Microbial fermentation culture medium optimization method for agricultural feed production

Abstract

The invention provides a microbial fermentation culture medium optimization method for agricultural feed production, which comprises the following steps: grouping and calibrating strain metabolic pathways according to a potential decomposition demand list to obtain an optimized metabolic pathway set; screening nutrient element combinations from a resource information base through the optimized metabolic pathway set, and determining element proportion configuration to obtain an initial culture medium formula draft; according to the initial culture medium formula draft, adopting a prediction simulation method to predict the strain growth state, and if the predicted growth activity is insufficient, iteratively adjusting the element proportion to obtain a refined culture medium formula; according to the adaptability enhancement scheme, a machine learning optimization method is adopted to optimize strain culture condition parameters, and stable culture condition setting is obtained; and setting and generating a strain culture protocol sequence through the stable culture condition, and determining a subsequent raw material decomposition efficiency index to obtain an overall fermentation optimization path.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for optimizing microbial fermentation culture media for agricultural feed production. Background Technology

[0002] In the agricultural and livestock sector, improvements in feed formulation directly impact livestock growth efficiency and health, serving as a core element for enhancing farming profitability and sustainable development. High-quality feed not only reduces farming costs but also minimizes environmental pollution. Therefore, optimizing feed composition and function through technological innovation has become a crucial issue that the industry urgently needs to address. In particular, the development of fermented feeds, utilizing microbial fermentation technology to improve nutritional value and digestibility, is widely considered a future direction for development.

[0003] However, a common problem in the current preparation of fermented feed is the lack of targeted cultivation and application of microbial strains, leading to unstable fermentation results. Many methods fail to fully consider the metabolic characteristics and environmental adaptability of the strains during cultivation, often resulting in insufficient strain activity or fermentation products that do not meet expectations. This problem not only affects the conversion efficiency of nutrients in the feed but may also lead to resource waste and increased production costs. At a deeper level, existing technologies often neglect the metabolic regulation of the strains during the cultivation stage, failing to provide a good starting point for the subsequent fermentation process.

[0004] Focusing on the technical challenges, the design of the culture medium for probiotics in fermented feed has become a key bottleneck. The composition of the culture medium directly determines the growth state and metabolic direction of the strains. If the needs of the strains cannot be precisely matched, it will be difficult for them to develop specific metabolic capabilities in the early stages. Furthermore, this lack of matching will affect the efficiency of the strains in decomposing feed ingredients during fermentation. For example, some strains may not be able to effectively decompose proteins, thus failing to generate bioactive substances beneficial to livestock growth. Taking practical application as an example, in pig feed fermentation, if the strains fail to develop the ability to decompose specific proteins beforehand, the nutrients in the fermented feed may not be fully absorbed by the pigs, leading to slow growth or frequent health problems.

[0005] Therefore, how to regulate the metabolic direction of microorganisms by optimizing the culture medium composition during the microbial culture stage and ensure that they can efficiently decompose feed ingredients in subsequent fermentation has become a key issue in the improvement of agricultural and livestock feed formulations. Summary of the Invention

[0006] This invention provides a method for optimizing microbial fermentation culture medium for agricultural feed production, mainly comprising: Data on feed ingredient composition and metabolic characteristics of microbial strains are acquired. A preliminary matching index is determined by comparing the feed ingredient composition data with the metabolic characteristics, resulting in a list of potential decomposition requirements of the strains for the raw materials. Based on this list, the strain's metabolic pathways are grouped and calibrated to obtain an optimized set of metabolic pathways. Nutrient element combinations are selected from a resource database using this optimized set of metabolic pathways to determine element ratios, resulting in an initial draft culture medium formulation. A predictive simulation method is used to predict the strain's growth status based on the initial draft culture medium formulation. If the predicted growth activity is insufficient, the element ratios are iteratively adjusted to obtain a refined culture medium formulation. Key nutrient element characteristics are extracted from the refined culture medium formulation, and fermentation environment adaptability indicators are determined by comparing these characteristics with historical fermentation data, resulting in an adaptability enhancement scheme. Based on this adaptability enhancement scheme, machine learning optimization methods are used to optimize the strain's culture condition parameters, resulting in stable culture condition settings. A strain culture protocol sequence is generated using these stable culture condition settings to determine subsequent raw material decomposition efficiency indicators, resulting in an overall optimized fermentation path.

[0007] Furthermore, the step of determining a preliminary matching index by comparing the feed ingredient composition data with the metabolic characteristic parameters to obtain a list of potential decomposition requirements of the microbial strain for the raw materials includes: extracting the metabolic characteristic parameters of the microbial strain from a preset database; comparing each component in the feed ingredient composition data with the metabolic characteristic parameters one by one; calculating the degree of matching between each component and the metabolic characteristic parameters; generating a preliminary matching index; if the preliminary matching index is lower than a preset threshold, marking the corresponding component as a high decomposition requirement component; and summarizing all high decomposition requirement components to form a list of potential decomposition requirements of the microbial strain for the raw materials.

[0008] Furthermore, the step of grouping and calibrating the bacterial metabolic pathways according to the potential decomposition demand list to obtain an optimized set of metabolic pathways includes: using a grouping algorithm to group the bacterial metabolic pathways corresponding to the potential decomposition demand list to obtain multiple metabolic pathway groups; judging the degree of path deviation within each metabolic pathway group; if the degree of deviation exceeds a preset range, adjusting the corresponding path parameters to reduce the deviation; and calibrating all metabolic pathway groups through multiple adjustments to obtain an optimized set of metabolic pathways.

[0009] Furthermore, the step of selecting nutrient element combinations from the resource information database through the optimized metabolic pathway set, determining the element ratio configuration, and obtaining an initial culture medium formula draft includes: acquiring data on multiple nutrient elements in the resource information database, selecting nutrient elements that match the metabolic pathways according to the optimized metabolic pathway set, combining the selected nutrient elements and calculating the proportional relationship between each element, determining the element ratio configuration, and generating an initial culture medium formula draft.

[0010] Furthermore, the method of predicting the growth status of the strain using a predictive simulation method for the initial culture medium formula draft, and iteratively adjusting the element ratios if the predicted growth activity is insufficient to obtain a refined culture medium formula, includes: inputting the initial culture medium formula draft into a metabolic model for simulation, obtaining the predicted growth status of the strain, determining whether the predicted growth activity meets the preset standard, and if the predicted growth activity is insufficient, adjusting the element ratios and re-inputting them into the metabolic model for simulation, and iterating and adjusting multiple times until the predicted growth activity meets the standard to obtain a refined culture medium formula.

[0011] Furthermore, the step of extracting key nutrient element characteristics from the refined culture medium formula, determining fermentation environment adaptability indicators by comparing the key nutrient element characteristics with historical fermentation data, and obtaining an adaptability enhancement scheme includes: extracting key nutrient element characteristics from the refined culture medium formula, obtaining environmental adaptation records from historical fermentation data, comparing the key nutrient element characteristics with corresponding characteristics in historical fermentation data, calculating fermentation environment adaptability indicators, generating targeted adjustment measures based on the adaptability indicators, and forming an adaptability enhancement scheme.

[0012] Furthermore, the step of optimizing the strain culture condition parameters using machine learning optimization methods according to the adaptive enhancement scheme to obtain stable culture condition settings includes: inputting adaptive enhancement scheme data into the machine learning model for training, outputting culture condition parameter adjustment values, determining whether the adjustment values ​​cause a shift in the metabolic direction, and if so, rolling back to the previous parameter set and continuing training until the adjustment values ​​no longer cause a shift in the metabolic direction, thereby obtaining stable culture condition settings.

[0013] Furthermore, the step of generating a strain culture protocol sequence through the stable culture conditions, determining the subsequent raw material decomposition efficiency index, and obtaining the overall fermentation optimization path includes: generating a strain culture protocol sequence according to the stable culture conditions, extracting fermentation start point data from the protocol sequence, calculating the subsequent raw material decomposition efficiency index in combination with the stable culture conditions, and summarizing all indicators to form the overall fermentation optimization path.

[0014] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method for generating optimized fermentation pathways based on the matching of feed ingredients and microbial strain metabolic characteristics. Addressing the complex relationship between feed ingredient decomposition efficiency and strain culture condition adaptability, it innovatively constructs a comprehensive solution from initial matching to optimized pathways by integrating feed ingredient composition data and strain metabolic characteristic parameters. The invention first determines potential decomposition needs by comparing feed ingredient data and metabolic parameters, then optimizes the metabolic pathway using a grouping algorithm. Next, it screens nutrient element combinations, iteratively adjusts the culture medium formula to enhance strain growth activity, and simultaneously optimizes culture conditions using historical fermentation data and machine learning methods to ensure stable metabolic direction, ultimately generating a highly efficient optimized fermentation pathway. The core of this invention lies in achieving a dual improvement in feed ingredient decomposition efficiency and strain adaptability through multi-level parameter calibration and predictive simulation. This provides precise and operable technical support for the feed fermentation field, significantly improving resource utilization and production efficiency in the fermentation process. Attached Figure Description

[0015] Figure 1 This is a flowchart of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0016] Figure 2 This is a schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0017] Figure 3 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0018] Figure 4 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0019] Figure 5 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0020] Figure 6 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0021] Figure 7 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0022] Figure 8 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0023] Figure 9 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0024] Figure 10 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0025] Figure 11 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0026] Figure 12 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0027] Figure 13 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention.

