Black fungus high-iron strain hybridization breeding optimization method based on big data analysis

Abstract

The invention provides a black fungus high-iron strain hybridization breeding optimization method based on big data analysis. The method relates to the field of big data analysis and black fungus variety cultivation, and comprises the following steps: S1, acquiring strain basic data, hybridization process data, culture environment parameters and external reference data, and carrying out standardization and structuring processing; s2, coding and optimizing the hybridization combination by adopting an offset random key genetic algorithm, and generating an optimal hybridization scheme in combination with a double-elite evolution mechanism and a dynamic parameter adjustment strategy; s3, constructing an environment dynamic model through a model mixed inverse reinforcement learning algorithm, and searching an optimal culture strategy for maximizing the iron content in the virtual environment; and S4, constructing a multi-index comprehensive scoring system based on an analytic hierarchy process, identifying abnormal samples, triggering model backtracking adjustment, and outputting high-performance strains to enter the next round of breeding. According to the method, the breeding efficiency and the accuracy of strain breeding are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis and black fungus variety breeding, and in particular to a black fungus high-iron strain hybridization breeding optimization method based on big data analysis. Background Art

[0002] Black fungus, a nutritious edible and medicinal fungus, is known for its iron content, which is important for improving anemia and replenishing trace elements. Currently, there are few reports on the selection and breeding of functional black fungus varieties. However, research has been conducted on other edible and medicinal fungi, such as Ganoderma lucidum, Cordyceps militaris, and Hericium erinaceus, resulting in the breeding of functional varieties, such as polysaccharide-rich Ganoderma lucidum, cordycepin-rich Cordyceps militaris, and polysaccharide Hericium erinaceus. These functional varieties are often bred through mutagenesis or a combination of hybridization and systematic selection. However, these methods suffer from non-directional and random breeding, unstable trait expression, and a high workload. Furthermore, traditional strain hybridization and breeding processes commonly face the following challenges: Low efficiency of combination screening: Traditional hybridization typically involves testing a large number of combinations, resulting in a low success rate and a lack of effective screening and process simulation mechanisms. Breeding cycles are long and costly: Extensive field experiments are required to verify target traits, such as iron content, yield, and growth period. Hybrid combination evaluation relies on manual experience: During the strain performance evaluation phase, there is often a lack of systematic quantitative standards, leading to inaccurate identification of superior strains. Sensitive to the culture environment but lacks intelligent prediction methods: Traits such as iron content are significantly affected by factors such as culture medium formula, temperature and humidity, and traditional methods cannot dynamically predict and optimize.

[0003] Although big data and artificial intelligence technologies have been applied in crop breeding, they still have limitations in the field of edible and medicinal fungi: existing genetic algorithms are difficult to adapt to the high-dimensional combinatorial optimization problem of strain hybridization, and premature convergence leads to insufficient global search capabilities; inverse reinforcement learning technology has not yet been used in fields such as multi-focus robot control for modeling culture conditions, and there is a lack of a hybrid driving mechanism of expert data and real-time data; the dynamic environment-gene expression coupling model is missing, and the probabilistic effect of gene recombination on iron content cannot be predicted.

[0004] Therefore, it is urgent to build an intelligent breeding method for high-iron black fungus strains that integrates combinatorial optimization algorithms, environmental dynamic modeling and multi-objective evaluation to achieve closed-loop optimization of hybrid design-cultivation regulation-performance feedback. Summary of the Invention

[0005] In response to the above problems, the present invention aims to construct a set of black fungus high iron strain hybridization breeding optimization methods based on big data analysis, forming a closed-loop automatic optimization process from data collection, combination optimization, environmental prediction to performance evaluation.

[0006] To achieve the above objectives, the following technical solutions are adopted:

[0007] In a first aspect, the present invention provides a method for hybridization and breeding optimization of black fungus high iron strains based on big data analysis, comprising:

[0008] Step S1: Collect basic strain data, hybridization process data, culture environment parameters and external reference data, and perform standardization and structured processing;

[0009] Step S2: using a biased random bond genetic algorithm to encode and optimize the hybrid combination, and combining the dual elite evolution mechanism and dynamic parameter adjustment strategy to generate the optimal hybridization scheme;

[0010] Step S3: constructing an environmental dynamic model through a model-based hybrid inverse reinforcement learning algorithm to search for the optimal cultivation strategy that maximizes iron content in the virtual environment;

[0011] Step S4: Construct a multi-index comprehensive scoring system based on the hierarchical analysis method to identify abnormal samples and trigger model backtracking adjustments, and output high-performance strains for the next round of breeding.

[0012] Furthermore, the implementation of the biased random key genetic algorithm in step S2 includes:

[0013] Encode the hybrid combination as a random key chromosome vector X represented by n-dimensional continuous real numbers;

[0014] The random bond chromosome vector X is decoded into a specific combination strategy and fitness score by an encoder, and an objective function is constructed with the iron content super-parent rate and agronomic trait stability as the evaluation core;

[0015] By introducing parental gene expression data, we simulated the gene recombination behavior when different monokaryotic strains merged, predicted the probability distribution of iron content in the offspring, and formed a dynamic hybridization simulation model.

[0016] Adopting the dual elite evolution mechanism, each generation retains the top Elite individuals are selected and biased crossover is performed with probability p>0.5: the offspring genes inherit the elite parent 1 gene with probability p and the non-elite parent 2 gene with probability 1-p;

[0017] As the number of iterations advances, parameters such as the elite ratio and mutation probability are dynamically adjusted to adapt to the search progress.

