A microbial inoculant compounding method and system for molecular weight directional regulation in composting process

Abstract

This invention provides a method and system for molecular weight-directed regulation of microbial agent compounding during composting, belonging to the interdisciplinary field of environmental microbiology and artificial intelligence. It includes quantifying the molecular characteristics of dissolved organic matter (DOM) in composting materials based on various ratios, classifying it into low molecular weight organic matter (LMW), medium molecular weight organic matter (MMW), and high molecular weight organic matter (HMW). Feature screening is performed on LMW, MMW, and HMW to obtain characteristic variable data. This characteristic variable data is then used to train a machine learning model to obtain a DOM molecular weight prediction model. The original DOM molecular characteristic data of the multi-source composting materials to be compounded are acquired. Using the DOM proportions of LMW, MMW, and HMW as optimization targets, the original DOM molecular characteristic data are input into a coupled DOM molecular weight prediction model and the NSGA-II algorithm for optimization, resulting in an organic component ratio scheme. The organic component ratio scheme is then compounded to obtain the final microbial agent addition ratio. This achieves targeted regulation of DOM molecular weight, efficiently identifying key microbial communities and providing specific microbial agent addition schemes.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of environmental microbiology and artificial intelligence, specifically relating to a method and system for molecular weight-directed regulation of microbial agent compounding during composting. Background Technology

[0002] Composting is a key biotechnology for the resource utilization of organic solid waste. Essentially, it involves the orderly succession of microbial communities to degrade complex and unstable macromolecular organic matter into stable, humic substances. From a molecular perspective, this organic matter transformation is a dynamic process involving significant changes in the molecular weight distribution of organic matter: initial macromolecules such as proteins, polysaccharides, and lignin (high molecular weight substances) are degraded by extracellular enzymes secreted by microorganisms into intermediate products such as oligopeptides and oligosaccharides (medium molecular weight substances), which are further metabolized into small-molecule organic acids, CO2, and H2O (low molecular weight substances). Simultaneously, through microbial metabolism and a series of condensation reactions, complex macromolecules such as humic acids are resynthesized. The use of compound microbial agents in the composting process involves the scientific combination of microbial strains with different functions or characteristics to achieve synergistic effects on organic solid waste, expand its application scope, or enhance its environmental adaptability.

[0003] Traditional optimization and control of composting processes have largely focused on analyzing macroscopic parameters (such as temperature, C / N ratio, and aeration rate) or the overall structure of the microbial community. However, these traditional methods have significant limitations: they cannot precisely guide the targeted enrichment of functional microorganisms to achieve accurate control over the molecular weight distribution of compost products. With the application and development of machine learning in the environmental and composting fields, a new trend of intelligent composting has gradually emerged, improving upon the shortcomings of traditional methods to some extent. However, because machine learning lacks analysis of the molecular weight hierarchical evolution of soluble organic matter (DOM) in multi-source organic solid waste composting systems, it is difficult to target and control the molecular weight of soluble organic matter (DOM), and it is also impossible to efficiently screen inoculants from complex microbial communities to achieve inoculant compounding, thus hindering the optimal role of synergistic and targeted regulation of microbial communities during the composting process. Summary of the Invention

[0004] To address the shortcomings of molecular weight-directed regulation during composting, this invention provides a method and system for molecular weight-directed regulation of microbial agents during composting.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for compounding microbial agents with molecular weight-directed regulation during composting includes the following steps: The molecular characteristics of soluble organic matter (DOM) were quantified for compost materials with different ratios to obtain the first molecular weight organic matter (LMW), the second molecular weight organic matter (MMW), and the third molecular weight organic matter (HMW); the molecular weights of LMW, MMW, and HMW increase sequentially. The importance of LMW, MMW and HMW was assessed, and organic components associated with different molecular weights were screened. Using the content of organic components as input and the proportion of organic matter with different molecular weights as label data, a machine learning model was trained to obtain the DOM molecular weight prediction model. The molecular weight data of the compost material to be formulated is obtained, and different molecular weight optimization targets are set. The NSGA-II algorithm is used to optimize the organic component allocation ratio of the target molecular weight. The organic component ratio combination of the individuals in the population during optimization is input into the DOM molecular weight prediction model to obtain the predicted value of the molecular weight ratio. It is determined whether the predicted value of the molecular weight ratio exceeds the set threshold. Individuals that exceed the threshold are penalized and eliminated. The population is iterated step by step through genetic operations to obtain the optimal individual. Based on the optimal individual, the optimal organic component allocation ratio scheme under the corresponding target is obtained. The population includes multiple material compounding schemes, and each individual in the population represents a specific organic component allocation ratio scheme. Functional bacterial genera associated with the formation of the target molecular weight (DOM) are screened based on the genus-level bacterial community during the composting process. The contribution weight of key functional bacterial genera is calculated based on the predicted molecular weight percentage. The organic component allocation scheme is then compounded based on the contribution weight to obtain the final bacterial agent addition ratio.