[0028] Figure 14 This is another schematic diagram of a method for optimizing a microbial fermentation culture medium for agricultural feed production according to the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] like Figures 1-14 This embodiment of a method for optimizing a microbial fermentation culture medium for agricultural feed production may specifically include: S101. Obtain feed ingredient composition data and microbial strain metabolic characteristic parameters. By comparing the feed ingredient composition data with the metabolic characteristic parameters, determine the preliminary matching degree index and obtain a list of potential decomposition requirements of the strain for the raw materials.

[0031] In one embodiment, feed ingredient composition data and microbial strain metabolic characteristic parameters are obtained. A preliminary matching index is determined by comparing the feed ingredient composition data and the metabolic characteristic parameters to obtain a list of potential decomposition requirements of the strain for the raw materials. Specifically, this includes step S1, obtaining feed ingredient composition data and microbial strain metabolic characteristic parameters.

[0032] Step S11: Extract the composition data of feed ingredients from a preset database, including the proportions of components such as carbohydrates, proteins and cellulose.

[0033] Step S12: Extract metabolic characteristic parameters of microbial strains from the same database, including enzyme activity levels and substrate preference data.

[0034] For example, these acquisition steps ensure that the data source is consistent, providing a reliable basis for subsequent comparisons and helping to improve matching accuracy.

[0035] In one embodiment, step S2 involves determining a preliminary matching index by comparing the feed ingredient composition data with the metabolic characteristic parameters.

[0036] Step S21: Convert the feed ingredient composition data into a numerical vector, for example, the proportion of carbohydrates is 0.4 and the proportion of protein is 0.3.

[0037] Step S22: Convert the metabolic characteristic parameters of the microbial strain into corresponding vectors, for example, the enzyme’s efficiency in decomposing carbohydrates is 0.8 and in decomposing proteins is 0.6.

[0038] Step S23: Use the cosine similarity algorithm to calculate the similarity between two vectors. This algorithm is defined as the dot product of two vectors divided by the product of their respective magnitudes to obtain a preliminary matching index. For example, a similarity value of 0.75 indicates a moderate match.

[0039] Step S24: If the matching score is higher than 0.7, it is marked as a high potential match.

[0040] Specifically, this comparison method is implemented using the cosine similarity algorithm, which is often used in vector space models to evaluate similarity. Here, it is applied to the matching of feed ingredients and microbial metabolism, which can quantify decomposition potential, avoid subjective judgment, and help optimize the efficiency of the feed fermentation process.

[0041] For example, when processing corn-based feed, the raw material vector is carbohydrate 0.6 and cellulose 0.2, and the microbial parameter vector is metabolic efficiency of carbohydrate 0.9 and cellulose 0.5. The calculated cosine similarity is 0.82, indicating a high match, which helps to improve the decomposition rate.

[0042] For example, in another scenario, for soybean-based feed, the raw material vector is adjusted to 0.5 for protein and 0.3 for carbohydrates, and the strain parameter vector is 0.7 for protein and 0.8 for carbohydrates. The similarity calculation is 0.76, which is also higher than the threshold, supporting list generation. This multi-scenario coverage ensures the robustness of the method under different feed types and is beneficial for adaptability in practical applications.

[0043] In one embodiment, step S3 involves obtaining a list of potential decomposition requirements of the microbial strain for raw materials.

[0044] Step S31: Sort the bacterial strains according to the preliminary matching degree index, for example, arrange them from high to low matching degree.

[0045] Step S32: Generate a demand list item for each strain, including the raw material components to be decomposed and the expected efficiency, such as a high demand for carbohydrates.

[0046] Step S33: Compile the results into a list, ensuring that each item is linked back to the matching metric.

[0047] Specifically, this list generation is based on sorting and summarizing logic, forming a specific chain from matching degree to demand, which is beneficial for subsequent fermentation path planning.

[0048] For example, in a corn feed scenario, the list shows that strain A has the primary need for carbohydrate decomposition, with an expected efficiency of 0.9, supporting the optimization of culture medium design.

[0049] For example, in soybean feed, the list highlights the protein requirement with an efficiency of 0.7, which connects with the previous similarity calculation to ensure logical consistency and is beneficial to the continuity of overall fermentation optimization.

[0050] S102. Group and calibrate the bacterial metabolic pathways according to the potential decomposition requirement list to obtain an optimized set of metabolic pathways; screen nutrient element combinations from the resource information database using the optimized set of metabolic pathways, determine the element ratio configuration, and obtain an initial culture medium formula draft.

[0051] In one embodiment, step S1 involves grouping and calibrating the bacterial metabolic pathways according to the potential decomposition demand list to obtain an optimized set of metabolic pathways. Specifically, step S11 involves grouping the bacterial metabolic pathways using a grouping algorithm. The grouping algorithm refers to a standard method based on k-means clustering, which divides the metabolic pathways into several groups based on similarity. If the grouping results show a deviation in the metabolic pathways, the path parameters are adjusted to calibrate the deviation. Here, the metabolic pathway deviation refers to the difference between the enzyme activity parameter in the path and the decomposition demand value in the demand list exceeding a preset threshold, for example, when the threshold is 0.5. The difference is reduced by iteratively adjusting parameter values ​​such as the enzyme concentration coefficient to obtain the optimized set of metabolic pathways.

[0052] Step S12: After grouping, calculate the average deviation value for each metabolic pathway group. If the average deviation value is greater than the threshold, roll back the grouping and readjust the parameters to ensure the stability of the pathway set.

[0053] Specifically, this implementation extends the process mentioned in the technical disclosure, which involves grouping the metabolic pathways of microorganisms according to a list of potential decomposition needs using a grouping algorithm and adjusting parameters based on deviations. The specific process of the grouping algorithm is to represent the metabolic pathways as vectors, calculate the Euclidean distance between the vectors for clustering, and use the gradient descent method to minimize the difference function when calibrating deviations. This approach can improve the efficiency of microorganisms in decomposing feed ingredients because the optimized pathways can better match the raw material composition data.

[0054] In one embodiment, step S2, which involves selecting nutrient element combinations from the resource information database using the optimized metabolic pathway set, determining the element ratio configuration, and obtaining an initial culture medium formulation draft, specifically includes step S21, extracting nutrient elements, such as carbon sources and nitrogen sources, that match key metabolic nodes in the pathway set from the resource information database, and selecting matching combinations.

[0055] Step S22: Calculate the element ratios based on the screening combination, for example, determine a carbon-nitrogen ratio of 5:1 using linear programming, and generate an initial culture medium formulation draft.

[0056] Specifically, this implementation method is based on the process of obtaining resource information database and screening nutrient element combinations that match the path set in the technical disclosure document. It is simply extended to direct matching and ratio calculation to avoid complex adjustments, thus ensuring that the draft formula initially meets the needs of the strain.

[0057] For example, in a scenario where the feed ingredients are corn and soybean meal, for lactic acid bacteria strains, the potential decomposition demand list in step S1 shows that starch and protein need to be decomposed. The grouping algorithm divides the metabolic pathway into a glycolysis group and an amino acid metabolism group. If the deviation shows that the enzyme activity of the glycolysis pathway is low, the parameters are adjusted to increase the glucose transporter coefficient to 1.2 to obtain an optimized set, which can improve the decomposition efficiency by 20%.

[0058] In one embodiment, for Bacillus subtilis, the raw material is wheat bran. If the deviation is large after grouping in step S1, the cellulase pathway parameter is adjusted from 0.8 to 1.0 by calibration. The optimized set supports better fiber decomposition and is beneficial to fermentation stability.

[0059] For example, in step S2, the optimized set is used to screen the resource library and select a combination of glucose and yeast extract with a ratio of 60% carbon source and 40% nitrogen source to obtain a draft, which simplifies the subsequent culture process and increases the strain growth rate by 15%.

[0060] In one embodiment, when calibrating the deviation in step S1 for yeast strains and fishmeal raw materials, the parameters are adjusted iteratively three times. After the path set is optimized, the matching degree reaches 95%, which has the beneficial effect of reducing ineffective metabolic waste.

[0061] S103. For the initial culture medium formula draft, a predictive simulation method is used to predict the growth status of the strain. If the predicted growth activity is insufficient, the element ratio is iteratively adjusted to obtain a refined culture medium formula.

[0062] S1 uses a predictive simulation method to predict the growth status of the strain for the initial culture medium formulation draft. If the predicted growth activity is insufficient, the element ratio is iteratively adjusted to obtain a refined culture medium formulation.