[0018] Furthermore, the biased random key genetic algorithm further includes:

[0019] Dithering mechanism: Probability for elite individuals Apply disturbance;

[0020] Reset mechanism: When there is no fitness improvement for the preset number of consecutive generations, non-elite individuals are cleared and random chromosomes are regenerated.

[0021] Furthermore, the objective function is:

[0022]

[0023] in, : The fitness score corresponding to chromosome X; : Iron content super-parent rate, that is, the ratio of the predicted iron content of the offspring to be higher than the average of the parents; : Agronomic trait stability index; α, β: objective function weight parameters, set according to task requirements.

[0024] Furthermore, the step S3: constructing an environmental dynamic model by using a model-based hybrid inverse reinforcement learning algorithm, and searching for an optimal cultivation strategy that maximizes iron content in the virtual environment, includes:

[0025] Construct a hybrid dataset using expert demonstration data and learner online data;

[0026] Use neural network to fit the environmental state transition probability to obtain the environmental dynamic model , and adopts strategy optimization and reward function iteration to form a closed-loop learning mechanism to efficiently simulate the culture medium formula and environmental condition adjustment behavior in a virtual model environment.

[0027] Furthermore, the step S3 further includes:

[0028] Randomly sample states from the expert demonstration dataset as the initial states for policy optimization.

[0029] In the environmental dynamic model In the process, a gradient-free optimization algorithm or model predictive control method is used to optimize the policy π to maximize the cumulative reward, and the reward function is updated by gradient ascent to generate a new policy π Further feedback to the real environment to generate new data , forming a data closed loop.

[0030] Furthermore, the reward function iteration formula is:

[0031]

[0032] : The current reward function, used in the tth iteration; : The reward function after the update at the t+1th iteration; : Learning rate, used to control the update step size of the reward function; : Find the gradient of the reward function f; :In the reward function Next, expert strategy Expected cumulative reward of :In the reward function Next, the current strategy Expected cumulative reward of : Expert strategies, usually derived from human experience or high-quality demonstrations; : The current strategy, that is, the strategy optimized in the tth iteration.

[0033] Furthermore, step S4: constructing a multi-index comprehensive scoring system based on the analytic hierarchy process, identifying abnormal samples and triggering model retrospective adjustment, outputting high-performance strains for the next round of breeding, includes:

[0034] The analytic hierarchy process was used to assign weights to each indicator: iron content accounted for 50%, yield accounted for 30%, and growth period accounted for 20%. The comprehensive score of each strain was calculated as follows:

[0035] Comprehensive score = 0.5 × iron content score + 0.3 × yield score + 0.2 × growth period score

[0036] Among them, each score is standardized;

[0037] When the comprehensive score of the strain is lower than the preset threshold or the iron content does not meet the standard, the system automatically goes back to the parent selection or hybrid design stage and updates the parameters of the genetic algorithm or recommended model to optimize the prediction quality of the mating combination.

[0038] Furthermore, the step S4 further includes:

[0039] If the strain compliance rate is lower than the threshold, the weight of the high-iron trait is increased or a penalty item is added. The total loss is calculated as follows:

[0040] Total loss = original loss + penalty coefficient × (average iron content of parents - actual iron content) 2

[0041] Among them, the penalty coefficient is dynamically adjusted according to the degree of non-compliance with the iron content.

[0042] In a second aspect, the present invention further provides a black fungus high iron strain hybridization breeding optimization system based on big data analysis, comprising:

[0043] Data collection and preprocessing module: collects basic strain data, hybridization process data, culture environment parameters and external reference data, and performs standardization and structured processing;

[0044] Hybrid combination optimization module: uses biased random bond genetic algorithm to encode and optimize hybrid combinations, and combines dual elite evolution mechanism and dynamic parameter adjustment strategy to generate the optimal hybridization scheme;

[0045] Culture condition prediction module: This module uses a model-based hybrid inverse reinforcement learning algorithm to build an environmental dynamic model and search for the optimal culture strategy that maximizes iron content in the virtual environment.

[0046] Strain performance evaluation module: Build a multi-index comprehensive scoring system based on the hierarchical analysis method to identify abnormal samples and trigger model backtracking adjustments, outputting high-performance strains for the next round of breeding.

[0047] Compared with the prior art, the present invention achieves the following beneficial effects:

[0048] 1. This paper proposes random-bond chromosomes and a dual-elite evolutionary mechanism, constructing a biased random-bond genetic algorithm. By encoding crossover combinations as continuous real-valued random-bond vectors, this algorithm constructs a chromosome representation decoupled from the problem structure. The dual-elite mechanism of elite retention and biased mating is introduced to enhance the retention and heritability of high-quality solutions. Biased crossover and mutation operations are used to avoid premature convergence and achieve efficient global search of the strain combination space.

[0049] 2. This invention proposes a dynamic adaptive parameter adjustment strategy: As the evolutionary process progresses, the system dynamically adjusts parameters such as the elite ratio, mutation probability, and population size. This improves the adaptability of the strategy during the early exploration and late convergence phases, enhancing the algorithm's stability and generalization capabilities.

[0050] 3. This paper proposes a model-based hybrid inverse reinforcement learning algorithm: This algorithm utilizes expert demonstration data and learner interaction data to construct a hybrid environment model. A neural network is used to fit the state transition probability distribution. Through policy optimization and reward function iteration, a closed-loop learning mechanism (reward function → strategy → model) is formed. This algorithm efficiently simulates the adjustment of culture medium formulation and environmental conditions in a virtual model environment, automatically searching for the optimal culture strategy.

[0051] 4. This paper proposes a regularized loss function design to prevent model overfitting: During dynamic model training, the loss function introduces weight decay (L2 norm regularization) and gradient clipping mechanisms. This prevents overfitting of the model in small sample sizes or with expert data distributions, improving generalization to new samples and complex environments.