[0006] Preferably, the NSGA-II algorithm is used to optimize the organic component allocation ratio for the target molecular weight to obtain the optimal organic component allocation ratio scheme for the corresponding target, specifically including the following steps: Initialize the population, where each individual is a three-dimensional list of the DOM proportions in the LMW, MMW, and HMW; set content threshold boundaries for the organic components of the compost DOM; Based on a preset population size, an initial set of individuals satisfying the constraints is randomly generated, and the organic component characteristic value of each individual is obtained. The organic component ratio combination of the individuals in the population during optimization is input into the DOM molecular weight prediction model to obtain the predicted molecular weight ratio. It is then evaluated whether to apply a penalty term to guide the individuals in the population away from invalid solutions. The fitness weight is adjusted according to the directional regulation requirements, the fitness value of each individual is calculated and sorted, and individuals with high fitness are selected as parents. Offspring are generated using single-point crossover. After crossover, it is checked whether the characteristic value exceeds the boundary. If it does, it is truncated to the boundary value. The individual characteristic values ​​are randomly perturbed to introduce new mutations and avoid the population from getting trapped in local optima. The parent generation, crossover offspring, and mutated offspring are merged into a temporary population. Non-dominated sorting and crowding calculation are performed on the temporary population again. The optimal individuals are selected according to the preset population size to form a new generation population. Through multiple generations of evolution, the Pareto optimal front is gradually approached to obtain the optimal solution under multiple objectives. Based on the optimal solution under multiple objectives, the optimal organic group allocation scheme is obtained.

[0007] Preferably, Fourier transform ion cyclotron resonance mass spectrometry is used to quantify the molecular characteristics of soluble organic matter (DOM) in compost materials with different ratios.

[0008] Preferably, the genus-level bacterial community is analyzed for its response to different organic components to screen out functional genera associated with the formation of the target molecular weight DOM.

[0009] Preferably, the calculation of the contribution weight of key functional bacterial genera specifically involves: quantifying the contribution weight of independent bacterial genera through variance decomposition analysis (VPA); calculating the comprehensive weight based on the absolute value of the Spearman correlation coefficient between the contribution weight of the genera and the molecular weight of the target DOM; and normalizing the comprehensive weight of the screened genera to obtain the contribution weight of the key functional bacterial genera.

[0010] Preferably, the importance assessment of LMW, MMW, and HMW, and the screening of organic components associated with different molecular weights, specifically involves using Random Forest (RF) to assess the importance of features, and selecting the top n features of importance for each of LMW, MMW, and HMW as associated organic components.

[0011] Preferably, the machine learning model includes: a GBR model and an XGBoost model; wherein, the GBR model is used to predict the organic matter of LMW and MMW, and the XGBoost model is used to predict the organic matter of HMW.

[0012] This invention also provides a microbial agent compounding system for molecular weight-directed regulation during composting, specifically comprising: The data processing module is used to quantify the molecular characteristics of soluble organic matter (DOM) in compost materials with different ratios, and obtain the first molecular weight organic matter (LMW), the second molecular weight organic matter (MMW), and the third molecular weight organic matter (HMW); the molecular weights of LMW, MMW, and HMW increase sequentially.

[0013] The model module is used to assess the importance of LMW, MMW, and HMW and screen organic components associated with different molecular weights. Using the content of organic components as input and the proportion of organic matter with different molecular weights as label data, a machine learning model is trained to obtain the DOM molecular weight prediction model.

[0014] The analysis and compounding module is used to acquire molecular weight data of the compost materials to be compounded and set different molecular weight optimization targets. The NSGA-II algorithm is used to optimize the organic component allocation ratio of the target molecular weight. The organic component ratio combination of the individuals in the population during optimization is input into the DOM molecular weight prediction model to obtain the predicted value of the molecular weight ratio. It is determined whether the predicted value of the molecular weight ratio exceeds the set threshold. Individuals that exceed the threshold are penalized and eliminated. The population is iterated step by step through genetic operations to obtain the optimal individual. Based on the optimal individual, the optimal organic component allocation ratio scheme under the corresponding target is obtained. The population includes multiple material compounding schemes, and each individual in the population represents a specific organic component allocation ratio scheme.

[0015] Functional bacterial genera associated with the formation of the target molecular weight (DOM) are screened based on the genus-level bacterial community during the composting process. The contribution weight of key functional bacterial genera is calculated based on the predicted molecular weight percentage. The organic component allocation scheme is then compounded based on the contribution weight to obtain the final bacterial agent addition ratio.