[0063] In one embodiment, step S1 uses a predictive simulation method to predict the growth status of the strain for the initial culture medium formulation draft. Specifically, step S11 involves inputting the initial culture medium formulation draft into a metabolic model to run the simulation. The metabolic model is constructed based on flux balance analysis, which solves the metabolic flux distribution of the strain under given nutrient conditions through linear programming to ensure that the simulation reflects the actual growth dynamics.

[0064] Step S12: Calculate growth activity indicators, such as biomass production rate, based on the simulation results. If the rate is lower than a preset threshold, the growth activity is deemed insufficient.

[0065] For example, in a feed ingredient fermentation scenario, for lactic acid bacteria strains, the initial formula contains 5% carbon source, and the simulation shows that the biomass production rate is 0.2 g / L / h, which is lower than the threshold of 0.5 g / L / h, thus triggering an adjustment.

[0066] In one possible implementation, if the predicted growth activity is insufficient in step S1, the element ratio is iteratively adjusted. Specifically, step S13 involves using a gradient descent algorithm to iteratively modify the nutrient element ratio, such as gradually increasing the nitrogen source ratio by 0.1% and resimulating until the growth activity reaches the threshold.

[0067] Step S14: Record the formula changes after each iteration to form an optimization path, and finally output the refined culture medium formula.

[0068] For example, in the application of yeast strains in the decomposition of protein feed, the initial nitrogen source ratio was 2%, which was adjusted to 3.5% after three iterations. The simulated growth rate was 0.6 g / L / h, which met the requirements. This process improved fermentation efficiency and reduced resource waste.

[0069] It should be noted that this iterative adjustment, through multiple simulation cycles, ensures that the formula adapts to the metabolic needs of the microbial strain, which can increase yield by more than 10% in feed production.

[0070] S104. Extract key nutrient element characteristics from the refined culture medium formula, determine fermentation environment adaptability indicators by comparing the key nutrient element characteristics with historical fermentation data, and obtain an adaptability enhancement scheme.

[0071] In one embodiment, key nutrient element characteristics are extracted from the refined culture medium formula, and fermentation environment adaptability indicators are determined by comparing the key nutrient element characteristics with historical fermentation data to obtain an adaptability enhancement scheme. Specifically, step S1 is to extract key nutrient element characteristics from the refined culture medium formula, wherein the refined culture medium formula contains components such as carbon source, nitrogen source and trace elements, and the concentration and type of these elements are directly extracted as characteristics by analyzing the formula data.

[0072] Step S2 involves determining fermentation environment adaptability indicators by comparing the key nutrient element features with historical fermentation data. Specifically, this includes: Step S21, converting the extracted key nutrient element features into standardized vectors, for example, normalizing carbon source concentration and nitrogen source concentration to a numerical range of 0 to 1; Step S22, selecting a matching subset from historical fermentation data, which includes environmental parameters from past fermentation experiments such as temperature and pH, as well as the corresponding microbial growth rate; Step S23, using a cosine similarity algorithm to calculate the similarity between the key nutrient element feature vector and the historical fermentation data vector. The cosine similarity algorithm is obtained by calculating the dot product of two vectors and dividing it by the product of their moduli, and is used to quantify the degree of feature matching; Step S24, quantifying the adaptability indicators based on the similarity results. If the similarity is higher than 0.8, the indicator is considered highly adapted; otherwise, it is considered low adapted, thus forming a list of quantified indicators.

[0073] For example, in the scenario of feed ingredient fermentation, assuming that the carbon source concentration in the refined culture medium formula is 5%, historical fermentation data shows that the strain growth rate reaches 90% at similar concentrations. The cosine similarity is calculated to obtain a similarity of 0.85, which determines the high adaptability index. This helps to predict fermentation stability, avoid strain growth inhibition caused by low adaptability, and is beneficial to improving the overall fermentation efficiency.

[0074] In one embodiment, the comparison process in step S2 can be extended to different microbial species. For example, for lactic acid bacteria, historical fermentation data focuses on growth in an acidic environment, while for yeast, it emphasizes sugar source utilization. By calculating the adaptability index through this distinction, environmental parameters can be optimized for specific species, resulting in higher decomposition efficiency.

[0075] Step S3, obtaining the adaptation enhancement scheme, specifically includes: Step S31, based on the adaptation index determined in Step S2, screening environmental adjustment parameters corresponding to high indices in historical fermentation data, such as increasing the temperature by 2 degrees or adjusting the pH by 0.5 units; Step S32, combining these parameters to generate an enhancement scheme, for example, integrating temperature and pH adjustments into a scheme list; Step S33, verifying the feasibility of the scheme, checking whether the adjusted indexes are improved through a simulated fermentation model. If they are improved, the scheme is confirmed; otherwise, iterate back to Step S31.

[0076] For example, for culture media with low adaptability indicators, enhancement programs include adding buffers to stabilize pH, which can improve the matching of the strain's decomposition requirements for feed ingredients, resulting in a more stable fermentation pathway and higher nutrient utilization, thus helping to reduce resource waste.

[0077] In one embodiment, step S3 can be applied to different raw material compositions, such as corn-based feed versus soybean-based feed. By adjusting the proportion of trace elements through an adaptability enhancement scheme, the scheme can be ensured to be effective in multiple scenarios, thereby improving the versatility of the fermentation environment.

[0078] S105. Based on the aforementioned adaptive enhancement scheme, the strain culture condition parameters are optimized using machine learning optimization methods to obtain stable culture condition settings.

[0079] In one embodiment, step S6 involves using a machine learning optimization method to optimize the bacterial culture condition parameters according to the adaptive enhancement scheme, thereby obtaining stable culture condition settings.

[0080] Step S61: Input the adaptive enhancement scheme data during training and output the adjustment value.

[0081] Specifically, the fermentation environment adaptability indicators in the adaptive enhancement scheme are processed by a neural network model. These indicators are used as input features. The neural network model is a multilayer perceptron, which includes an input layer, a hidden layer, and an output layer. The input layer receives adaptive indicators such as temperature adaptation range and pH adaptation threshold. The hidden layer uses an activation function such as ReLU to perform nonlinear transformation. The output layer generates adjustment values ​​such as temperature adjustment amplitude and pH adjustment step size. These adjustment values ​​are used to optimize culture condition parameters such as culture temperature and stirring speed.

[0082] Step S62: If the adjustment value causes a shift in the metabolic direction, then roll back to the previous parameter set.

[0083] For example, the output adjustment value is applied to the metabolic model simulation. The metabolic model calculates the metabolic flux of the strain based on the Flux Balance Analysis method. If the simulation results show that the metabolic flux deviates from the path specified in the preset decomposition demand list, such as the carbon source utilization path, it is judged as an offset and rolled back to the previous parameter set, such as the parameter combination of initial temperature 25 degrees Celsius and pH 6.5, to ensure stability.

[0084] In one embodiment, after obtaining stable culture condition settings, a strain culture protocol sequence is generated through these settings, but this embodiment focuses on the optimization process itself.

[0085] For example, in a scenario where the feed ingredient is corn stalks, the adaptability enhancement scheme provides a temperature adaptability index of 20-30 degrees Celsius. After inputting into the neural network model, the output adjustment value is to increase the temperature by 2 degrees Celsius. If the simulation shows a shift in the metabolic direction, such as a 10% decrease in glucose metabolic flux, the process is rolled back to ensure that the final stable conditions are set at a temperature of 26 degrees Celsius and a pH of 6.8. This can improve the decomposition efficiency of the strain on the raw materials, resulting in beneficial effects such as a 15% increase in fermentation yield.

[0086] In one embodiment, for the optimization of lactic acid bacteria strains, the input adaptive scheme data includes oxygen adaptation indexes, the neural network model outputs adjustment values ​​such as a 0.5% reduction in oxygen concentration, and after determining that there is no deviation, a stable setting is obtained, such as a stirring speed of 200 rpm under anaerobic conditions, which is beneficial to enhancing the growth activity of the strains.

[0087] For example, in the application of yeast strains, if the adjustment value causes a deviation, such as a deviation from the ethanol metabolism pathway, the rollback process uses the nutrient element ratio of the previous parameter set to ensure that the stable culture conditions are adapted to historical fermentation data, resulting in technical effects such as reducing the accumulation of metabolic byproducts and improving the reliability of the overall fermentation optimization pathway.

[0088] S106. By setting the stable culture conditions, a strain culture protocol sequence is generated, and the subsequent raw material decomposition efficiency index is determined to obtain the overall fermentation optimization path.

[0089] In one embodiment, a microbial culture protocol sequence is generated by setting stable culture conditions to determine the subsequent raw material decomposition efficiency index, thereby obtaining an overall optimized fermentation path. Specifically, this includes step S1, constructing a time-sequential operation sequence based on parameters such as temperature, pH, and stirring speed in the stable culture condition settings. This sequence defines the duration and condition switching points of each stage from inoculation to the end of fermentation, for example, gradually increasing the temperature from an initial 28 degrees Celsius to 32 degrees Celsius to match the peak growth of the microbial strain.

