Biological fermentation intelligent optimization method based on multi-objective optimization

Abstract

The invention discloses a biological fermentation intelligent optimization method based on multi-objective optimization, and the method comprises the following steps: 1) collecting whole-process parameters of a biological fermentation process, including but not limited to raw material parameters, process parameters, regulation and control parameters and result parameters; 2) carrying out data cleaning and preprocessing on the collected parameters, and carrying out data enhancement by adopting a generative adversarial network (GAN) so as to solve the problem of insufficient samples and improve data representativeness; by integrating various advanced optimization algorithms and data enhancement technologies, the efficiency and benefits of the biological fermentation process are remarkably improved. Meanwhile, various control parameters are optimized, and the stability and precision of the process are ensured. Besides, through Pareto frontier analysis, the optimal balance among a plurality of targets can be realized, the production efficiency and the cost are further optimized, and the technical scheme is proved to have remarkable advantages in the aspects of improving the yield, reducing the cost and simplifying the process.

Description

Technical Field

[0001] The invention belongs to the field of artificial intelligence and biological fermentation, and specifically is a biological fermentation intelligent optimization method based on multi-objective optimization. Background Art

[0002] Traditional bio-fermentation technology mainly relies on manual experience and trial-and-error methods to optimize the production process. This method has the disadvantages of insufficient data utilization, high manual dependence, high control complexity, insufficient samples and poor data representativeness, lack of multi-objective optimization, and insufficient consideration of actual constraints. These shortcomings limit the yield improvement and efficiency optimization of the fermentation process, and also increase the cost and complexity of operation.

[0003] This patent effectively overcomes the limitations of traditional methods by using advanced optimization algorithms, data enhancement technology and metabolic pathway optimization. It uses generative adversarial networks to enhance data, multi-objective optimization algorithms to balance different objectives, and graph neural networks to optimize metabolic pathways, which significantly improves the yield and efficiency of bio-fermentation, while reducing costs and operational complexity, and promotes the industrial application of bio-fermentation technology. Summary of the invention

[0004] The present invention aims to address the characteristics of the biological fermentation field: the high complexity of biological processes, incomplete data, multiple objectives, and the diversity of process parameters and controls. The data enhancement technology is used to solve the problem of insufficient samples; the multi-objective optimization algorithm framework is used to find the balance between multiple objectives, ensuring the constraints while optimizing the output, volume, chemical efficiency and cost. Constrained optimization is added to constrain the control parameters to avoid solutions that do not meet the production conditions. The Pareto frontier analysis technology is used to find the optimal solution among multiple optimization objectives to improve the flexibility and practicality of the algorithm.

[0005] To achieve the above object, the present invention provides the following technical solution: a biological fermentation intelligent optimization method based on multi-objective optimization, comprising the following steps: 1) Collect the whole process parameters of the biological fermentation process, including but not limited to raw material parameters, process parameters, control parameters and result parameters; 2) Clean and preprocess the collected parameters, and use generative adversarial networks (GANs) for data enhancement to solve the problem of insufficient samples and improve data representativeness; 3) Perform feature engineering on the preprocessed data, including polynomial expansion and logarithmic transformation of continuous variables, unique hot encoding and combined features of discrete variables; 4) Optimize fermentation parameters using multi-objective optimization algorithms, including maximizing yield, volume, and chemical efficiency, minimizing costs, and screen the optimal solution through Pareto frontier analysis; 5) Based on the optimization solution set, the metabolic pathway of the strain is optimized, and graph neural network modeling is used to identify key metabolites, key enzymes and metabolic pathways, and design gene editing strategies.

[0006] The parameters collected in step 1) include: Raw material parameters: water-soluble fiber content, amino acid composition, mineral content, salt additives, etc. Process parameters: stirring mode, temperature, pH change, dissolved oxygen concentration, reactor pressure, carbon dioxide concentration, etc.; Control parameters: sugar source addition, nitrogen source adjustment, gas flow rate control, pH regulator dosage, etc. Result parameters: product concentration, metabolite content, by-product accumulation, protein purity, etc.

[0007] In step 2), data cleaning includes outlier detection and normalization processing, specifically: using the quartile method to detect outliers; processing the detected outliers by mean correction or replacement; normalizing all parameters and adjusting the value range to between 0 and 1.