[0016] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps described in the method for molecular weight-directed regulation of microbial agent compounding during composting.

[0017] The present invention also provides a computer-readable storage medium storing a computer program, which, when loaded by a processor, is capable of executing the steps described in the method for molecular weight-directed regulation of microbial agent compounding during composting.

[0018] The present invention provides a method for compounding microbial agents with molecular weight-directed regulation during composting, which has the following beneficial effects: This invention classifies and categorizes the molecular characteristic data of dissolved organic matter (DOM) in compost from different materials, clarifying the target objectives for different application scenarios. A multi-objective optimization framework based on the NSGA-II algorithm is constructed, achieving synergistic optimization of DOM components with different molecular weights in multi-source materials by coupling machine learning prediction models and evolutionary algorithms. This ensures that the optimized compound formulation strictly adheres to the stoichiometry of organic matter, guaranteeing the physical feasibility of the final solution. Simultaneously, the functionality of compost products can be directionally controlled according to actual needs, precisely guiding the generation of HMW or LMW organic matter through the addition of specific proportions of substances. In addition to screening dominant microbial strains, the proportions of various bacteria are reverse-optimized, providing specific schemes for adding microbial agents, realizing a shift from experience-driven to model-driven approaches. Attached Figure Description

[0019] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a method for compounding microbial agents with molecular weight directional control during composting, according to an embodiment of the present invention.

[0021] Figure 2 This is the analysis result of the random forest model variable selection for feature variables in this embodiment of the invention.

[0022] Figure 3 These are the prediction results of the SVR, GBR, and XGBoost models for three target variables in embodiments of the present invention. Among them, Figure 3 (a) represents the prediction results for low molecular weight molecules; Figure 3 (b) represents the predicted result for the medium molecular weight; Figure 3 (c) represents the high molecular weight prediction result.

[0023] Figure 4 This is an optimized compounding scheme for organic components corresponding to low, medium, and high molecular weight organic matter in this embodiment of the invention. Wherein, Figure 4 (a), (b), and (c) correspond to the optimal organic component blending schemes for low, medium, and high molecular weight organic matter, respectively.

[0024] Figure 5 The VPA analysis results are based on low, medium, and high molecular weight DOM from compost in this embodiment of the invention. Figure 5 (a), (b), and (c) correspond to VPA analysis results based on low, medium, and high molecular weight DOM, respectively. Detailed Implementation

[0025] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0026] Example This invention provides a method for compounding microbial agents with molecular weight-directed regulation during composting, such as... Figure 1 As shown, the specific steps include: Step 1: Analysis of DOM molecular properties of multi-source materials.

[0027] Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) was used to quantify DOM molecular characteristics. In the experimental treatment groups, various composts with different ratios were used: chicken manure compost treatment group (CM), pig manure compost treatment group (PM), chicken manure mixed with cow manure compost treatment group (CCM), pig manure mixed with cow manure compost treatment group (PCM), chicken manure mixed with kitchen waste compost treatment group (CKW), and pig manure mixed with kitchen waste compost treatment group (PKW).

[0028] To visualize the FT-ICRMS data, the elemental ratios of O / C and H / C were determined and visualized in a van Krevelen diagram. Based on the elemental ratios of O / C and H / C, combined with the modified aromaticity index (AImod), DOM can be divided into different categories, including lipids (0 < O / C < 0.2 and 1.5 < H / C < 2.3), proteins (0.2 < O / C < 0.52 and 1.5 < H / C < 2.2), amino sugars (0.52 < O / C < 0.7 and 1.5 < H / C < 2.2), carbohydrates (0.7 < O / C < 1.1 and 1.5 < H / C < 2.4), lignin / CRAM-like (0.25 < O / C < 0.67 and 0.75 < H / C < 1.5), tannins (0.67 < O / C < 0.97 and 0.53 < H / C < 1.5), and condensed aromatics (0 < O / C < 0.25 and 0.5 < H / C < 1.25). The molecular weights of DOM detected in the experiments mainly distributed in the range of 100 - 800 daltons (Da). According to the differences in their molecular characteristics, they were divided into three categories: organic matter with a molecular weight below 350 Da was defined as low molecular weight (LMW) organic matter, which has a simple molecular structure and prominent biological availability, and can quickly participate in the microbial metabolic process and drive the cycling of nutrient elements; DOM in the range of 350 - 500 Da was defined as medium molecular weight (MMW) organic matter, which serves as an intermediate bridging component of DOM molecules, with a polymorphic molecular structure, showing certain interfacial reaction activity and dissolution and migration ability in the environmental medium; DOM in the range of 500 - 800 Da was classified as high molecular weight (HMW) organic matter, whose molecules show a relatively complex spatial configuration and chemical stability, endowing it with strong resistance, and may become an important storage form of soil organic carbon pool.