[0090] Step S11: Extract fermentation start point data from the sequence, including the concentration of the starting strain and the amount of raw materials input. Calculate the preliminary decomposition rate using these data. For example, when the starting concentration is 10^6 CFU / mL, the raw material decomposition rate is expected to be 5% per day.

[0091] Step S12: Based on the extracted data, simulate the raw material decomposition process in the subsequent stages and determine whether the decomposition efficiency reaches the preset threshold. If the efficiency is lower than 80%, it is marked as a point that needs optimization.

[0092] In one possible implementation, the generation process of step S1 uses the aforementioned optimized set of metabolic pathways as a basis to ensure that the protocol sequence matches the metabolic characteristics of the strain. For example, in the scenario where lactic acid bacteria are used in corn feed fermentation, the sequence includes an acidification stage to improve decomposition efficiency. The beneficial effect is to improve the utilization rate of feed nutrients and reduce waste.

[0093] Step S2: Determine the subsequent raw material decomposition efficiency index. Specifically, calculate the average decomposition efficiency using stage data in the protocol sequence. For example, compare the starting point data with the intermediate sampling point data to obtain an efficiency index of 85%. This index reflects the overall utilization of raw materials by the strain.

[0094] Step S21: Use historical fermentation data as a reference to quantify efficiency indicators, such as comparing the efficiency values ​​of similar corn feed fermentation in the past, and adjust the current indicators to correct deviations.

[0095] Step S22: If the efficiency index shows fluctuations, apply a linear regression method to fit the trend line and predict a stable efficiency value. For example, the fitting result shows that the efficiency increases from an initial 70% to a final 90%.

[0096] For example, in the application of protein raw materials such as soybean meal fermentation, the efficiency index determination in step S2 can identify the bottleneck of nitrogen source decomposition, which has the beneficial effect of guiding subsequent adjustments and optimizing fermentation output.

[0097] Step S3 yields the overall fermentation optimization path, specifically integrating protocol sequences and efficiency indicators to form a complete path diagram from raw material preparation to product collection. For example, the path includes three modules: pretreatment, fermentation, and posttreatment, with efficiency indicators labeled for each module.

[0098] Step S31: Verify the continuity of the path by checking the data flow between modules to ensure there are no information silos. For example, efficiency indicators are output from the fermentation module and then used for adjustments in the post-processing module.

[0099] Step S32: Output the optimized path as a guide document for actual feed production. For example, in a grain feed scenario, this path can improve decomposition efficiency by 15%, and the beneficial effect is to enhance the stability and repeatability of the fermentation process.

[0100] In one embodiment, when generating the protocol sequence in step S1 for the application of lactic acid bacteria on different feed ingredients, the sequence can be adjusted according to the type of raw material. For example, for high-fiber raw materials, the stirring stage can be extended to improve decomposition. Step S2 calculates the efficiency index. For example, if the raw material is wheat bran, the initial decomposition efficiency is 75%, which is subsequently increased to 92% through path optimization. This reflects the logical chain from condition setting to path generation. The beneficial effect is that it adapts to diverse feed scenarios and improves the targeting of microbial fermentation.

[0101] For example, in the scenario where yeast is used in fishmeal feed, the overall fermentation optimization path in step S3 integrates efficiency indicators and forms a chain from the starting point to the end.

[0102] Specifically, the pathway shows that the decomposition efficiency peaks at pH 5.5, with the beneficial effects of reduced energy consumption and improved product quality.

[0103] In one possible implementation, the efficiency index determination in step S2 can be extended to multi-strain mixed fermentation. For example, comparing the decomposition rate of a single strain with that of a mixture yields an index difference of 10%, which supports the path optimization in step S3. The beneficial effect is to achieve more efficient feed processing.

[0104] S107. The step of determining a preliminary matching index by comparing the feed ingredient composition data with the metabolic characteristic parameters to obtain a list of potential decomposition requirements of the microbial strain for the raw materials includes: extracting the metabolic characteristic parameters of the microbial strain from a preset database; comparing each component in the feed ingredient composition data with the metabolic characteristic parameters one by one; calculating the degree of matching between each component and the metabolic characteristic parameters; generating a preliminary matching index; if the preliminary matching index is lower than a preset threshold, marking the corresponding component as a high decomposition requirement component; and summarizing all high decomposition requirement components to form a list of potential decomposition requirements of the microbial strain for the raw materials.

[0105] In one embodiment, a preliminary matching index is determined by comparing the feed ingredient composition data with the metabolic characteristic parameters to obtain a list of potential decomposition requirements of the microbial strain for the raw materials, including the following steps.

[0106] Step S1: Extract the metabolic characteristic parameters of the microbial strain from a preset database.

[0107] Specifically, the pre-set database stores relevant data on various microbial species, such as enzyme activity levels and substrate preference parameters of lactic acid bacteria or yeast. These parameters can be directly extracted by querying the species identifier to ensure the accuracy of subsequent comparisons.

[0108] Step S2 involves comparing each component and metabolic characteristic parameter in the feed ingredient composition data one by one.

[0109] In one possible implementation, the feed ingredient composition data includes components such as carbohydrates, proteins, and fiber, and the chemical structural characteristics of each component are compared with the metabolic characteristic parameters of the strain, such as enzyme affinity.

[0110] Step S3: Calculate the degree of matching between each component and its metabolic characteristic parameters.

[0111] For example, the matching degree is calculated using the cosine similarity method, where the vector representation of the raw material component is multiplied by the metabolic parameter vector and normalized to obtain a value between 0 and 1. For example, for the carbohydrate component, if its vector is (0.8, 0.2, 0.5) and the metabolic parameter vector is (0.7, 0.3, 0.4), the similarity is approximately 0.95. This reflects the potential utilization efficiency of the strain for the component, which is beneficial for identifying efficient decomposition pathways and improving the overall yield of feed fermentation.

[0112] Step S4: Generate preliminary matching index.

[0113] It should be noted that the preliminary matching index is derived from the weighted average of the matching degree of all components. For example, the weight is based on the proportion of the component in the raw material, ensuring that the index fully represents the overall compatibility between the strain and the raw material.

[0114] Step S5: If the initial matching degree index is lower than the preset threshold, the corresponding component is marked as a component with high decomposition demand.

[0115] In one embodiment, a preset threshold is set to 0.6. If the matching degree of a certain component is 0.4, it is marked as a component with high decomposition requirements. For example, in corn-based feed, fiber components with low matching degree are marked. This helps to optimize the culture medium in a targeted manner to enhance the ability of the strain to metabolize difficult-to-decompose components, thereby improving fermentation efficiency and reducing waste generation.

[0116] Specifically, for different feed scenarios, such as soybean protein raw materials, if the protein composition matching degree is 0.5, which is lower than the threshold, it is marked as having high decomposition requirements. The beneficial effect is that it can guide the subsequent addition of specific enzyme catalysts to improve the decomposition rate.

[0117] In one possible implementation, for grain raw materials such as wheat, if the starch matching degree is 0.3, the pH value can be adjusted to assist decomposition after labeling. Multiple aspects support the view that this labeling mechanism can optimize feed utilization from the perspective of metabolic compatibility, forming a consistent argument for improved fermentation.

[0118] Step S6: Summarize all components with high decomposition requirements to form a list of potential decomposition requirements of the strain for the raw materials.

[0119] For example, a list of all labeled components is compiled, such as fiber and high molecular weight protein, and the intensity of demand is indicated. This directly supports the subsequent culture medium design, ensuring that the potential decomposition needs of the strain in feed ingredients are met.

[0120] S108. The step of grouping and calibrating the bacterial metabolic pathways according to the potential decomposition demand list to obtain an optimized set of metabolic pathways includes: using a grouping algorithm to group the bacterial metabolic pathways corresponding to the potential decomposition demand list to obtain multiple metabolic pathway groups; judging the degree of path deviation within each metabolic pathway group; if the degree of deviation exceeds a preset range, adjusting the corresponding path parameters to reduce the deviation; and calibrating all metabolic pathway groups through multiple adjustments to obtain an optimized set of metabolic pathways.

[0121] In one embodiment, the step of grouping and calibrating the bacterial metabolic pathways according to the potential decomposition requirement list to obtain an optimized set of metabolic pathways specifically includes the following steps.

[0122] Step S21: The metabolic pathways of the bacterial species corresponding to the potential decomposition demand list are grouped using a grouping algorithm to obtain multiple metabolic pathway groups.

[0123] Specifically, the potential decomposition demand list includes the strain's decomposition requirements for feed ingredients, such as its preference for carbon and nitrogen sources, which are then mapped onto metabolic pathways. The grouping algorithm employs k-means clustering, which groups similar paths into the same group by calculating the Euclidean distance between path feature vectors.