[0008] In step 2), a generative adversarial network (GAN) is used for data augmentation, including: the generator takes the fermentation parameters other than the target variable as input; the discriminator performs feedback training on the generator by comparing the statistical characteristics of the generated data and the real data; and the trained generative adversarial network is used to generate a new augmented data set.

[0009] In step 4), the multi-objective optimization algorithm adopts a differential evolution algorithm, and the optimization objectives include: maximizing fermentation yield; maximizing fermentation volume; maximizing chemical efficiency; minimizing production cost; and the weight of the objective function is dynamically adjusted according to actual production needs.

[0010] The penalty function method is used to improve the objective function and constrain the optimization process. The constraints include but are not limited to: fermentation temperature range; pH value range; dissolved oxygen concentration range; reactor pressure range.

[0011] The Pareto frontier analysis in step 4) includes: comparing the pros and cons of various optimization solutions on different objectives, screening out non-dominable solutions; constructing the Pareto frontier through non-dominable solutions; Select the optimal solution in the Pareto frontier based on actual needs, such as maximizing output and efficiency while avoiding significant increases in costs.

[0012] The optimization of metabolic pathways in step 5) includes: converting the microbial metabolic network into a graph model, where nodes represent metabolites, enzymes or metabolic fluxes, and edges represent metabolic reaction relationships; using graph neural networks to analyze the node and edge weights in the metabolic network to identify key enzymes and key metabolic pathways; designing gene editing strategies based on the analysis results of key pathways and enzymes to optimize the metabolic performance of the strain.

[0013] The optimization algorithm described in step 4) includes but is not limited to: gradient-based optimization algorithms, such as gradient descent, Adam, and RMSProp; population-based optimization algorithms, such as genetic algorithms and particle swarm optimization; heuristic-based optimization algorithms, such as taboo search and local search; reinforcement learning algorithms or Bayesian optimization algorithms.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention significantly improves the efficiency and benefits of the biofermentation process by integrating multiple advanced optimization algorithms and data enhancement technologies. At the same time, various control parameters are optimized to ensure the stability and accuracy of the process. In addition, through Pareto frontier analysis, the optimal balance between multiple objectives can be achieved, and production efficiency and cost are further optimized, which proves the significant advantages of this technical solution in improving yield, reducing costs and simplifying processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] like Figure 1 As shown, a bio-fermentation intelligent optimization method based on multi-objective optimization includes the following steps: 1) Collect the whole process parameters of the biological fermentation process, including but not limited to raw material parameters, process parameters, control parameters and result parameters; Among them, the raw material parameters represent the various index parameters of the raw materials used in biological fermentation, including but not limited to water-soluble fiber content, mineral content, acidity regulator, amino acid composition, etc. In the present invention, the raw material characteristic parameters involved include grain content, amino acid content, mineral composition, water-soluble fiber, salt additives, etc., covering more than 1,000 items.

[0018] Process parameters represent various index parameters in the biological fermentation process, including but not limited to stirring mode, temperature, pH change, dissolved oxygen, reactor pressure, carbon dioxide concentration, reactor type and volume, etc. These parameters are crucial to microbial growth, metabolic pathways and the formation of final products. In the example of the present invention, the process control parameters involved include stirring speed, temperature control, dissolved oxygen level, gas concentration change, reactor type, reaction pressure, fermentation time, etc., covering about 100 parameters.

[0019] The control parameters represent the index parameters that can be considered to be controlled during the biological fermentation process, including but not limited to the amount of sugar supplementation, nitrogen source addition, gas flow, dissolved oxygen control, pH regulator addition, etc. Through these adjustments, the fermentation process can be better controlled and the yield and quality of the product can be improved. The control parameters involved in the present invention include sugar source addition, nitrogen source adjustment, gas flow rate control, pH adjustment, etc., totaling about 20 items.

[0020] The result parameters represent the quality, yield and purity of the final product in the biological fermentation process, including but not limited to product purity, pH value after fermentation, gas release, product concentration, total yield, protein content, cell concentration, accumulation of by-products, etc. These indicators help to evaluate the efficiency of the fermentation process and the quality of the product. In the present invention, the result parameters involved include product concentration, metabolite content, gas generation rate, protein purity, by-product accumulation, pH value after fermentation, etc.

