbon dioxide detection and a laser methane sensor for methane detection. The measurement points are consistent with the spectral data sampling points, and a static box method is used. In some embodiments, measurements can be taken weekly for 30 minutes each time, and flux data are recorded in milligrams per square meter per hour.

[0045] All collected data is transmitted in real time to the cloud platform via 4G or 5G networks and stored in a distributed database, such as MongoDB, in a time-series format. Each data point is accompanied by a timestamp and geographic coordinates to ensure traceability. The cloud platform preprocesses the data, including denoising, formatting, and missing value imputation, providing high-quality input for subsequent analysis.

[0046] As shown in Figure 2, the layered identification and modeling unit of this invention is responsible for identifying the initial mass proportions and decomposition progress of soluble components, cellulose, hemicellulose, and lignin in straw based on near-infrared spectral data and image data, and establishing a layered progress bar. The implementation method is described in detail below.

[0047] Component identification employs a supervised learning-based neural network model. In some embodiments, a combination architecture of convolutional neural networks (CNNs) and fully connected layers may be used. The inputs are near-infrared spectral data and image data, and the outputs are the mass percentage and degradation progress of each component. The model architecture includes:

[0048] Input layer: In a specific embodiment, the near-infrared spectral data is a one-dimensional vector with a length of 1600, corresponding to 900 nm to 1700 nm, with one data point per 1 nm; the image data can be a 128×128 pixel RGB image, which, after preprocessing such as normalization, is flattened into a 49152-dimensional vector (128×128×3). The two types of data are combined to form a mixed input.

[0049] Convolutional layers: Three convolutional operations are applied to the image data. In specific embodiments, the kernel sizes are 5×5, 3×3, and 3×3, and the number of channels are 32, 64, and 128, respectively. The ReLU activation function is used to extract spatial features. One-dimensional convolution is applied to the spectral data, with a kernel size of 5 and 32 channels, also using ReLU activation.

[0050] Fully connected layers: The feature vectors output from the convolutional layers are concatenated and input into a three-layer fully connected network with 512, 256, and 4 neurons, respectively, to predict the mass percentage of soluble components, cellulose, hemicellulose, and lignin. Another set of fully connected layers with 256, 128, and 4 neurons predicts the decomposition progress of each component, ranging from 0% to 100%.

[0051] Output layer: Outputs eight values. The first four are the mass percentages of each component, initially outputting the initial mass percentages. The last four are the degradation progress of each component. In some embodiments, the degradation progress can be set as the ratio of the degradation mass to the initial mass, where the degradation mass is the initial mass minus the currently estimated mass. Alternatively, the degradation progress can be trained directly through labels during training, allowing the model to directly output the degradation progress.

[0052] The training dataset consists of laboratory chemical analysis results and field measurement data. Chemical analysis uses wet chemical methods to determine the actual content of each component; for example, soluble components are determined by hot water extraction, cellulose and hemicellulose by acid hydrolysis, and lignin by sulfuric acid method. In a specific embodiment, the training sample includes 1000 sets of data, each containing spectral data, image data, and corresponding chemical analysis labels.

[0053] The training uses the mean squared error (MSE) as the loss function, and the formula is as follows:

[0054] ;

[0055] in, This refers to the actual content or decomposition progress obtained from chemical analysis. These are the model's predicted values. The sample size is denoted as . In a specific embodiment, the optimization algorithm uses Adam with a learning rate of 0.001, a batch size of 32, and 100 training epochs. Ten-fold cross-validation can be used during training to ensure the model's generalization ability.

[0056] The model outputs the mass percentage of each component. and decomposition progress For example, a prediction result might be: soluble component percentage 0.4%, decomposition progress 0.95%; cellulose percentage 0.3%, decomposition progress 0.25%; hemicellulose percentage 0.2%, decomposition progress 0.7%; lignin percentage 0.1%, decomposition progress 0.05%. These values ​​are directly mapped to layered progress bars, displayed on the cloud platform interface. The length of each progress bar is proportional to the decomposition progress. In some embodiments, different components can be distinguished by color, such as green for soluble components and blue for cellulose.

[0057] To ensure logical integrity, the model in this invention performs a quality check on the input data before each prediction. If the spectral data is abnormal, such as an absorbance value exceeding the reasonable range of 0 to 2, the abnormal points are removed and the data is filled in using interpolation with neighboring points; if the image data is blurry, the camera is triggered to retake the image. Example: In a certain field, the initial detected soluble component percentage was 0.45%, and the decomposition progress reached 0.92 after 10 days, indicating rapid decomposition, while the lignin progress was only 0.03, consistent with its slow decomposition characteristics.

[0058] The present invention integrates a progress calculation unit that weights the layered progress bar based on the initial mass proportion of each component to generate an overall corrosion progress bar. Its core formula is:

[0059] ;

[0060] in, For the first i The completion status of each layered progress bar. For the first i The initial mass percentage of each component satisfies n=4 corresponds to four components.