[0052] 5. This paper proposes a strain performance evaluation method based on a comprehensive scoring mechanism and penalty factors. This method constructs a multi-index evaluation system based on the Analytic Hierarchy Process (AHP), weightedly integrating iron content (50%), yield (30%), and growth period (20%). This standardized scoring mechanism and penalty factors automatically identify low-performing or substandard strains, and iteratively improves breeding strategies by tracing back to upstream optimization steps.

[0053] 6. Based on the evaluation of numerous black fungus germplasm resources, this invention selectively selects strains with high iron content as hybrid parents, making it easier to create super-iron-tolerant strains. This method, combined with the addition of culture media at appropriate exogenous iron ion concentrations, specifically cultivates high-iron strains. Compared to hybridization and mutation breeding techniques alone, this method is more targeted and efficient, improving breeding efficiency. This invention not only improves the genetic traits of black fungus but also enhances high-iron traits through altered culture conditions, significantly improving the accuracy of strain selection and providing new insights into the selection of functional edible fungi strains.

[0054] It should be understood that the contents described in the summary of the invention are not intended to limit the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are provided for a better understanding of the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, among which:

[0056] Figure 1 This is a flow chart of a method for optimizing hybridization and breeding of high-iron-rich black fungus strains based on big data analysis according to an embodiment of the present invention;

[0057] Figure 2 This is a module diagram of a black fungus high iron strain hybridization breeding optimization system based on big data analysis according to an embodiment of the present invention;

[0058] Figure 3 The monokaryotic hyphae of the black fungus strain JAUA628 under the optical microscope of the present invention;

[0059] Figure 4 The binucleate hyphae of the black fungus strain JAUA628 under the optical microscope of the present invention;

[0060] Figure 5 The black fungus strain JAUA628 of the present invention is antagonistic to its parent;

[0061] Figure 6 The invention discloses a black fungus strain JAUA628. DETAILED DESCRIPTION

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of 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 making creative efforts are within the scope of protection of the present invention.

[0063] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the related objects are in an "or" relationship.

[0064] The present invention starts from the black fungus strain with high iron content, creates a super high iron affinity strain through hybrid breeding, and then combines exogenous iron ion addition culture medium to cultivate the high iron black fungus strain, which is an effective way to efficiently obtain the high iron black fungus strain.

[0065] The high-iron black fungus strain JAUA628 is derived from the wild black fungus strain A356 (161.4 mg / kg) with high iron content and the high-yield cultivated strain A592. Spores were collected and isolated from single spores to obtain a monokaryotic strain. Single-spore hybridization was then performed to create a true hybrid strain. The hybrid strain AA48, which exhibited the highest iron content and exceeded the parent strain, was screened. The monokaryotic strain obtained from the parent strain A356 was then backcrossed with the monokaryotic strain of the parent A356 to identify the hybrid strain JAUA628, which exhibited a high iron content and exceeded the parent strain by 20%. When grown in a culture medium composed of 78% sawdust, 20% wheat bran, 1% lime, 1% gypsum, 2.0 g / kg ferrous sulfate, and 60% moisture, the iron content of the hybrid strain JAUA628 was increased to 221 mg / kg.

[0066] The high-iron black fungus strain JAUA628 was deposited in the China Center for Type Culture Collection on April 19, 2024, with the deposit number CCTCCM2024717.

[0067] The high-iron black fungus strain of the present invention was created by improving its specific traits through genetic breeding and combining it with supporting cultivation methods. However, research on the selection and breeding of functional black fungus varieties has not yet been reported. Other functional edible fungi are mostly achieved by modifying culture conditions and utilizing the fungus's enrichment capacity, without genetically modifying their traits, which has certain limitations.

[0068] Based on the evaluation of a large number of black fungus germplasm resources, this method selectively selects strains with high iron content as hybrid parents, making it easier to create super-iron-tolerant strains. By adding culture media with appropriate exogenous iron ion concentrations, high-iron-content strains can be cultivated. This method is more targeted and efficient than hybridization or mutation breeding alone, improving breeding efficiency.

[0069] The present invention not only improves the genetic traits of black fungus, but also combines changes in culture conditions to double-strengthen the high-iron trait, significantly improving the accuracy of strain selection and breeding, providing new ideas for the selection and breeding of functional edible fungi strains.

[0070] This invention uses artificial intelligence to build a hybrid breeding optimization system for black fungus high-iron strains based on big data analysis, forming a closed-loop automatic optimization process from data collection, combination optimization, environmental prediction to performance evaluation, providing an intelligent decision-making tool for the breeding of edible and medicinal fungi such as black fungus, and significantly improving the efficiency and breeding accuracy in hybrid combination optimization, environmental prediction modeling and performance evaluation.

[0071] Figure 1 The following is a flow chart showing a method for optimizing the hybridization and breeding of high-iron black fungus strains based on big data analysis according to an embodiment of the present invention. Figure 1 As shown, a hybridization and breeding optimization method 100 for black fungus high-iron strains based on big data analysis includes:

[0072] Step S1: Collect basic strain data, hybridization process data, culture environment parameters and external reference data, and perform standardization and structured processing;

[0073] This step S1 is used to realize multi-source data collection and preprocessing. In order to build an intelligent black fungus high iron strain hybridization and breeding optimization system, it is necessary to first establish a complete, standardized, and structured data foundation. By systematically collecting and cleaning multi-source heterogeneous data, high-quality support is provided for subsequent hybridization optimization, model prediction, and performance evaluation. It mainly includes the following steps:

[0074] Step S1.1: Collection of four types of data

[0075] Step S1.1.1: Collection of basic strain data

[0076] The genetic background and agronomic trait information of key parent strains were collected as the basic input for hybridization design.