[0021] 2) Clean and preprocess the collected parameters, and use generative adversarial networks (GANs) for data enhancement to solve the problem of insufficient samples and improve data representativeness; Data cleaning is performed on all parameters, and the generative adversarial network (GAN) is used to enhance the data to solve the problem of insufficient samples and improve the representativeness of the data.

[0022] Specifically, the dissolved oxygen concentration in the original data set is used as the data to be filled, and the other data is used as the training set.

[0023] Subsequently, for the GAN generator, other fermentation process parameters (such as temperature, pH value, etc.) in addition to the dissolved oxygen concentration are input, and the generator generates a new dissolved oxygen concentration sample.

[0024] Subsequently, for the GAN discriminator, the generated dissolved oxygen concentration data is compared with the real dissolved oxygen concentration data, and training is performed to ensure that the generated data and the real data are consistent in statistical characteristics.

[0025] Subsequently, the dissolved oxygen concentration data is generated through the trained GAN model and combined with the original dataset to generate a new dataset containing more dissolved oxygen concentration data.

[0026] In addition, outliers are recorded using quartiles to facilitate their replacement in subsequent preprocessing. 3) Perform feature engineering on the preprocessed data, including polynomial expansion and logarithmic transformation of continuous variables, unique hot encoding and combined features of discrete variables; The data in the sample data set are preprocessed, where the outlier data recorded in S1 are normalized. The normalization formula is: Then, a random floating quantity z (mean 0, standard deviation σ) is generated that conforms to the Gaussian distribution. The Gaussian distribution formula is:

[0027] Then, a Gaussian float is added to the mean. Assuming the mean of the padding is μ, the new padding value is:

[0028] Then, the new fill value is normalized to the range 0 to 1:

[0029] Then, to ensure that the values ​​after padding do not exceed 0 and 1, use the clipping operation:

[0030] In addition, feature engineering is used to upgrade the dimension of all features. Examples include: polynomial expansion and logarithmic transformation for continuous variables; unique hot encoding and combined features for discrete variables.

[0031] 4) Optimize fermentation parameters using multi-objective optimization algorithms, including maximizing yield, volume, and chemical efficiency, minimizing costs, and screen the optimal solution through Pareto frontier analysis; A differential evolution algorithm is used as the optimization algorithm, and a multi-objective optimization method is used to set the objective function as maximizing output, maximizing volume, maximizing chemical efficiency, and minimizing cost. The specific formula is as follows: Goal 1: Maximize yield

[0032] Goal 2: Maximize volume

[0033] Goal 3: Maximize efficiency

[0034] Goal 4: Minimize costs

[0035] At this time, the objective function is: + + -

[0036] in: are the weights of each goal, which determine the priority between goals.

[0037] Set up a constrained optimization function and incorporate the constraints into the optimization model. Use the penalty function method to improve the original objective function.

[0038] Identify the constraints: (1) Temperature range:

[0039] (2) pH range:

[0040] For constraints , the penalty function is defined as: +

[0041] in: (1) is a penalty term that indicates the degree of constraint violation:

[0042] (2) is each constraint function, if >0 indicates that the constraint is violated; (3) Is the penalty coefficient, which controls the influence of the penalty function.

[0043] Based on the multiple solutions obtained by iterating the above algorithm multiple times, the dominance relationship is calculated using Pareto frontier analysis.

[0044] The partial solutions obtained by the difference algorithm are shown in the following table:

[0045] When calculating the dominance relationship, we can get: (1) Compared with solution 2, solution 1 has a higher yield, but solution 2 has a higher volume. If solution 1 has no obvious disadvantages in other goals, then solution 1 may dominate solution 2.

[0046] (2) Compared with solution 5, solution 1 has higher cost, while solution 5 has higher output and lower cost. Therefore, solution 5 may dominate solution 1.

[0047] Therefore, it can be concluded that solution 3 and solution 4 are not dominable, which means that they each have advantages in certain objectives.

[0048] Therefore, solution number 3, solution number 4, and solution number 5 constitute the Pareto frontier.

[0049] In addition, when selecting the optimal solution, based on actual application requirements, assuming that we want to maximize output and efficiency without significantly increasing costs, we can choose solution number 3 or solution number 4 as the final optimization solution.

[0050] 5) Based on the optimization solution set, the metabolic pathway of the strain is optimized, and graph neural network modeling is used to identify key metabolites, key enzymes and metabolic pathways, and design gene editing strategies.