[0061] This formula is based on the principle of mass conservation and the differences in the contribution of each component to the overall decomposition. Straw decomposition is a multi-component synergistic process, and the initial mass proportion of each component determines its weight in influencing the overall progress. Weighted summation reflects this nonlinear dynamic, ensuring that the overall progress bar accurately reflects the biochemical process.

[0062] The progress output unit simultaneously displays layered progress bars and an overall progress bar through a cloud platform interface. In a specific embodiment, the interface is developed using web technologies, employing HTML5 and JavaScript, and the progress bars are drawn using Canvas. Each layered progress bar is displayed as a horizontal bar, with a length equal to... Proportional to the overall progress bar, which is displayed in bold and its length is proportional to the progress bar's length. Proportional.

[0063] As shown in Figure 3, in a further embodiment, the pattern mining unit mines patterns in the decay progress through cluster analysis, similarity comparison, and feature extraction, identifies potential blockage causes, and provides optimization suggestions.

[0064] First, cluster analysis was performed on the time series data of the historical stratified progress bars using the K-means algorithm. The number of clusters K was determined by the elbow rule, ranging from 3 to 10. The time series data consisted of the daily progress value for each component, with a length equal to the decomposition cycle, approximately 90 days. After clustering, progress bars with similar shapes were grouped together to form a prototype library. For example, a certain type of progress bar showed that soluble components were rapidly completed, reaching 90% within 10 days, while lignin remained stagnant for a long time, reaching only 10% after 90 days; this was classified as "slow lignin".

[0065] When the overall decomposition completion rate of a certain type is significantly lower than the average level, some embodiments may set it to be two standard deviations below the mean, and extract the corresponding operational and environmental data features for that type. Feature extraction includes:

[0066] Time series environmental data: Calculate the mean, variance, range, and trend slope.

[0067] Operational data: Calculate the operation frequency, such as watering twice a week, the operation interval, such as watering interval of 3 days, and the average parameter value, such as the amount of microbial agent applied is 10 grams per square meter.

[0068] Feature selection employs an information gain-based approach to identify features that occur frequently in the clogging category but infrequently in the non-clogging category. For example, low soil moisture (e.g., below 30%) and low inoculant application rate (e.g., below 5 g / m²) may be causes of clogging.

[0069] The current hierarchical progress bar is compared with historical progress bars in the prototype library. The Dynamic Time Warping (DTW) algorithm is used to calculate the time series similarity, with a threshold of 0.8. If the similarity exceeds the threshold, the corresponding historical blockage reason is extracted and analyzed for overlap with the current operation and environmental data. The overlap is calculated using the cosine similarity of the feature vectors, with a threshold of 0.7. If the overlap meets the standard, the overlapping features are output. For example, if the current progress bar is similar to "slow lignin" and the soil moisture is low, the output "It is recommended to increase the irrigation frequency to 3 times per week" is provided.

[0070] To further optimize the system, this invention selects several historical progress bars with the highest similarity to the previous progress bars, such as the top 5. It calculates the progress bar with the highest average similarity and selects the one with the fastest overall decomposition as a reference, recommending its operation and environmental data. For example, if the reference progress bar corresponds to a microbial agent dosage of 15 grams per square meter, it is recommended to adjust the current dosage to this value.

[0071] As shown in Figure 4, in a further embodiment, the present invention also includes an anchor point calibration unit for calibrating the stratified progress bar based on carbon dioxide and methane flux data. The specific rules are as follows:

[0072] When the carbon dioxide flux reaches an early peak, it indicates that the soluble components are rapidly decomposing, and its progress bar should be calibrated to no less than 90%; when cellulose decomposition begins, the progress bar should be calibrated to 20% to 30%.

[0073] When a lagged peak appears in the methane flux, it indicates active hemicellulose decomposition, and the progress bar is calibrated to 70% to 80%. Calibration is achieved through linear interpolation to ensure a smooth transition of the progress bar. For example, if the current progress of the soluble component is 85%, it will be directly set to 90% after calibration.

[0074] As shown in Figure 5, this embodiment illustrates the layered progress bars and overall progress bars of the rice straw decomposition process in the early and middle stages. It can be seen that the soluble components, due to their easy solubility, have reached approximately 90% decomposition, essentially complete; the cellulose components are in the initial stage of slow decomposition, with a completion rate of approximately 25%; the hemicellulose components have a faster degradation rate, with a completion rate of approximately 75%; and the lignin components are the most stable, with only about 40% complete. Based on the above layered progress bars and a weighted calculation combining their initial mass percentages, the overall decomposition progress bar completion rate is approximately 53%.