[0077] Genetic information: such as the genotype data of parents A356 and A592, the lock-shaped joint characteristics of monokaryotic hyphae, and karyotype information, etc., are used to assist in inferring the genetic expression ability of the fusion offspring.

[0078] Iron content measurement value: For example, the iron content of the wild black fungus strain A356 is 161.4 mg / kg, which serves as a reference baseline for the super-parent performance of the offspring.

[0079] Agronomic trait data: including mycelium growth rate, yield performance, growth cycle (such as ear emergence time, maturity time), etc., are used for subsequent comprehensive performance evaluation.

[0080] Step S1.1.2: Hybridization process data acquisition

[0081] Data is collected for key steps in each round of hybridization experiments, covering microscopic operation records and experimental result indicators.

[0082] Single spore isolation data: record the number of monokaryotic strains isolated from fruiting bodies in each round of experiments (e.g., 18 in the first round) to provide a diversity basis for subsequent combination design.

[0083] Hybridization combination design: record the number of combinations generated in each round of hybridization (e.g., 81 combinations generated in the first round), and number them according to the naming rules for easy tracking and management.

[0084] Antagonistic reaction experimental data: Collect antagonistic reaction images and response levels (such as fusion, refusal to fusion, and excessive reaction) between different mycelial combinations to eliminate non-true hybrid combinations or homologous parent groups.

[0085] Agronomic trait data: including mycelium growth rate, yield performance, growth cycle (such as ear emergence time, maturity time), etc., are used for subsequent comprehensive performance evaluation.

[0086] Step S1.1.3: Collection of culture environment and condition parameters

[0087] Collect important environmental factors that affect strain growth and iron accumulation to support subsequent reinforcement learning environment modeling and simulation.

[0088] Culture medium formula composition: record the basic materials (such as 78% sawdust, 20% wheat bran, lime / gypsum and other additives) as the culture environment input parameters.

[0089] Iron ion concentration setting: covers the set concentration gradient (such as 1.0-2.5g / kg ferrous sulfate), which is used as an input variable to associate the target iron content in the prediction model.

[0090] Temperature and humidity conditions: including continuous or staged environmental control data such as temperature during the sterilization stage (22-28°C) and humidity during the ear removal stage (85-90%).

[0091] Agronomic trait data: including mycelium growth rate, yield performance, growth cycle (such as ear emergence time, maturity time), etc., are used for subsequent comprehensive performance evaluation.

[0092] Step S1.1.4: External reference data acquisition

[0093] Introduce the technical routes and successful cases of functional breeding of other edible and medicinal fungi (such as Ganoderma lucidum and Cordyceps militaris) as a priori information for system design reference and model training.

[0094] Functional variety breeding methods: such as polysaccharide enrichment, breeding process for selenium adsorption capacity, metabolic pathway enhancement strategy, etc.

[0095] Breeding evaluation index system: refer to the multi-objective evaluation method used in functional evaluation, compare and absorb its effective experience.

[0096] Agronomic trait data: including mycelium growth rate, yield performance, growth cycle (such as ear emergence time, maturity time), etc., are used for subsequent comprehensive performance evaluation.

[0097] Step S1.2: Data preprocessing process

[0098] To ensure data availability and model training effectiveness, the collected multi-source data is standardized and structured, including:

[0099] Missing values ​​were filled and outliers were eliminated (e.g., iron content below 10 mg / kg was directly eliminated).

[0100] Numerical normalization processing (applicable to various continuous indicators, such as temperature, concentration, etc.).

[0101] Categorical feature encoding (e.g., antagonistic response level, strain identifier).

[0102] Time series alignment and stage labeling (for growth period and dynamic simulation modeling).

[0103] Through the above-mentioned multi-dimensional data collection and preprocessing operations, the system has built a high-quality strain breeding data foundation that integrates genetics, physiology, environment and process behavior, laying a solid digital platform foundation for subsequent algorithm optimization and performance improvement.

[0104] Step S2: using a biased random bond genetic algorithm to encode and optimize the hybrid combination, and combining the dual elite evolution mechanism and dynamic parameter adjustment strategy to generate the optimal hybridization scheme;

[0105] In order to effectively improve the efficiency of hybrid breeding of high-iron strains of black fungus and reduce ineffective combinations and experimental costs, this system further introduces a biased random bond genetic algorithm based on data collection to construct an intelligent hybrid optimization mechanism and process simulation model. By comprehensively considering the two goals of "iron content super-parent rate" and "agronomic trait stability", the hybrid combination strategy and backcross algebra are dynamically optimized, significantly improving the screening accuracy of real hybrid offspring. Step S2 is used to achieve hybrid combination optimization and process simulation, in which the implementation of the biased random bond genetic algorithm includes:

[0106] Encode the hybrid combination as a random key chromosome vector X represented by n-dimensional continuous real numbers;

[0107] The random bond chromosome vector X is decoded into a specific combination strategy and fitness score by an encoder, and an objective function is constructed with the iron content super-parent rate and agronomic trait stability as the evaluation core;

[0108] By introducing parental gene expression data, we simulated the gene recombination behavior when different monokaryotic strains merged, predicted the probability distribution of iron content in the offspring, and formed a dynamic hybridization simulation model.

[0109] Adopting the dual elite evolution mechanism, each generation retains the top Elite individuals are selected and biased crossover is performed with probability p>0.5: the offspring genes inherit the elite parent 1 gene with probability p and the non-elite parent 2 gene with probability 1-p;

[0110] As the number of iterations advances, parameters such as the elite ratio and mutation probability are dynamically adjusted to adapt to the search progress.