[0029] Step 2: Construct a machine learning model coupled with multiple parameters for quantitative prediction of the formation rules of different DOM molecular weight differentiation in a multi-source material composting system.

[0030] Through data preprocessing, outlier removal, and feature importance evaluation, key variables were identified and screened, aiming to provide strong data support and a scientific basis for model optimization.

[0031] Data preprocessing: Using IQR combined with box plots, the distribution of the data was measured sequentially using five representative data points: maximum, third quartile, median, first quartile, and minimum. Outlier handling: Box plots were used to identify outliers for each feature indicator. A very small number of outliers were found in tannins, amino sugars, and lipids. Considering the limited data volume, the median from the feature samples was used for replacement. Furthermore, high collinearity among feature variables could distort parameter estimation in the machine learning model; therefore, pairwise correlations between features needed to be assessed. Spearman's rank correlation coefficient was used for correlation testing, and the results showed good independence among the feature variables.

[0032] Further selection of input features for the prediction model was performed using the feature importance evaluation module of Random Forest (RF). Calculations were performed separately for three different target variables: LMW, MMW, and HMW. First, the hyperparameters of the RF model were tuned using a random search algorithm. The optimal parameters were then substituted into the RF model for training, and the importance ranking of each feature was calculated. The top five most important features were selected as the input features for the final prediction model. After hyperparameter tuning, the key parameters of the Random Forest model were configured as follows: n_estimators=200, min_samples_split=2, min_samples_leaf=2. The feature importance analysis results are as follows: Figure 2 As shown, the feature selection results for LMW are: condensed aromatics, amino sugars, lignin / CRAM-like substances, tannins, and carbohydrates; the feature selection results for MMW are: condensed aromatics, tannins, amino sugars, carbohydrates, lignin / CRAM-like substances; and the feature selection results for HMW are: carbohydrates, proteins, condensed aromatics, lignin / CRAM-like substances, and tannins.

[0033] Step 3: Three regression prediction models—SVR, GBR, and XGBoost—were built, and the corresponding SVR, GradientBoostingRegressor, and XGBRegressor modules from the integrated toolkit were called to initialize the model framework. A random search algorithm was introduced to optimize the hyperparameters of each model. This algorithm traverses all possible parameter combinations within a predefined parameter search space to locate the optimal parameter configuration for the model. The optimized parameters for each model are as follows:

[0034] SVR model: kernel function type, gamma value, epsilon, regularization parameter C.

[0035] GBR model: number of decision trees (n_estimators), minimum number of split samples (min_samples_split), maximum tree depth (max_depth), and learning rate (learning_rate), etc.

[0036] XGBoost model: parameters such as subsample rate, number of decision trees, maximum depth, learning rate, and feature sampling ratio (colsample_bytree).

[0037] The model was trained based on the feature data selected in step two, and the performance of multiple models was compared to select the optimal prediction model. The ensemble models (GBR and XGBoost) showed better prediction accuracy than the single model SVR. Specifically, GBR performed best in predicting LMW and MMW organic matter (R² reached 0.79 and 0.74 respectively after hyperparameter optimization), while XGBoost showed a significant advantage in predicting HMW organic matter (R² reached 0.90). The ensemble models, through a boosting mechanism that strings together weak learners, effectively capture the interactions and nonlinear relationships between features, significantly improving prediction accuracy compared to SVR and providing technical support for intelligent control of composting. Figure 3 As shown, the optimal prediction model was selected as the XGBoost model.

[0038] Step 4: Based on the prediction model, the machine learning prediction model is coupled with an evolutionary algorithm. Using the NSGA-II algorithm-based multi-objective optimization framework, the synergistic optimization of DOM components with different molecular weights in multi-source materials is achieved, and the quantitative output of the reverse compounding scheme is completed.

[0039] The multi-objective optimization framework specifically (1) constructs a dynamic weight adjustment strategy, adjusts the weight coefficients of each objective in the fitness function, and combines the personalized needs of specific application scenarios to configure exclusive constraints for the prediction model, so as to realize effective constraint control of the output of other objectives during the single objective optimization process and meet the differentiated optimization needs of diverse scenarios; (2) integrates a constraint maintenance genetic operation module, and monitors the content threshold boundary of each organic component of compost DOM in real time during the execution of crossover and mutation operators to ensure that the optimized results strictly follow the stoichiometric laws of organic matter; (3) through the joint setting of multi-dimensional constraints and optimization objectives, the output results are mapped to the index inverse normalization to restore to the original feature space, and combined with the adaptive penalty function to carry out multiple rounds of iterative optimization to ensure the physical feasibility of the final solution.