[0051] Specifically, the metabolic network of microorganisms is converted into a graph, where nodes represent metabolites or enzymes, such as metabolite concentrations, enzyme expression levels, metabolic flux, etc., and edges represent metabolic reactions or conversion relationships between them, such as enzyme activity, reaction rate, reaction balance, etc. Graph neural networks are used for modeling optimization, to analyze the weights of nodes and edges, identify key enzymes and pathways, and help design gene editing strategies.

[0052] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A bio-fermentation intelligent optimization method based on multi-objective optimization, characterized in that: The following steps are involved: 1) Collect the whole process parameters of the biological fermentation process, including but not limited to raw material parameters, process parameters, control parameters and result parameters; 2) Clean and preprocess the collected parameters, and use generative adversarial networks (GANs) for data enhancement to solve the problem of insufficient samples and improve data representativeness; 3) Perform feature engineering on the preprocessed data, including polynomial expansion and logarithmic transformation of continuous variables, unique hot encoding and combined features of discrete variables; 4) Optimize fermentation parameters using multi-objective optimization algorithms, including maximizing yield, volume, and chemical efficiency, minimizing costs, and screen the optimal solution through Pareto frontier analysis; 5) Based on the optimization solution set, the metabolic pathway of the strain is optimized, and graph neural network modeling is used to identify key metabolites, key enzymes and metabolic pathways, and design gene editing strategies.

2. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 1, characterized in that: The parameters collected in step 1) include: Raw material parameters: water-soluble fiber content, amino acid composition, mineral content, salt additives, etc. Process parameters: stirring mode, temperature, pH change, dissolved oxygen concentration, reactor pressure, carbon dioxide concentration, etc.; Control parameters: sugar source addition, nitrogen source adjustment, gas flow rate control, pH regulator dosage, etc. Result parameters: product concentration, metabolite content, by-product accumulation, protein purity, etc.

3. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 1, characterized in that: In step 2), data cleaning includes outlier detection and normalization processing, specifically: using the quartile method to detect outliers; processing the detected outliers by mean correction or replacement; normalizing all parameters and adjusting the value range to between 0 and 1.

4. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 1, characterized in that: In step 2), a generative adversarial network (GAN) is used for data augmentation, including: the generator takes the fermentation parameters other than the target variable as input; the discriminator performs feedback training on the generator by comparing the statistical characteristics of the generated data and the real data; and the trained generative adversarial network is used to generate a new augmented data set.

5. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 1, characterized in that: In step 4), the multi-objective optimization algorithm adopts a differential evolution algorithm, and the optimization objectives include: maximizing fermentation yield; maximizing fermentation volume; maximizing chemical efficiency; minimizing production cost; and the weight of the objective function is dynamically adjusted according to actual production needs.

6. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 5, characterized in that: The penalty function method is used to improve the objective function and constrain the optimization process. The constraints include but are not limited to: fermentation temperature range; pH value range; dissolved oxygen concentration range; reactor pressure range.

7. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 1, characterized in that: The Pareto frontier analysis in step 4) includes: comparing the pros and cons of various optimization solutions on different objectives, screening out non-dominable solutions; constructing the Pareto frontier through non-dominable solutions; Select the optimal solution in the Pareto frontier based on actual needs, such as maximizing output and efficiency while avoiding significant increases in costs.

8. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 1, characterized in that: The optimization of metabolic pathways in step 5) includes: converting the microbial metabolic network into a graph model, where nodes represent metabolites, enzymes or metabolic fluxes, and edges represent metabolic reaction relationships; using graph neural networks to analyze the node and edge weights in the metabolic network to identify key enzymes and key metabolic pathways; designing gene editing strategies based on the analysis results of key pathways and enzymes to optimize the metabolic performance of the strain.

9. The method for intelligent optimization of biological fermentation based on multi-objective optimization according to claim 1, characterized in that: The optimization algorithm described in step 4) includes but is not limited to: gradient-based optimization algorithms, such as gradient descent, Adam, and RMSProp; population-based optimization algorithms, such as genetic algorithms and particle swarm optimization; heuristic-based optimization algorithms, such as taboo search and local search; reinforcement learning algorithms or Bayesian optimization algorithms.

Images