[0111] Step S2 specifically includes the following steps:

[0112] Step S2.1: Hybrid combination optimization strategy

[0113] The number of combinations in traditional hybridization is huge (e.g. 81 combinations in the first round), but the success rate of actual hybridization is low (e.g. 36 true hybrid strains were screened out). Therefore, this system adopts the following strategy:

[0114] Antagonistic reaction screening: Using the results of the mycelial antagonistic reaction experiment, combinations with significant kinship differences are preferentially retained, and homogeneous or non-fusion combinations are preliminarily excluded.

[0115] Dynamic adjustment of backcrossing strategy: Based on the mid-term test results, the number of backcrossing of offspring with obvious high-iron traits, such as "hybrid strain AA48 and wild black fungus strain A356-4 with high iron content", is dynamically increased to ensure the genetic stability of the target traits.

[0116] Multi-objective optimization: By constructing a joint objective function, we can balance the improvement of iron content with the stable performance of agronomic traits such as yield and growth period.

[0117] Step S2.2: Constructing a biased random bond genetic algorithm

[0118] The system uses the Biased Random Key Genetic Algorithm (BRKGA) for combinatorial optimization. Its core feature is the random key chromosome combined with the double elite evolution mechanism, which has good search space coverage and global optimization efficiency.

[0119] Random key chromosome vector is a virtual code designed in biased random key genetic algorithm to represent the optimization strategy of hybrid combination. It is a mathematical abstract model. Corresponding to a breeding decision variable, a decoder can be used to convert it into an actionable hybrid breeding strategy (such as parent selection, number of backcrosses, and recommended environmental parameters such as iron concentration in culture medium). Ultimately, the optimal strategy is output to guide experimental design. The random key chromosome vector is defined as each individual represented by a real number vector of dimension n:

[0120]

[0121] X: random key chromosome vector, used to encode the selection order and strategy parameters of strain combinations; n is the encoding dimension, which is determined by the number of decision variables to be optimized (such as parent selection, number of backcrosses, environmental parameter weights, etc.); : The i-th random key represents the priority or proportional allocation parameter in the decoder, which is a continuous real number in the interval [0,1], indicating the probability weight or priority of a decision; the decoder decodes the random key vector X into a specific combination strategy and fitness score, such as by threshold decoding (for example, when , selected as the i-th parent) to generate a hybrid combination strategy. Taking iron content and trait stability as the core of the evaluation, the objective function is constructed:

[0122]

[0123] : The fitness score corresponding to the random key chromosome vector X; : Iron content super-parent rate, which predicts the ratio of offspring iron content higher than the average value of parents; : Agronomic trait stability, a trait consistency index based on fitting data such as ear emergence period and yield; α, β: objective function weight parameters, set according to task requirements. Step S2.3: Dynamic simulation of hybridization process

[0124] By introducing parental gene expression data, simulating the gene recombination behavior when different monokaryotic strains fuse, and predicting the probability distribution of iron content in offspring, a dynamic hybridization simulation model is formed:

[0125]

[0126] : The expected iron content of the offspring of the kth combination; : expression level of the jth iron-related gene; : The weight factor of the jth gene in the kth combination is obtained by fitting the linear regression model of the parental gene expression data, and the model training uses the least squares method; : The number of target genes involved in the simulation; : The summation symbol for subscript j from 1 to g represents the accumulation of weighted expression values ​​of all iron-related genes; : Mathematical expectation operation, which represents the weighted average of all possible offspring gene combinations under the condition of uncertainty in gene expression and combinatorial recombination.

[0127] Step S2.4: Evolution process and dynamic parameter adjustment

[0128] During the algorithm iteration process, a dual elite mechanism and adaptive parameter adjustment strategy are adopted:

[0129] Elite replication: Each generation retains fitness before Individual

[0130] Mutation Operation: Generate random chromosomes to increase diversity

[0131] Biased crossover: with probability Biased selection of elite genes to generate the next generation of individuals:

[0132]

[0133] : The gene value of the i-th position of the offspring, that is, the encoding value of the new individual generated after the mating operation at the i-th position, which is part of the chromosome of the new generation; : The gene from elite parent 1 indicates that the gene position is selected from individuals with higher fitness (i.e., better performance) in the current population; : Gene from non-elite parent 2, indicating that this gene position is selected from a non-optimal individual in the current population; p: Elite gene retention probability, indicating the probability that gene position i is directly inherited from an elite individual when generating the next generation of chromosomes. Since p > 0.5, this strategy has a clear "elite bias," enhancing the transmission of excellent genes; 1 - p: The probability of non-elite genes being selected, which is used to maintain population diversity and avoid falling into local optima. Optionally, the bias probability p can be linearly increased from 0.6 to 0.8.

[0134] With the number of iterations As the search progresses, the system dynamically adjusts parameters such as the elite ratio and mutation probability to adapt to the search progress. The formula is as follows:

[0135]

[0136] : The number of elite individuals at the tth algorithm iteration; : Maximum number of iterations; , : Set the minimum and maximum elite ratio thresholds, Initial value of is 0.2, The range of dynamic adjustment is: , .

[0137] Step S2.5: Dither and reset mechanism

[0138] In order to avoid the algorithm from falling into local optimum, The proportion of mutant individuals (completely randomly generated new solutions) is increased to improve population diversity. Elite individuals are slightly disturbed by the jitter operator, and the elite individuals are Apply disturbance, the expression is:

[0139]

[0140] : disturbance amplitude parameter; : uniformly distributed random variable; : disturbance probability; : The i-th random key value, with the same meaning as step S2.2.

[0141] When there is no fitness improvement for several consecutive generations, a reset operation is performed to clear non-elite individuals and regenerate new random key individuals (random chromosomes) to avoid premature convergence.