[0040] Based on the aforementioned multi-objective optimization framework, the target variables to be optimized are set according to different application requirements. By inversely solving the input feature space of machine learning, the compounding schemes of organic heavy components are obtained, enabling quantitative analysis of the formation conditions of organic matter with different molecular weights. Specifically, the following steps are included:

[0041] First, an initial set of candidate solutions (population) satisfying the constraints is generated. Each individual is a three-dimensional list (LMW, MMW, HMW), corresponding to the proportion of the three molecular weights. The constraint ranges for the eigenvalues ​​are defined (e.g., 'Lipids':(0.0001,0.62), 'Proteins':(0.035,0.46)). The population size is set, i.e., the number of initially generated random individuals.

[0042] The fitness value is then used to evaluate the quality of each individual. The generated individual features are input into the previously trained XGBoost model to obtain the predicted molecular weight percentage. Based on the predicted molecular weight percentage, it is evaluated whether to apply a penalty term to filter out invalid solutions and avoid meaningless molecular weight percentages. If the sum of features exceeds a threshold (e.g., total LMW > 0.86), a linear penalty (10×) is applied to guide the population away from invalid solutions.

[0043] The weights are dynamically adjusted based on the optimization objective: Default mode (maximizing HMW): Fitness value is (-LMW, -MMW, HMW) (negative values ​​are used for objectives where smaller values ​​are better, and positive values ​​are used for objectives where larger values ​​are better). max_lmw mode (maximizing LMW): Fitness value is (LMW, -MMW, -HMW).

[0044] Individuals with high fitness are selected from the current population to serve as parents for offspring generation. Offspring are generated through gene recombination of parent individuals, increasing population diversity. Single-point crossover is used to generate offspring: eigenvalues ​​at the same position in parent individuals are exchanged with a 50% probability.

[0045] Constraints must be met: immediately after the crossover, check whether the eigenvalue exceeds the boundary (e.g., Lipids cannot exceed 0.62). If it does, truncate to the boundary value.

[0046] Randomly perturb individual eigenvalues ​​to introduce new mutations and prevent the population from getting trapped in local optima. Each eigenvalue is perturbed by a small amount (±0.1) with a 20% probability, simulating random mutations in natural evolution. After mutation, the eigenvalue boundaries are checked again to ensure the feasibility of the solution. By using small-scale perturbations to escape local optima, the population's exploratory ability is increased, while constraints ensure that the mutated solution remains valid.

[0047] The main loop of the genetic algorithm is executed to generate a new generation of population.

[0048] By gradually approaching the Pareto optimal frontier through multiple generations of evolution, the optimal solution set under multiple objectives is finally obtained.

[0049] Multi-objective optimization framework parameter settings: mu=100: parent population size (retain 100 individuals); lambda_=100: offspring population size (generate 100 offspring); cxpb=0.7: crossover probability (70% of parent pairs will crossover); mutpb=0.2: mutation probability (20% of offspring will mutate); ngen=50: number of generations (execute 50 selection-crossover-mutation cycles).

[0050] Quantitative analysis shows that when the proportion of DOM (domestic organic matter) components in compost meets a specific threshold, it can directionally induce the optimal formation of the target molecular weight. In the optimized LMW (lumenstone millihydrate) organic matter model, the proportion of amino sugars added is 19.36%, while the relative content of lignin / CRAM-like substances needs to reach 64.91%. For the optimal conditions for the directional formation of MMW (medium millihydrate millihydrate), the relative content of lignin / CRAM-like substances needs to be controlled at 71.91%, while the proportions of tannins and carbohydrates are maintained at 1.79% and 1.83%, respectively. For the optimal guided formation scheme of HMW (high-molecular-weight millihydrate millihydrate), a synergistic system of tannins (22.22%) and proteins (12.60%) needs to be constructed, while maintaining the relative content of lignin / CRAM-like substances at the critical threshold of 53.20%. Figure 4 As shown.

[0051] Step 5: Compound preparation of microbial agents.

[0052] Response relationship analysis was conducted between genus-level bacterial communities and different types of organic components in the organic matter of multi-source organic waste compost. After screening and filtering, bacterial communities that were significantly associated with seven organic components of organic matter were selected for network analysis (P<0.05).