[0124] For example, clustering the glycolysis and amino acid metabolism pathways of lactic acid bacteria yields carbon and nitrogen metabolomics. This grouping helps identify similarities between pathways and improves the efficiency of subsequent calibration.

[0125] In one embodiment, for the decomposition of feed ingredients such as corn flour, if the potential decomposition demand list shows that lactic acid bacteria need to efficiently utilize starch, then the grouping algorithm will group the starch hydrolysis path and the downstream glycolysis path into one group, with 5 paths within the group and a distance threshold of 0.2 between groups to ensure compact grouping.

[0126] Step S22: Determine the degree of path deviation within each metabolic pathway group.

[0127] Specifically, the degree of path deviation is quantified by calculating the variance of the path parameters within a group. The variance formula is the sum of squares of the difference between each path parameter and the group mean, divided by the number of paths. If the variance is greater than a preset range, such as 0.1, it indicates a significant deviation.

[0128] For example, in carbon metabolomics, if the enzyme activity parameter of a pathway is 1.5, while the group mean is 1.0, the variance calculation result is 0.15, which is outside the range.

[0129] In one embodiment, for the application of yeast on wheat bran raw materials, the deviation of the nitrogen metabolome was judged. The intra-group path parameters included amino acid transport rate. The variance was calculated to be 0.12, which exceeded the 0.1 threshold. This reflects the inconsistency between paths, which may lead to uneven fermentation.

[0130] Step S23: If the deviation exceeds the preset range, adjust the corresponding path parameters to reduce the deviation.

[0131] Specifically, the adjustment employs the gradient descent method, iteratively updating parameter values ​​to minimize the deviation. Gradient descent updates the parameters along the negative gradient direction by calculating the partial derivative of the deviation with respect to the parameters, with a step size of 0.01.

[0132] For example, the enzyme concentration parameter for the deviation path was adjusted from 1.2 to 1.05, reducing the variance to 0.08. This adjustment ensures path consistency and enhances the strain's potential for decomposing raw materials.

[0133] In one embodiment, in the scenario of Bacillus decomposing soybean meal raw materials, the carbon metabolome deviation was 0.13. By adjusting path parameters such as glucose uptake rate, the deviation was reduced from the initial 2.0 to 1.8. After three iterations, the deviation was reduced to below 0.09. The beneficial effect was to improve the overall metabolic efficiency and reduce fermentation waste.

[0134] In one embodiment, when the same Bacillus is applied to fishmeal raw materials, the nitrogen metabolome deviation is 0.14. The amino acid synthesis parameters are adjusted from 0.9 to 1.1. After 4 iterations, the deviation is reduced to 0.07. This optimizes the pathway balance, enhances the protein breakdown capacity, and avoids nutrient waste.

[0135] Step S24: By adjusting and calibrating all metabolic pathway groups multiple times, an optimized set of metabolic pathways is obtained.

[0136] Specifically, multiple adjustments refer to repeatedly executing step S23 on all groups until the deviation of all groups is lower than the preset range, forming a complete set.

[0137] For example, a total of 10 adjustments were made to the three metabolic pathway groups, resulting in a final set containing 20 calibrated pathways. This calibration process ensured the stability of the pathway set for subsequent culture medium design.

[0138] In one embodiment, in the optimization of mixed feed ingredients by lactic acid bacteria, the three groups of carbon, nitrogen and vitamins were adjusted, and a total of 15 iterations were performed to obtain a path set. The average path deviation in the set was 0.05, which improved the fermentation yield by 20%. The beneficial effect is to achieve efficient utilization of raw materials and reduce production costs.

[0139] In one embodiment, for the application of yeast on a single grain raw material, four groups were adjusted, and a total of 12 iterations were performed. The path set bias was reduced to 0.04, which enhanced the adaptability and increased the acid production by 15%, demonstrating the robustness of calibration under different raw material scenarios.

[0140] For example, in the field of feed fermentation, the grouping in step S21 uses k-means clustering to divide the 10 pathways of lactic acid bacteria into 2 groups with an intra-group distance of 0.15, ensuring the preliminary classification of the pathways and facilitating the rapid identification of key metabolic modules.

[0141] For example, the deviation judgment in step S22 uses variance calculation. In the Bacillus group, the variance of 0.11 exceeds 0.1, revealing the problem of uneven path, which helps to optimize in a targeted manner and avoid fermentation failure.

[0142] For example, the adjustment in step S23 uses gradient descent to iteratively update the parameters, such as enzyme activity from 1.3 to 1.1 and deviation from 0.13 to 0.08, which improves path consistency and enhances the growth stability of the strain.

[0143] For example, the multiple adjustments in step S24 cover all groups, such as performing 8 iterations on the 4 groups of yeast, with a final deviation of 0.06. This forms a reliable set of paths, supports efficient fermentation, and reduces energy consumption.

[0144] In one possible implementation, the application of the above grouping and calibration process to corn flour feed showed that optimizing the path set improved decomposition efficiency by 15%, demonstrating its practical value in the feed industry.

[0145] In one possible implementation, for soybean meal feedstock, iterative calibration of the path set reduces nutrient loss due to bias, increases output by 10%, and highlights the sustainability of the technology's effectiveness.

[0146] S109. The step of selecting nutrient element combinations from the resource information database through the optimized metabolic pathway set, determining the element ratio configuration, and obtaining an initial culture medium formula draft includes: acquiring multiple nutrient element data from the resource information database, selecting nutrient elements that match the metabolic pathways according to the optimized metabolic pathway set, combining the selected nutrient elements and calculating the ratio relationship between each element, determining the element ratio configuration, and generating an initial culture medium formula draft.

[0147] In one embodiment, the combination of nutrient elements is screened from the resource information database through the optimized set of metabolic pathways, and the element ratio configuration is determined to obtain an initial culture medium formula draft. Specifically, the process includes step S1, obtaining data on multiple nutrient elements from the resource information database.

[0148] Step S11: Extract nutrient element data such as carbon source, nitrogen source, and trace elements from the resource information database. These data include element type, concentration range, and bioavailability indicators.

[0149] Step S12: Organize the extracted data into a structured list for easy filtering later.

[0150] For example, in a scenario where the feed ingredient is corn flour, the resource information database provides data on carbon sources such as glucose, with a concentration range of 10-50 g / L and a bioavailability index of 85%. The data obtained in this way directly supports subsequent screening and ensures the compatibility of nutrient elements with the metabolic pathways of microbial strains.

[0151] In one embodiment, step S2 involves screening for nutrients that match the metabolic pathways based on the optimized set of metabolic pathways.

[0152] Step S21: Analyze the optimized metabolic pathway set and extract the key enzymatic reactions and required substrate types involved in the pathway.

[0153] Step S211: Decompose the set of pathways into individual metabolic pathways, and label each pathway with the required nutrients, such as the carbohydrate breakdown pathway which requires glucose as a substrate.

[0154] Step S212: Compare the path label with the nutrient element data in the resource information database, and calculate the matching score. The matching score is obtained by the substrate type overlap rate and enzyme affinity parameter. For example, if the overlap rate is greater than 70% and the affinity is greater than 0.5, it is considered to be a match.

[0155] Step S22: Select nutrient elements with high matching scores from the resource information database to form a preliminary element set.

[0156] Step S221: For each metabolic pathway, prioritize selecting the 3-5 elements with the highest matching degree to ensure coverage of pathway requirements.

[0157] Step S222: If the element set coverage is less than 90%, then supplement with similar elements, such as replacing some glucose with fructose.

[0158] Step S23: Verify the metabolic compatibility of the screened elements by confirming that the elements do not inhibit the pathway through a simulated path activation test.

[0159] For example, in the set of metabolic pathways of lactic acid bacteria strains, the pathways involve lactic acid fermentation and require carbon sources such as lactose. During screening, the pathway parameters are compared with lactose data in the library, and the matching degree score is 92%. In this way, the selected elements can improve the decomposition efficiency of feed raw materials by the strains and are beneficial to the stability of the fermentation process.

[0160] In one embodiment, for the Bacillus subtilis scenario, the metabolic pathway set includes amino acid synthesis pathways, and nitrogen sources such as ammonium salts are screened with a matching score of 88%. Through this screening, the culture medium can better support the growth of the strain and reduce metabolic deviations.

[0161] For example, in the application of yeast, the pathway set focuses on glycolysis, screening for carbon sources such as maltose with a matching degree of 95%, which has the beneficial effect of improving the adaptability of the strain to complex feed ingredients.

[0162] In one embodiment, step S3 involves combining the screened nutrient elements and calculating the proportional relationship between each element.

[0163] Step S31: Group the selected nutrients into primary elements and secondary elements. The primary elements are based on the core requirements of the pathway, and the secondary elements are based on the auxiliary reactions of the pathway.

[0164] Step S32: Calculate the proportional relationship using the element demand coefficient formula, where the demand coefficient is the element consumption rate in the path divided by the total consumption rate. For example, the carbon source coefficient is 0.6 and the nitrogen source coefficient is 0.3.