[0142] Step S3: constructing an environmental dynamic model through a model-based hybrid inverse reinforcement learning algorithm to search for the optimal cultivation strategy that maximizes iron content in the virtual environment;

[0143] This step S3 constructs an iron content prediction model by analyzing the nonlinear relationship between the iron ion concentration in the culture medium (range: 0–2.5 g / kg) and the iron content in the fruiting body (range: 194.3–221 mg / kg). Based on a model-based hybrid inverse reinforcement learning algorithm, it automatically searches for the optimal culture recipe, with "maximizing the iron content in the fruiting body" as the reward goal. The specific process of implementing intelligent culture condition prediction in step S3 includes:

[0144] Step S3.1: Construct model-based hybrid inverse reinforcement learning

[0145] This step S3.1 constructs an environmental dynamic model through a model-based hybrid inverse reinforcement learning algorithm to search for the optimal cultivation strategy that maximizes iron content in the virtual environment, including: constructing a hybrid data set through expert demonstration data and learner online data; using a neural network to fit the environmental state transition probability to obtain the environmental dynamic model , and adopts strategy optimization and reward function iteration to form a closed-loop learning mechanism, which can effectively simulate the adjustment behavior of culture medium formula and environmental conditions in the virtual model environment.

[0146] Input data is expert demonstration data (high-quality priors) and learner online data (Real-time correction), mixed to form the training set ( is a union operation). By mixing data, using expert demonstration Provide high-quality state-action pairs, narrow the strategy search space, avoid the inefficiency of random exploration; and use the learner trajectory Correct model errors, enhance adaptability to environmental dynamics, improve generalization of environmental dynamics predictions, and prevent the model from overfitting expert data. Expert demonstration data Including: artificially optimized culture medium formula (such as ferrous sulfate 1.5-2.5 g / kg), temperature and humidity control curve (25±2℃ during the fermentation period), and harvesting time point (initial spore ejection stage).

[0147] Use a neural network (such as Transformer or multi-layer perceptron) to simulate environmental dynamics, define the state space S = {culture medium formula C, temperature and humidity T, iron ion concentration Fe, mycelial growth rate G, metabolite concentration M}, and the action space A = {adjust Fe concentration ΔFe, change temperature ΔT, modify formula ratio ΔC}. Reward function ,in, is the actual iron concentration of the fruiting body (e.g., the iron content of JAUA628 in 2.0g / kg culture medium = 221mg / kg); ΔT is the temperature difference before and after the action is executed (e.g., from 25°C to 28°C, ); ΔC is the adjustment range of culture medium composition (e.g. wheat bran ratio changes from 18% to 20%, then ); is the corresponding weight coefficient, The typical ratio is 20:1:0.5, ensuring that the iron content contribution accounts for more than 95% of the total reward value, and the weight can be dynamically adjusted with the iteration stage, with the initial focus on exploration ( Lower), and later focus on convergence ( Approaching 1.0). Prediction in state Next action The training goal is to minimize the KL divergence between the predicted state distribution and the true state distribution. The loss function is Avoid overfitting by adding weight decay and gradient clipping:

[0148]

[0149] : Model loss function, used to measure the current learning model Dynamic with real environment the gap; : Mixed dataset, data demonstrated by experts and learner-generated data The combined dataset : The learned dynamic model of the environment; s: current state; a: action; s': next state; :In real environment, given state and actions The probability distribution of the next state after ; : The next state distribution predicted by the learning model, i.e., the neural network’s estimate of the environment dynamics; are model parameters; is the regularization coefficient; : Norm square, i.e. regular term, is used to constrain the parameter scale; : KL divergence, which measures the difference between two probability distributions.

[0150] This loss function can prevent the model from over-relying on expert data distribution and enhance its adaptability to complex environments with small samples.

[0151] Step S3.2: In-model policy search

[0152] This step S3.2 randomly extracts states from the expert demonstration dataset as the initial states for policy optimization. In the process, a gradient-free optimization algorithm or model predictive control method is used to optimize the policy π to maximize the cumulative reward, and the reward function is updated by gradient ascent to generate a new policy π Further feedback to the real environment to generate new data , forming a data closed loop. Specifically:

[0153] Randomly sample states from the expert trajectory , as the initial state of the strategy optimization in the model, simulating the starting point of expert behavior. In the current reward function Optimization strategy , maximize the cumulative reward:

[0154]

[0155] Among them, π: strategy, which defines the agent's behavior in a given state Next select action The probability distribution of : The new strategy optimized at the t+1th iteration; : The trajectory ξ is formed by The state-action sequence is composed of In the environmental dynamic model The simulation sampling is generated; ξ represents a complete trajectory path, including all steps from the initial state to the final state; : The learned dynamic model of the environment is used to simulate the state transition process (i.e., predict the next state based on the current state and action); : In the tth iteration, the reward function is applied to the state at the hth step of the trajectory and actions The score is used to measure the quality of the decision of this step; H: trajectory length, that is, the maximum number of steps or rounds in the simulation path, indicating that a trajectory contains at most H state-action pairs (from arrive ; : In the model Under this circumstance, the trajectory obtained by sampling the strategy π Expected value, used to calculate the long-term average effect of statistics such as cumulative rewards; : Optimization objective is to select the strategy that maximizes the expected cumulative reward from all possible strategies π, that is, to find the optimal strategy.

[0156] Gradient-free optimization algorithms (such as cross entropy method, proximal policy optimization) or model predictive control methods can be used to efficiently iterate strategies in the model and reduce frequent dependence on the real environment. Low-cost iterative strategies in the virtual environment can reduce the number of real experiments and be used through reward functions. Dynamically guided strategies approach expert behavior.