[0053] The bacterial community structure influencing DOM transformation varies significantly across different composting systems. In single-source composting systems, the CM group exhibits a richer abundance of protein and carbohydrate metabolic bacteria, but its ability to transform recalcitrant compounds is limited. The PM group's bacteria, through intensive interactions, enhance the synergistic transformation of complex organic compounds such as condensed aromatics, which is related to the concentrated molecular weight distribution and high unsaturation of the initial DOM molecules in pig manure. In mixed composting systems, both the CCM and PCM groups show complex microbial functional interactions. However, the CCM group is dominated by a core group of readily degradable nitrogen-containing bacteria, which guides the initial DOM molecules to be rich in highly active unsaturated reduced compounds. In contrast, the PCM group shows a closer cooperative relationship between components and bacteria, demonstrating a certain degree of ability to transform complex structures. In mixed food waste systems, readily degradable carbon sources provide a resource advantage, resulting in a relatively simple microbial metabolic network. The CKW group, due to its advantage in reducing compounds, exhibits a short-term resource-based metabolic preference that inhibits the transformation of other components, while the PKW group, due to the abundance of saturated oxidized compounds in its initial DOM molecules, shows an advantage in aromatic structure degradation. The study, combining changes in microbial community structure and interspecies interaction networks during the composting of different materials, indicates that initial differences in organic components within the compost guide multidirectional transformation of the microbial community. Furthermore, the diverse microbial community structure within different composting systems is a crucial factor influencing the directional transformation of organic components in compost. In conclusion, the interactions between microorganisms during composting are a significant driving force influencing the composition of the organic matter matrix (DOM) in compost.

[0054] Therefore, by combining different proportions of microbial agents, the targeted synthesis of DOM molecular weight can be achieved, thus enabling targeted regulation of compost product functions. The screening of functional microbial agents is based on DOM molecular weight characteristics, combined with the functional characteristics of the microbial community for initial screening and optimization. Furthermore, considering that the functions of the microbial genera must meet the specific needs of compost products and be able to effectively participate in the synthesis or degradation of the target molecular weight DOM, strains with potential for microbial formulation or practical application in the market are prioritized. Table 1 shows the target microbial genera after screening.

[0055] Table 1 Target bacterial genera after screening Based on the established DOM molecular weight prediction model, a method combining variance partitioning analysis (VPA) and correlation coefficients was used to synthesize the contribution weights of fungal genera and construct an optimization model for the compound ratio of fungal agents. VPA quantifies the proportion of specified environmental factors explaining changes in community structure by decomposing the response variable (DOM target molecular weight) into the independent contributions and interaction effects of explanatory variables (relative abundance of key functional genera). The mathematical expression of VPA is as follows:

[0056] Var(Y) = Independent contribution ∑Var(Y|X) i + Interactive contribution Var(Y|X) i ∩X j + residual; Where Y: DOM molecular weight (HMW / LMW / MMW ratio); X i : Abundance of key bacterial genera.

[0057] Independent contribution weight W i =Var(Y|X) i ) / ∑Var(Y∣X i )×100%; Total contribution weight W i,total =Var(Y|X) i )+0.5×Var(Y|X i ∩X j ) / Total explained rate × 100% (interaction effects are evenly distributed 50% to each genus); There are still many undefined variables (residuals) in the generation and transformation of the three molecular weight DOMs, which may be affected by environmental factors (such as temperature and oxygen concentration) or other microbial communities. Therefore, in order to comprehensively quantify the proportion of functional bacteria added and achieve their precise application and synergistic effect in composting, a comprehensive weighting method combining variance decomposition and correlation coefficients was used to optimize the compounding ratio.

[0058] Based on VPA analysis results and correlation coefficients as comprehensive weighting factors, the formulation ratio of functional microbial agents was optimized. First, the weights of independent genera and synergistic contributions were quantified using VPA. The synergistic contribution weight of each genera was combined with its correlation coefficient with the target DOM molecular weight, and the comprehensive weight was calculated by averaging the sum of these two values. Finally, the comprehensive weights of the screened genera were normalized to obtain the final formulation ratio.

[0059] W 综合 = (Total contribution weight + Relevance weight) / 2; Relevance weight = |ρ Spearman ∣ / ∑∣ρ∣; The final ratio of microbial agent added is obtained by normalizing the overall weight: Add ratio = *100%; By comprehensively considering independent contributions and correlation data, and combining variance decomposition with optimized calculations of correlation coefficients, the optimal ratio of the three DOM molecules in the compound bacterial agent was determined to be: LWM:Pseudomonas:Paracoccus:Lysobacter=58.1%:25.4%:16.5%.

[0060] MWM:Prosthecobacter:Acidovorax:Paenibacillus:Aeribacillus=59.5%:19.1%:11.5%:9.9%.

[0061] HWM:Paenisporosarcina:Thermoactinomyces:Novibacillus:Stackebrandtia=41.7%:28.2%:16.3%:13.8%.