[0165] Step S33: Adjust the proportions to balance the overall ratio, ensuring the sum is 1.

[0166] For example, in the microbial culture of corn flour feed, glucose and ammonium salts are combined in a calculated ratio of 6:3. This ratio optimizes resource utilization and results in higher fermentation yield.

[0167] In one embodiment, step S4 involves determining the element ratio configuration.

[0168] Step S41: Based on the calculated proportional relationship, set the specific concentration value of each element, such as the carbon source concentration of 20 g / L.

[0169] Step S42: Verify the feasibility of the configuration by checking whether it supports complete metabolism through path simulation.

[0170] For example, after configuration, the ratio of carbon source to nitrogen source is 2:1, and the concentrations are 30 g / L and 15 g / L, respectively, which is beneficial to the rapid proliferation of the strain.

[0171] In one embodiment, step S5 involves generating an initial culture medium formulation draft.

[0172] Step S51: Convert the determined element ratio configuration into a formula list, including element names and concentrations.

[0173] Step S52: Add standard solvent and pH adjuster to form a complete draft.

[0174] For example, the generated draft includes 20 g / L glucose, 10 g / L ammonium salt, and pH 6.5. This draft can be directly used for subsequent prediction simulations to improve the efficiency of the overall fermentation optimization pathway.

[0175] S1010. The method of predicting the growth status of the strain using a predictive simulation method for the initial culture medium formula draft, and iteratively adjusting the element ratios if the predicted growth activity is insufficient, to obtain a refined culture medium formula, includes: inputting the initial culture medium formula draft into a metabolic model for simulation, obtaining the predicted growth status of the strain, determining whether the predicted growth activity meets the preset standard, and if the predicted growth activity is insufficient, adjusting the element ratios and re-inputting them into the metabolic model for simulation, and iterating and adjusting multiple times until the predicted growth activity meets the standard to obtain a refined culture medium formula.

[0176] In one embodiment, a predictive simulation method is used to predict the growth status of the strain for the initial culture medium formulation draft. If the predicted growth activity is insufficient, the element ratio is iteratively adjusted to obtain a refined culture medium formulation, including the following steps.

[0177] Step S1: Input the initial culture medium formula draft into the metabolic model for simulation and obtain the predicted results of the bacterial growth status.

[0178] Specifically, the nutrient element combination and ratio data in the initial culture medium formulation draft are used as input parameters and imported into a pre-constructed metabolic model. This model simulates the growth process based on the metabolic pathway of the strain and outputs the predicted growth rate and activity index after running.

[0179] Step S2: Determine whether the predicted growth activity meets the preset standard.

[0180] For example, the growth activity value is compared with a preset threshold based on the output result. If the threshold is 0.5 per hour growth rate, it is considered to meet the standard if it is higher than this value, otherwise it will enter the adjustment stage.

[0181] Step S3: If the predicted growth activity is insufficient, adjust the element ratio and re-enter it into the metabolic model for simulation.

[0182] In one possible implementation, the specific cause of insufficient activity is first identified, such as metabolic obstruction due to an excessively low carbon source ratio. Then, the carbon source ratio is increased in a targeted manner, for example, from the initial 20% to 25%. The adjusted formula is then re-input into the model to run the simulation and obtain new prediction results.

[0183] Step S31: Analyze the specific nutrient element deviations that are insufficient in activity.

[0184] For example, by comparing the actual metabolic flux with the ideal value through the model output log, the deviation amount, such as insufficient nitrogen source, can be determined.

[0185] Step S32: Adjust the proportion of the corresponding elements based on the deviation.

[0186] Specifically, if the nitrogen source deviation is negative, the nitrogen source ratio will be increased by a proportionality factor such as 1.1 times, and the draft formulation will be updated.

[0187] Step S4 involves iterative adjustments until the predicted growth activity meets the standard, resulting in a refined culture medium formulation.

[0188] For example, during the iteration process, a simulation is run after each adjustment. If it is still insufficient, adjustments are continued. If the activity reaches the target after three consecutive iterations, the final ratio configuration is output as the refined formula.

[0189] In one embodiment, considering the corn flour as the main component of the feed ingredients, the initial draft formulation has a carbon source ratio of 30%, a nitrogen source ratio of 15%, and a trace element ratio of 5%. In step S1, this draft is input into a metabolic model, which uses a flux balance analysis method to simulate the glycolysis pathway of lactic acid bacteria and predicts a growth activity of 0.4 per hour, which is lower than the preset standard of 0.5.

[0190] After step S2, the system enters the insufficient state.

[0191] In step S3, the nitrogen source was adjusted to 18%, and the activity was re-simulated to 0.45, which was still insufficient; the carbon source was adjusted to 35% again, and the activity reached 0.52 after simulation, which met the standard.

[0192] Through the iteration of step S4, a refined formula is obtained: 35% carbon source, 18% nitrogen source, and 5% trace elements. This improves the efficiency of the microbial strain in decomposing raw materials, resulting in higher fermentation yield and benefiting microbial optimization in feed production.

[0193] In another embodiment, for a raw material primarily composed of soybean meal, the initial draft contains 25% carbon source, 20% nitrogen source, and 4% trace elements. Step S1 simulates yeast growth, with a predicted activity of 0.3, which is below 0.5.

[0194] Step S2 confirms insufficiency.

[0195] Step S3 first adjusts the trace elements to 5%, simulating an activity of 0.35; then adjusts the nitrogen source to 22%, simulating an activity of 0.48; in step S4, the third iteration adjusts the carbon source to 28%, achieving an activity of 0.51, thus obtaining the refined formula. This iterative mechanism ensures the adaptability of the formula, reduces the trial-and-error costs in actual experiments, and improves the accuracy of the culture medium.

[0196] For example, in the application of Bacillus, the initial draft contained 40% carbon source, 10% nitrogen source, and 3% trace elements, with a simulated predicted activity of 0.45. After two adjustments, the carbon source was reduced to 38% and the nitrogen source increased to 12%, achieving an activity of 0.55. This process demonstrates model-driven optimization, effectively avoiding growth inhibition caused by nutrient imbalance.

[0197] It should be noted that the specific implementation of the metabolic model in the above embodiments is based on known academic definitions, such as flux balance analysis. This analysis solves the flux distribution in the metabolic network through linear programming, assumes flux balance under steady-state conditions, calculates indicators such as growth rate, and thus predicts the adjustment direction when activity is insufficient, ensuring the logical rigor of the iteration.

[0198] S1011. The step of extracting key nutrient element characteristics from the refined culture medium formula, determining fermentation environment adaptability indicators by comparing the key nutrient element characteristics with historical fermentation data, and obtaining an adaptability enhancement scheme includes: extracting key nutrient element characteristics from the refined culture medium formula, obtaining environmental adaptation records from historical fermentation data, comparing the key nutrient element characteristics with corresponding characteristics in historical fermentation data, calculating fermentation environment adaptability indicators, generating targeted adjustment measures based on the adaptability indicators, and forming an adaptability enhancement scheme.

[0199] In one embodiment, key nutrient element characteristics are extracted from the refined culture medium formulation, and fermentation environment adaptability indicators are determined by comparing the key nutrient element characteristics with historical fermentation data to obtain an adaptability enhancement scheme, including the following steps.

[0200] Step S1: Extract key nutrient characteristics from the refined culture medium formula.

[0201] Specifically, by analyzing the element ratio configuration in the refined culture medium formula, key nutrients such as carbon source, nitrogen source and trace elements are identified and their concentration values ​​are quantified to form a feature vector. For example, the carbon source concentration is represented as the first dimension in the vector.

[0202] Step S2: Obtain environmental adaptation records from historical fermentation data.

[0203] For example, historical fermentation records related to similar strains can be retrieved from a pre-defined database, including data on their adaptation performance under environmental parameters such as temperature, pH, and oxygen levels.

[0204] Step S3: Compare the characteristics of key nutrients with the corresponding characteristics in historical fermentation data.

[0205] In one possible implementation, the cosine similarity algorithm is used to calculate the similarity between the key nutrient element feature vector and the corresponding feature vector in historical data. The cosine similarity algorithm obtains the similarity value by calculating the dot product of two vectors and dividing it by the product of their respective norms. The formula is cosθ = (A·B) / (|A| |B|), where A is the current feature vector and B is the historical feature vector. This algorithm is often used for feature matching and can effectively quantify the degree of matching of nutrient elements.

[0206] Step S31: Filter out records with a matching degree higher than the threshold based on the similarity results.

[0207] Step S32: Perform a weighted average on the selected records to obtain the comprehensive feature value after comparison.

[0208] Specifically, this comparison process can improve the accuracy of fitness assessment. By using the cosine similarity algorithm, it avoids the limitations of simple Euclidean distance and is beneficial for processing high-dimensional nutritional data.

[0209] Step S4: Calculate the fermentation environment adaptability index.