[0157] Step S3.3: Reward function update

[0158] Optimize the reward function through gradient ascent for regret-free learning, so that it gradually approaches the expert strategy:

[0159]

[0160] in, : The current reward function, used in the tth iteration; : The reward function after the update at the t+1th iteration; : Learning rate, used to control the update step size of the reward function; : Find the gradient of the reward function f; :In the reward function Next, expert strategy Expected cumulative reward of :In the reward function Next, the current strategy Expected cumulative reward of : Expert strategies, usually derived from human experience or high-quality demonstrations; : The current strategy, that is, the strategy optimized in the tth iteration.

[0161] Optionally, after preliminary experiments, when the cumulative reward of strategy π When the improvement rate is less than 1% for 5 consecutive iterations, and the contribution of further optimization to the improvement of iron content is less than 0.5%, the optimization is terminated.

[0162] By continuously correcting the reward function, the learner's strategy can continuously approach the expert behavior, realizing the closed-loop reinforcement learning process of "reward function update → strategy optimization → model correction".

[0163] The optimized strategy is applied to the real training system, new trajectory data is collected, the new data is screened for effectiveness, and trajectories with high rewards and high diversity are retained for further training of the model and reward function. The final output obtained through the experiment is:

[0164] The optimal culture formula is recommended: 78% sawdust + 20% wheat bran + 1% lime + 1% gypsum + 2.0g / kg ferrous sulfate (60% water content), with a predicted iron content of 221mg / kg.

[0165] Recommended environmental parameters: The initial temperature is 25-28℃, the middle and late temperature is 22-24℃, and the humidity outside the ear is 85-90%.

[0166] This step, S3, builds model-based inverse reinforcement learning, nesting and iterating environment modeling (step 3.1), strategy optimization (step 3.2), and reward updating (step 3.3). This allows for low-cost trial-and-error in a virtual environment (reducing the need for real-world breeding experiments) and self-correction of the reward function (avoiding the subjectivity of manually designed rewards). Furthermore, inverse reinforcement learning is deeply integrated with the dynamics model, forming a self-iterative decision-making loop, enabling intelligent prediction and autonomous optimization of black fungus cultivation conditions.

[0167] Step S4: Construct a multi-index comprehensive scoring system based on the hierarchical analysis method to identify abnormal samples and trigger model backtracking adjustments, and output high-performance strains for the next round of breeding.

[0168] This step S4 comprehensively evaluates the performance of candidate strains in terms of iron content, yield, and stress resistance (such as growth period), selects high-performance strains, and retrospectively adjusts the model strategy when the standards are not met to improve breeding efficiency. Step S4 implements the strain performance evaluation and iteration process specifically including:

[0169] Step S4.1: Construction of a multi-indicator comprehensive evaluation system

[0170] The analytic hierarchy process (AHP) was used to assign weights to each indicator: iron content accounted for 50%, yield accounted for 30%, and growth period (representing stress resistance) accounted for 20%. The comprehensive score of each strain was calculated as follows:

[0171] Comprehensive score = 0.5 × iron content score + 0.3 × yield score + 0.2 × growth period score

[0172] Each score is normalized. Iron content score / yield score = (measured value - minimum value) / (maximum value - minimum value) × 100. For example, for the iron content score, the minimum value = 100 mg / kg and the maximum value = 250 mg / kg.

[0173] Growth period score = 1 - |(growth period length of the strain to be evaluated - preset optimal growth period length) / preset optimal growth period length|.

[0174] Step S4.2: Automatic identification and elimination of abnormal samples

[0175] The following samples will be identified as abnormal and automatically removed:

[0176] (1) Iron content is lower than the average level of the parents

[0177] (2) Iron content is lower than the target threshold of 194.3 mg / kg

[0178] (3) Abnormal reproductive period (e.g. deviation from the normal range ±15%)

[0179] (4) Missing indicators or obvious data errors

[0180] The elimination logic is: if the iron content is <194.3 mg / kg or lower than the average value of the parent strain → eliminate the strain sample.

[0181] Step S4.3: Iterative optimization and model backtracking adjustment

[0182] If most strains do not meet the criteria in a round of evaluation (e.g., the overall score is lower than 0.6 or the iron content does not meet the criteria), the system will automatically go back to the previous step (e.g., parent selection or hybrid design) and perform the following optimizations:

[0183] (1) Increase the weight of high-strength iron markers and enhance the selection tendency for high-strength iron traits;

[0184] (2) Rescreen the parents and eliminate those with too low basic iron content;

[0185] (3) Update the parameters of the genetic algorithm or recommendation model to optimize the prediction quality of mating combinations.

[0186] The parameter adjustment logic can be expressed by adding a penalty term (effective only when the iron content does not meet the standard). That is, if the strain's compliance rate is lower than the threshold, the weight of the high iron trait is increased or a penalty term is added. The total loss is calculated as follows:

[0187] Total loss = original loss + penalty coefficient × (average iron content of parents - actual iron content) 2

[0188] Among them, the penalty coefficient is dynamically adjusted according to the degree of non-compliance with the iron content.

[0189] According to step S4.3, when the comprehensive score of the strain is lower than the preset threshold or the iron content does not meet the standard, the system automatically backtracks to the parent selection or hybridization design stage and updates the parameters of the genetic algorithm or recommendation model to optimize the prediction quality of the mating combination.

[0190] Step S4.4: Output and next round of evaluation

[0191] The top 10% of strains ranked by overall score are retained as excellent candidates for the next round of testing; the remaining samples are used for model training or eliminated. The system records and outputs the following information: a distribution chart of scores for each round; a list of excluded samples and the reasons for their exclusion; and a record of adjustments to model parameters and weights to facilitate tracking of optimization paths.