[0062] This invention also provides a microbial agent compounding system for molecular weight-directed regulation during composting, comprising: The data processing module is used to quantify the molecular characteristics of soluble organic matter (DOM) in compost materials with different ratios, and obtain the first molecular weight organic matter (LMW), the second molecular weight organic matter (MMW), and the third molecular weight organic matter (HMW); the molecular weights of LMW, MMW, and HMW increase sequentially.

[0063] The model module is used to assess the importance of LMW, MMW, and HMW and screen organic components associated with different molecular weights. Using the content of organic components as input and the proportion of organic matter with different molecular weights as label data, a machine learning model is trained to obtain the DOM molecular weight prediction model.

[0064] The analysis and compounding module is used to acquire molecular weight data of the compost materials to be compounded and set different molecular weight optimization targets. The NSGA-II algorithm is used to optimize the organic component allocation ratio of the target molecular weight. The organic component ratio combination of the individuals in the population during optimization is input into the DOM molecular weight prediction model to obtain the predicted value of the molecular weight ratio. It is determined whether the predicted value of the molecular weight ratio exceeds the set threshold. Individuals that exceed the threshold are penalized and eliminated. The population is iterated step by step through genetic operations to obtain the optimal individual. Based on the optimal individual, the optimal organic component allocation ratio scheme under the corresponding target is obtained. The population includes multiple material compounding schemes, and each individual in the population represents a specific organic component allocation ratio scheme.

[0065] Functional bacterial genera associated with the formation of the target molecular weight (DOM) were screened based on the genus-level bacterial community during the composting process. The contribution weight of key functional bacterial genera was calculated based on the predicted molecular weight percentage. Based on the contribution weight, the organic component allocation scheme was compounded to obtain the final bacterial agent addition ratio.

[0066] The modules in the aforementioned molecular weight-directed microbial agent compounding system for composting can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0067] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps in an embodiment of a method for molecular weight-directed regulation of microbial agent compounding during composting. Specific implementation methods can be found in the method embodiments, and will not be repeated here.

[0068] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions, on which a computer program is stored. For example, a memory containing instructions that can be executed by a processor of a computer device to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. When the computer program is executed by the processor, it can implement the steps in an embodiment of a method for molecular weight-directed regulation of microbial agent compounding during composting. Specific implementation methods can be found in the method embodiments, and will not be repeated here.

[0069] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0070] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0071] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0072] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0073] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail in this specification and embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered within the protection scope of the present invention patent. No reference numerals in the claims should be construed as limiting the scope of the claims. Any simple variations or equivalent substitutions of technical solutions that can be readily obtained by those skilled in the art within the scope of the technology disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A microbial inoculant compounding method for molecular weight directional regulation in composting process, characterized in that, Includes the following steps: The molecular characteristics of soluble organic matter (DOM) were quantified for compost materials with different ratios to obtain the first molecular weight organic matter (LMW), the second molecular weight organic matter (MMW), and the third molecular weight organic matter (HMW); the molecular weights of LMW, MMW, and HMW increase sequentially. The importance of LMW, MMW and HMW was assessed, and organic components associated with different molecular weights were screened. Using the content of organic components as input and the proportion of organic matter with different molecular weights as label data, a machine learning model was trained to obtain the DOM molecular weight prediction model. The molecular weight data of the compost material to be formulated is obtained, and different molecular weight optimization targets are set. The NSGA-II algorithm is used to optimize the organic component allocation ratio of the target molecular weight. The organic component ratio combination of the individuals in the population during optimization is input into the DOM molecular weight prediction model to obtain the predicted value of the molecular weight ratio. It is determined whether the predicted value of the molecular weight ratio exceeds the set threshold. Individuals that exceed the threshold are penalized and eliminated. The population is iterated step by step through genetic operations to obtain the optimal individual. Based on the optimal individual, the optimal organic component allocation ratio scheme under the corresponding target is obtained. The population includes multiple material compounding schemes, and each individual in the population represents a specific organic component allocation ratio scheme. Functional bacterial genera associated with the formation of the target molecular weight (DOM) are screened based on the genus-level bacterial community during the composting process. The contribution weight of key functional bacterial genera is calculated based on the predicted molecular weight percentage. The organic component allocation scheme is then compounded based on the contribution weight to obtain the final bacterial agent addition ratio.