[0210] For example, using the comprehensive feature values ​​after comparison, a linear regression model is applied to predict the fitness score. The linear regression model is obtained by fitting historical data using the least squares method. The process includes collecting independent variables such as nutrient concentration and dependent variables such as growth rate, and then solving the coefficients to minimize the sum of squared residuals, thereby calculating the index value.

[0211] Step S41: Input the comprehensive feature values ​​into the model and run the prediction.

[0212] In step S42, if the predicted score is lower than a preset threshold, it is marked as low fitness.

[0213] In one embodiment, this calculation method is applied to a lactic acid bacteria fermentation scenario, which can significantly improve the reliability of the indicators and bring more stable fermentation prediction results.

[0214] Step S5: Generate targeted adjustment measures based on the adaptability indicators.

[0215] Specifically, based on the calculated adaptation index, if the index shows insufficient nitrogen source, measures to increase the amino acid concentration are generated. The index value is mapped to specific adjustment parameters through a mapping table, such as increasing the nitrogen source ratio from 5% to 7%.

[0216] Step S51: Traverse the low-fitness items in the indicators and match them with predefined adjustment rules.

[0217] Step S52: Output a list of adjustment measures, including the amount of elements added and fine-tuning of environmental parameters.

[0218] For example, in feed ingredient fermentation, this measure can precisely correct for microbial metabolic deviations, which is beneficial to improving overall decomposition efficiency.

[0219] Step S6: Formulate an adaptive enhancement scheme.

[0220] In one possible implementation, the generated adjustment measures are integrated into a program document, including a sequence of steps and an assessment of expected effects, thereby obtaining a complete adaptive enhancement program.

[0221] Specifically, this formation process ensures the consistency of the scheme, closely connects with the aforementioned steps, and can bring higher strain adaptability and yield in actual fermentation.

[0222] In one embodiment, regarding the comparison process in step S3, considering the scenario of lactic acid bacteria fermenting in a corn substrate, key nutrient element characteristics are first extracted, such as glucose concentration of 20 g / L and ammonia nitrogen of 5 g / L. Then, environmental adaptation records similar to corn fermentation in historical data are obtained, such as a growth rate of 0.5 / h at pH 6.0. The cosine similarity calculated during comparison is 0.85, which is higher than the threshold of 0.7. Records are then selected for subsequent calculations. This method can reveal the impact of nutrient matching on growth and is beneficial for optimizing feed fermentation.

[0223] For example, in the calculation of step S4, assuming the comprehensive feature value is [20, 5, 0.1], the input linear regression model has model coefficients of [0.02, 0.1, 0.5], and the prediction fitness index is 1.2. If the threshold is 1.0, it is considered to be highly fitness. This example illustrates how the quantification process guides fermentation adjustment, resulting in more efficient resource utilization.

[0224] In one possible implementation, when generating measures in step S5, if the adaptability index is 0.8, the measures include adding 0.5 g / L of zinc. After verification with historical data, a plan is formed. This method can reduce trial iterations in microbial feed production and emphasizes the beneficial effects of targeted adjustments, such as increasing the survival rate of strains by 10%.

[0225] It should be noted that the above embodiments cover the key steps in the field of feed ingredient fermentation, and ensure the integrity of the adaptability enhancement scheme through the logical chain from extraction to formation.

[0226] S1012. The step of optimizing the strain culture condition parameters using machine learning optimization method according to the adaptive enhancement scheme to obtain stable culture condition settings includes: inputting adaptive enhancement scheme data into the machine learning model for training, outputting culture condition parameter adjustment values, determining whether the adjustment values ​​cause a shift in metabolic direction, and if a shift in metabolic direction is caused, rolling back to the previous parameter set and continuing training until the adjustment values ​​no longer cause a shift in metabolic direction, thereby obtaining stable culture condition settings.

[0227] The adaptive enhancement scheme is used to optimize the strain culture condition parameters using machine learning optimization methods to obtain stable culture condition settings. This includes: inputting adaptive enhancement scheme data into the machine learning model for training, outputting culture condition parameter adjustment values, determining whether the adjustment values ​​cause a shift in the metabolic direction, and if so, rolling back to the previous parameter set and continuing training until the adjustment values ​​no longer cause a shift in the metabolic direction, thus obtaining stable culture condition settings.

[0228] In one embodiment, the adaptive enhancement scheme data is input into the machine learning model for training. Specifically, this includes importing the adaptive enhancement scheme data extracted from historical fermentation data, such as the correspondence between key nutrient element features and fermentation environment adaptability indicators, into a support vector machine model. This model is a supervised learning algorithm based on statistical learning theory, used for classification and regression tasks, and constructs a hyperplane to separate the data by maximizing the margin.

[0229] Output the adjustment values ​​for culture condition parameters.

[0230] Specifically, the input data is processed by a trained support vector machine model to generate specific adjustment values ​​for temperature, pH, or oxygen concentration, such as adjusting the temperature from the initial 28 degrees Celsius to 30 degrees Celsius to optimize bacterial growth.

[0231] Determining whether the adjustment value leads to a shift in metabolic direction specifically includes step S1, simulating the bacterial metabolic pathway after applying the adjustment value, using the Flux Balance Analysis method, which is a linear programming-based metabolic network analysis technique that predicts metabolic flux distribution by optimizing objective functions such as biomass growth rate, and calculating the change vector of metabolic flux before and after adjustment.

[0232] Step S11: Compare the change vector with a preset threshold. If the magnitude of the change vector exceeds 0.5, it is determined to be a shift in metabolic direction. This threshold is set based on the statistical mean of stable metabolism in historical fermentation data to ensure the accuracy of the judgment.

[0233] In step S12, if there is no offset, the current adjustment value is retained as the optimized output.

[0234] If this causes a shift in the metabolic direction, the model will roll back to the previous parameter set. Specifically, step S2 involves discarding the current adjustment value and restoring the parameter set from the previous training iteration, for example, rolling back from the adjusted 30 degrees Celsius to 28 degrees Celsius, while updating the model's weights to avoid repeated shifts.

[0235] Continue training until the adjustment value no longer causes a shift in metabolic direction, thus obtaining a stable culture condition setting.

[0236] Specifically, the support vector machine model is trained through multiple iterations, and the adjustment values ​​are gradually refined until the simulated metabolic pathway is stable, resulting in the final temperature setting of 28.5 degrees Celsius and pH value of 6.8.

[0237] In one possible implementation, for the step of determining whether the adjustment value causes a shift in the metabolic direction, considering a scenario where corn flour is the main feed ingredient, the Flux Balance Analysis method is used to calculate the metabolic flux. For example, the initial flux is 100 units, and after adjustment it is 105 units, with a change vector magnitude of 5. If it exceeds the threshold of 0.5, it will shift. This ensures that the strain maintains an efficient carbon metabolism pathway when decomposing corn flour, avoids energy waste, and is beneficial to improving fermentation efficiency.

[0238] For example, in the pH adjustment embodiment, the input adaptive enhancement scheme data includes the correspondence between pH value and growth activity in historical fermentation. The support vector machine model is trained to output an adjustment value, such as from 6.5 to 6.7. Metabolic shift is judged by comparing flux distribution. If the shift rolls back, training continues to obtain a stable pH value of 6.6. This mechanism is beneficial to prevent metabolic inhibition caused by acidification and improve the matching of strain with the decomposition requirements of raw materials.

[0239] In one embodiment, for the extension of the rollback mechanism, assuming the oxygen concentration is adjusted from 20% to 22%, if FluxBalance Analysis shows that the flux deviation exceeds the threshold, it is rolled back to 20% and then trained to 21%. This can stabilize the aerobic metabolic direction and is beneficial to the long-term maintenance of microbial activity in feed fermentation.

[0240] For example, when the whole process is applied to lactic acid bacteria culture, the input data includes nutrients such as the proportion of nitrogen sources. After outputting the adjustment value, the deviation is judged. If it is rolled back multiple times, a stable setting such as a stirring speed of 150 rpm is finally obtained. This optimization is beneficial to enhance adaptability and reduce the fermentation failure rate.

[0241] S1013. The step of generating a strain culture protocol sequence by setting stable culture conditions, determining subsequent raw material decomposition efficiency indicators, and obtaining an overall fermentation optimization path includes: generating a strain culture protocol sequence according to stable culture conditions, extracting fermentation start point data from the protocol sequence, calculating subsequent raw material decomposition efficiency indicators in combination with stable culture conditions, and summarizing all indicators to form an overall fermentation optimization path.

[0242] In one embodiment, the step of generating a strain culture protocol sequence by setting stable culture conditions, determining subsequent raw material decomposition efficiency indicators, and obtaining an overall fermentation optimization path includes the following steps.

[0243] Step S1: Generate a microbial culture protocol sequence based on stable culture conditions.

[0244] In one embodiment, step S1 involves generating a microbial culture protocol sequence based on stable culture conditions. Specifically, step S11 involves sequentially arranging parameters such as temperature, pH, and stirring speed from the stable culture conditions into a protocol sequence, ensuring that each parameter corresponds to a culture stage. This ensures that the protocol sequence reflects the actual culture process, which is beneficial to the accuracy of subsequent data extraction.