[0075] As shown in the figure, the overall progress curve is not linear: in the early stage, the rapid decomposition of soluble components drives a rapid increase in overall progress; in the middle stage, the overall progress is mainly affected by the decomposition rate of cellulose and hemicellulose, and the growth slows down; in the later stage, the slow degradation of lignin becomes the main limiting factor, and the growth of the overall progress gradually slows down. This schematic diagram intuitively demonstrates the system's dynamic monitoring capability of the entire straw decomposition process, accurately reflecting the degradation differences of different components and the nonlinear evolution characteristics of the overall progress, providing reliable data support for excavation and blockage analysis.

[0076] Through the above implementation methods, the present invention system achieves comprehensive monitoring and optimization of the rice straw decomposition process, and combined with the visualization output of the cloud platform, provides a scientific basis for agricultural production.

[0077] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A cloud-based system for monitoring the entire lifecycle of rice straw decomposition, characterized in that, include: Field data acquisition terminals are used to collect near-infrared spectral data, image data, operational and environmental data during the straw decomposition process; The layered identification and modeling unit is used to identify the initial mass ratio of soluble components, cellulose components, hemicellulose components and lignin components of straw based on the near-infrared spectral data and the image data, as well as the decomposition progress of each component, and to establish a layered progress bar for the corresponding decomposition process. A fusion progress calculation unit is used to weight the layered progress bars according to the initial proportion to generate an overall decomposition progress bar; A progress output unit is used to simultaneously output the layered progress bar and the overall corrosion progress bar; The pattern mining unit is used to perform cluster analysis on the time series data of historical hierarchical progress bars in groups, and divide the multi-group hierarchical progress bars with similar shapes into a progress category to form a prototype library; when the overall decomposition completion rate in a progress category is lower than the average level and exceeds a preset threshold, the common features of the operation and environmental data in the corresponding progress category are extracted as potential blocking reasons and output. The operational and environmental data include any one or more of the following: water layer status, irrigation and drainage operation records, secondary fragmentation parameters, type and dosage of microbial agents, soil temperature, soil moisture, pH value, and redox potential. The field data acquisition terminal is also used to collect carbon dioxide flux data and methane flux data. The system also includes an anchor calibration unit, used to calibrate the progress bar of the soluble component to a completion rate of no less than 90% and the progress bar of the cellulose component to a completion rate of 20% to 30% when an early peak is detected in the carbon dioxide flux data; and to calibrate the progress bar of the hemicellulose component to a completion rate of 70% to 80% when a late peak is detected in the methane flux data.

2. The cloud-based rice straw decomposition process full life cycle monitoring system according to claim 1, characterized in that, The pattern mining unit is also used to compare the current layered progress bar with the historical layered progress bars in the prototype library, and when the similarity exceeds a preset threshold, extract the potential blocking reasons corresponding to the historical layered progress bars and perform overlap analysis with the current operation and environment data. When the overlap reaches a preset threshold, the overlapping features are output as potential blocking reasons.

3. The cloud-based rice straw decomposition process full life cycle monitoring system according to claim 2, characterized in that, The pattern mining unit is also used to perform a preliminary similarity comparison between the current layered progress bar and the historical layered progress bars in the prototype library, and select multiple matching historical layered progress bars with the highest average similarity based on the comparison results. The progress bar with the fastest completion of the overall decay progress bar among the matching historical layered progress bars is selected as a reference for corresponding operations and adjustments to the environmental data.

4. The cloud-based rice straw decomposition process full life cycle monitoring system according to claim 1, characterized in that, The hierarchical identification and modeling unit is constructed using a supervised learning-based training model. The training model takes near-infrared spectral data and image data as input, and uses the actual contents of soluble components, cellulose components, hemicellulose components and lignin components obtained through chemical analysis as supervision labels. It is trained by minimizing the loss function between the predicted content and the chemically measured content.

5. The cloud-based rice straw decomposition process full life cycle monitoring system according to claim 1, characterized in that, The completion percentage P of the overall corrosion progress bar is determined by the following formula: ； in, Let i be the completion status of the i-th layer progress bar. Let i be the initial mass percentage of the i-th component, satisfying... n=4.

6. The cloud-based rice straw decomposition process full life cycle monitoring system according to claim 1, characterized in that, When the pattern mining unit extracts common features of operation and environment data within a certain progress category, it performs feature extraction on each set of operation and environment data within the progress category, forming multiple sets of feature vectors. It then aggregates and statistically analyzes the feature vectors and uses a feature selection algorithm to identify features that appear frequently in the samples of that category but appear infrequently in the non-blocking category. These features are then used as common features of the progress category and output as potential blocking causes.

7. The cloud-based rice straw decomposition process full life cycle monitoring system according to claim 6, characterized in that, The feature extraction includes calculating the mean, variance, range, and trend slope for time-series environmental data, and calculating the operation frequency, operation interval, and parameter mean for operational data.

8. The cloud-based rice straw decomposition process full life cycle monitoring system according to claim 6, characterized in that, The feature selection algorithm can be any of the following: feature ranking method based on information gain, correlation screening method based on mutual information, or sparse feature selection method based on LASSO regression.

Images