2. The method for compounding microbial agents with molecular weight directional regulation during composting according to claim 1, characterized in that, The NSGA-II algorithm is used to optimize the organic component allocation ratio for the target molecular weight, resulting in the optimal organic component allocation ratio scheme for the corresponding target. The specific steps include: Initialize the population, where each individual is a three-dimensional list of the DOM proportions in the LMW, MMW, and HMW; set content threshold boundaries for the organic components of the compost DOM; Based on a preset population size, an initial set of individuals satisfying the constraints is randomly generated, and the organic component characteristic value of each individual is obtained. The organic component ratio combination of the individuals in the population during optimization is input into the DOM molecular weight prediction model to obtain the predicted molecular weight ratio. It is then evaluated whether to apply a penalty term to guide the individuals in the population away from invalid solutions. The fitness weight is adjusted according to the directional regulation requirements, the fitness value of each individual is calculated and sorted, and individuals with high fitness are selected as parents. Offspring are generated using single-point crossover. After crossover, it is checked whether the characteristic value exceeds the boundary. If it does, it is truncated to the boundary value. The individual characteristic values ​​are randomly perturbed to introduce new mutations and avoid the population from getting trapped in local optima. The parent generation, crossover offspring, and mutated offspring are merged into a temporary population. Non-dominated sorting and crowding calculation are performed on the temporary population again. The optimal individuals are selected according to the preset population size to form a new generation population. Through multiple generations of evolution, the Pareto optimal front is gradually approached to obtain the optimal solution under multiple objectives. Based on the optimal solution under multiple objectives, the optimal organic group allocation scheme is obtained.

3. The method according to claim 1, wherein the microbial inoculant is a microbial inoculant for composting process. Fourier transform ion cyclotron resonance mass spectrometry was used to quantify the molecular characteristics of dissolved organic matter (DOM) in compost materials with different ratios.

4. The method according to claim 1, wherein the microbial inoculant is a microbial inoculant for composting process. By analyzing the response relationship between genus-level bacterial communities and different organic components, functional genera associated with the formation of DOM (monomolecules with a target molecular weight) were screened.

5. The method according to claim 1, wherein the microbial inoculant is a microbial inoculant for composting process. The calculation of the contribution weight of key functional bacterial genera specifically involves: quantifying the contribution weight of independent bacterial genera through variance decomposition analysis (VPA); calculating the comprehensive weight based on the absolute value of the Spearman correlation coefficient between the contribution weight of the genera and the molecular weight of the target DOM; and normalizing the comprehensive weight of the screened genera to obtain the contribution weight of the key functional bacterial genera.

6. The method according to claim 1, wherein the microbial inoculant is a microbial inoculant for composting process. The importance assessment of LMW, MMW, and HMW, and the screening of organic components associated with different molecular weights, specifically uses Random Forest (RF) to assess the importance of features, and selects the top n features of importance for LMW, MMW, and HMW as associated organic components.

7. The method according to claim 1, wherein the microbial inoculant is a microbial inoculant for composting process. The machine learning model includes: a GBR model and an XGBoost model; wherein, the GBR model is used to predict the organic matter of LMW and MMW, and the XGBoost model is used to predict the organic matter of HMW.

8. A microbial inoculants compounding system for molecular weight directed regulation in composting process, characterized in that, include: The data processing module is used to quantify the molecular characteristics of soluble organic matter (DOM) in compost materials with different ratios, and obtain the first molecular weight organic matter (LMW), the second molecular weight organic matter (MMW), and the third molecular weight organic matter (HMW); the molecular weights of LMW, MMW, and HMW increase sequentially. The model module is used to assess the importance of LMW, MMW, and HMW and screen organic components associated with different molecular weights. Using the content of organic components as input and the proportion of organic matter with different molecular weights as label data, a machine learning model is trained to obtain the DOM molecular weight prediction model. The analysis and compounding module is used to acquire molecular weight data of the compost materials to be compounded and set different molecular weight optimization targets. The NSGA-II algorithm is used to optimize the organic component allocation ratio of the target molecular weight. The organic component ratio combination of the individuals in the population during optimization is input into the DOM molecular weight prediction model to obtain the predicted molecular weight ratio. It is determined whether the predicted molecular weight ratio exceeds a set threshold. Individuals that exceed the threshold are penalized and eliminated. The population is iterated step by step through genetic operations to obtain the optimal individual. Based on the optimal individual, the optimal organic component allocation ratio scheme under the corresponding target is obtained. The population includes multiple material compounding schemes, and each individual in the population represents a specific organic component allocation ratio scheme. Functional bacterial genera associated with the formation of the target molecular weight (DOM) are screened based on the genus-level bacterial community during the composting process. The contribution weight of key functional bacterial genera is calculated based on the predicted molecular weight percentage. The organic component allocation scheme is then compounded based on the contribution weight to obtain the final bacterial agent addition ratio.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is loaded by the processor, it is able to perform the steps of the method according to any one of claims 1 to 7.

Images