[0245] Step S2: Extract fermentation start point data from the protocol sequence.

[0246] In one embodiment, step S2, which involves extracting fermentation start point data from the protocol sequence, specifically includes step S21, identifying nodes in the protocol sequence that indicate the start of fermentation, such as the time point when the temperature reaches a preset value, and extracting relevant data such as the initial inoculum concentration. This extraction provides a starting point for efficiency calculations and is beneficial for optimizing the continuity of the process.

[0247] Step S3: Calculate the subsequent raw material decomposition efficiency index based on stable culture conditions.

[0248] In one embodiment, step S3 involves calculating the subsequent raw material decomposition efficiency index in conjunction with stable culture conditions. Specifically, this includes: step S31, obtaining the nutrient element ratios and environmental parameters from the stable culture conditions; step S32, inputting these parameters into a fermentation kinetic model to simulate the decomposition process. The fermentation kinetic model refers to a description of microbial growth and substrate consumption based on the Monod equation, for example, predicting decomposition efficiency by calculating the substrate consumption rate; and step S33, outputting the decomposition efficiency index, such as the percentage of raw material decomposition per hour. This calculation allows for accurate efficiency prediction and is beneficial for optimizing the fermentation process.

[0249] In one embodiment, for step S3, different feed raw material scenarios are considered. For example, when the raw material is corn-based feed, the proportion of nutrients obtained in step S31 is biased towards carbon source richness, and the Monod equation parameters in step S32 are adjusted to have a lower saturation constant to simulate rapid decomposition. The output index shows an efficiency of 85%. In the case of soybean meal-based raw material, the proportion is biased towards nitrogen source, the saturation constant is higher, and the efficiency index is 70%. This multi-scenario coverage can enhance the adaptability of the path and is beneficial to the stability in practical applications.

[0250] Step S4: Summarize all indicators to form an overall fermentation optimization path.

[0251] In one embodiment, step S4 involves summarizing all indicators to form an overall fermentation optimization path. Specifically, this includes: step S41, integrating the decomposition efficiency indicators calculated in step S3 with the protocol sequence data; step S42, applying a linear programming method to optimize the path, where linear programming refers to solving variable constraints by minimizing time or maximizing yield through an objective function, for example, setting efficiency indicators as constraints to solve for the optimal path sequence; and step S43, generating a complete path from the starting point to the end point. This summarization results in an efficient path, which is beneficial for guiding the overall fermentation process.

[0252] In one embodiment, for step S4, in the field of feed fermentation, when the decomposition efficiency index is 80%, the integrated path in step S41 emphasizes the increase of mid-stage temperature to improve efficiency; while in low efficiency index scenarios such as 60%, the path is adjusted to extend the initial stage, which is beneficial to the adaptation of the strain.

[0253] For example, the output increased by 15% after path optimization. This multifaceted example supports the practicality of the path and helps reduce resource waste.

[0254] In one embodiment, steps S3 and S4 are used in combination. For example, the efficiency index of step S3 is directly input into the linear programming of step S4 to generate a path that covers the entire chain from the initial data to the final optimization, which is beneficial to the overall efficiency improvement of feed raw material decomposition.

[0255] The above description of the embodiments is only for the purpose of helping to understand the technical solutions and core ideas of this application; those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for optimizing a microbial fermentation culture medium for agricultural feed production, characterized in that, include: Acquire feed ingredient composition data and microbial strain metabolic characteristic parameters, determine preliminary matching index by comparing the feed ingredient composition data with the metabolic characteristic parameters, and obtain a list of potential decomposition requirements of the strain for the raw materials; Based on the potential decomposition requirement list, the bacterial metabolic pathways are grouped and calibrated to obtain an optimized set of metabolic pathways. The optimized metabolic pathway set is used to screen nutrient element combinations from the resource information database, determine the element ratio configuration, and obtain an initial culture medium formula draft. The initial culture medium formulation draft was used to predict the growth status of the strain using a predictive simulation method. If the predicted growth activity was insufficient, the element ratio was iteratively adjusted to obtain a refined culture medium formulation. Key nutrient element characteristics are extracted from the refined culture medium formula, and fermentation environment adaptability indicators are determined by comparing the key nutrient element characteristics with historical fermentation data to obtain an adaptability enhancement scheme. Based on the aforementioned adaptive enhancement scheme, machine learning optimization methods are used to optimize the bacterial culture condition parameters to obtain stable culture condition settings; By setting the stable culture conditions, a strain culture protocol sequence is generated, and the subsequent raw material decomposition efficiency index is determined, thus obtaining the overall fermentation optimization path.

2. The method for optimizing a microbial fermentation culture medium for agricultural feed production according to claim 1, characterized in that, The process involves comparing the feed ingredient composition data with the metabolic characteristic parameters to determine a preliminary matching index, thereby obtaining a list of potential decomposition requirements of the microbial strain for the feed ingredients, including: The metabolic characteristic parameters of the microbial strains are extracted from the preset database. Each component in the feed ingredient composition data is compared with the metabolic characteristic parameters one by one. The degree of matching between each component and the metabolic characteristic parameters is calculated, and a preliminary matching index is generated. If the preliminary matching index is lower than a preset threshold, the corresponding component is marked as a component with high decomposition demand. A list of potential decomposition demand of the strains on the raw materials is formed based on all components with high decomposition demand.

3. The method for optimizing a microbial fermentation culture medium for agricultural feed production according to claim 1, characterized in that, The process involves grouping and calibrating the bacterial metabolic pathways according to the potential decomposition requirement list to obtain an optimized set of metabolic pathways, including: A grouping algorithm is used to group the metabolic pathways of the strains corresponding to the potential decomposition demand list, resulting in multiple metabolic pathway groups. The degree of deviation within each metabolic pathway group is judged. If the deviation exceeds the preset range, the corresponding path parameters are adjusted to reduce the deviation. By adjusting and calibrating all metabolic pathway groups multiple times, an optimized set of metabolic pathways is obtained.

4. The method for optimizing a microbial fermentation culture medium for agricultural feed production according to claim 1, characterized in that, The step of selecting nutrient element combinations from the resource information database using the optimized metabolic pathway set, determining the element ratio configuration, and obtaining an initial culture medium formulation draft includes: Obtain data on various nutrients from the resource information database, screen nutrients that match the optimized metabolic pathway set, combine the screened nutrients and calculate the proportional relationship between each element, determine the element ratio configuration, and generate an initial culture medium formula draft.

5. The method for optimizing a microbial fermentation culture medium for agricultural feed production according to claim 1, characterized in that, The initial culture medium formulation draft is used to predict the growth status of the microbial strain using a predictive simulation method. If the predicted growth activity is insufficient, the element ratios are iteratively adjusted to obtain a refined culture medium formulation, including: Input the initial culture medium formula draft into the metabolic model for simulation, obtain the predicted growth status of the strain, and determine whether the predicted growth activity meets the preset standard. If the predicted growth activity is insufficient, adjust the element ratio and re-input into the metabolic model for simulation. Through multiple iterations and adjustments until the predicted growth activity meets the standard, the refined culture medium formula is obtained.

6. The method for optimizing a microbial fermentation culture medium for agricultural feed production according to claim 1, characterized in that, The process of extracting key nutrient element characteristics from the refined culture medium formula, determining fermentation environment adaptability indicators by comparing the key nutrient element characteristics with historical fermentation data, and obtaining an adaptability enhancement scheme includes: Key nutrient element characteristics are extracted from the refined culture medium formula, environmental adaptation records are obtained from historical fermentation data, key nutrient element characteristics are compared with corresponding characteristics in historical fermentation data, fermentation environment adaptability indicators are calculated, and targeted adjustment measures are generated based on the adaptability indicators to form an adaptability enhancement program.

7. The method for optimizing a microbial fermentation culture medium for agricultural feed production according to claim 1, characterized in that, The step of optimizing the bacterial culture condition parameters using machine learning optimization methods based on the adaptive enhancement scheme to obtain stable culture condition settings includes: Input the adaptive enhancement scheme data into the machine learning model for training, output the culture condition parameter adjustment value, determine whether the adjustment value causes a shift in the metabolic direction, if it does, roll back to the previous parameter set, and continue training until the adjustment value no longer causes a shift in the metabolic direction, thus obtaining a stable culture condition setting.

8. The method for optimizing a microbial fermentation culture medium for agricultural feed production according to claim 1, characterized in that, The process of generating a strain culture protocol sequence by setting stable culture conditions, determining subsequent raw material decomposition efficiency indicators, and obtaining an overall fermentation optimization path includes: Based on the stable culture conditions, a strain culture protocol sequence is generated. Fermentation start point data is extracted from the protocol sequence. Combined with the stable culture conditions, the subsequent raw material decomposition efficiency index is calculated. All indicators are summarized to form an overall fermentation optimization path.

Images