[0192] The above is an introduction to the method embodiment. The following further illustrates the solution of the present invention through a system embodiment.

[0193] Figure 2 This is a module diagram of a black fungus high iron strain hybridization breeding optimization system based on big data analysis according to an embodiment of the present invention. Figure 2 As shown, a black fungus high iron strain hybridization breeding optimization system 200 based on big data analysis of the present invention includes:

[0194] Data collection and preprocessing module 210: collects basic strain data, hybridization process data, culture environment parameters and external reference data, and performs standardization and structured processing;

[0195] Hybrid combination optimization module 220: uses a biased random key genetic algorithm to encode and optimize hybrid combinations, and combines a dual elite evolution mechanism and a dynamic parameter adjustment strategy to generate an optimal hybridization scheme;

[0196] Culture condition prediction module 230: constructing an environmental dynamic model through a model-based hybrid inverse reinforcement learning algorithm to search for the optimal culture strategy that maximizes iron content in the virtual environment;

[0197] Strain performance evaluation module 240: Constructs a multi-index comprehensive scoring system based on the hierarchical analysis method, identifies abnormal samples and triggers model backtracking adjustments, and outputs high-performance strains for the next round of breeding.

[0198] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0199] Furthermore, based on the above-mentioned black fungus high iron strain hybridization breeding optimization method based on big data analysis and the optimal cultivation strategy determined by the system, the present invention conducted breeding experiments and the test results are as follows:

[0200] 1. The first round of high iron hybrid strain creation: The wild black fungus strain A356 with high iron content (iron content 161 mg / kg) and the high-yield cultivated strain A592 were used as parents. Spores were collected after the ears emerged and single spores were isolated to obtain monokaryotic strains. The authenticity of the monokaryotic strains was verified by optical microscopy observation of the presence of lock-like unions in the mycelium ( Figure 3 and Figure 4 ), 18 monokaryotic strains were prepared from the two parents, and 81 hybrid combinations were configured. After antagonistic reaction with the two parents ( Figure 5 and Figure 6 ), 36 true hybrid strains were screened and subjected to ear test. The iron content of the fruiting bodies was tested and shown in Table 1. The strain AA48 had the highest iron content, with a content of 173 mg / kg.

[0201] Table 1 Iron content of strains in the first round of hybridization

[0202]

[0203] 2. The second round of high-iron hybrid strain creation: The selected high-iron hybrid strain AA48 was used as the parent, spores were collected, and monokaryotic strains were obtained. After the authenticity of the monokaryotic strains was identified, a total of 53 monokaryotic strains were prepared, and backcross tests were carried out with the parent A356-4 of the hybrid strain AA48 to obtain 24 hybrid strains. After antagonistic reactions with the two parents, 17 authentic hybrid strains were screened, and ear tests were carried out to detect the iron content and agronomic traits of the fruiting bodies (Tables 2 and 3). Finally, the hybrid strain JAUA628 was screened out with the highest iron content of 194.3 mg / kg, which was 20% higher than the parent.

[0204] Table 2 Iron content of the second round of hybridization strains

[0205]

[0206] Table 3 Agronomic characteristics of hybrid strains

[0207]

[0208] 3. Culture material formula screening:

[0209] 1) Experimental Treatment: The iron ion concentration of the culture medium was set at a gradient of 1.0 g / kg, 1.5 g / kg, 2.0 g / kg, and 2.5 g / kg. Ferrous sulfate solutions of varying concentrations were added to the traditional culture medium (78% sawdust, 20% wheat bran, 1% lime, and 1% gypsum). The specific iron ion concentration gradients for the culture medium are shown in Table 4. Ferrous sulfate of the appropriate concentration was weighed and dissolved in the mixing water. Once completely dissolved, it was then mixed into the culture medium, mixed thoroughly, and bagged.

[0210] Table 4 Details of iron ion addition at different concentrations

[0211]

[0212] 2) Bag Preparation and Cultivation: Use 17×33cm polyethylene plastic bags and sterilize them by autoclaving for 2 hours. Allow the bags to cool to room temperature before inoculating with 10-15g of inoculum per bag. The bag temperature for incubation should be 25-28°C in the early stages and 22-24°C in the middle and late stages. Once the bags are full, allow 10 days for maturation.

[0213] 3) Management of ear emergence: punch holes to promote germination, temperature 15-20℃, relative humidity 70-80%, and start ear emergence management when primordia appear at the punched area. The temperature in the greenhouse should be 15-24℃, relative humidity 85-90%, spray water 2-3 times a day to maintain humidity, and expose to scattered light. When the ear pieces mature, the outer edge becomes thinner and a small amount of spores are ejected, which means the ear pieces are mature and can be harvested.

[0214] 4) Determination of iron content in the fruiting body of strain JAUA628

[0215] The iron content of the fruiting bodies of strain JAUA628 in different treatment groups and the control group was determined by inductively coupled plasma optical emission spectrometry.

[0216] 5) Screening of iron-rich culture medium formula for strain JAUA628

[0217] Based on the iron content of the fruiting bodies in the different treatment groups and the control group, the optimal culture medium formula for iron supplementation was screened. The results showed that the iron content of the fruiting bodies of strain JAUA628 increased with increasing iron concentration in the culture medium, reaching a maximum iron content of 221 mg / kg at a concentration of 2.0 g / kg (Table 5). Therefore, the optimal iron-enriched culture medium formula for strain JAUA628 is 78% sawdust, 20% wheat bran, 1% lime, 1% gypsum, 60% water content, and 2.0 g / kg ferrous sulfate (60% water content).

[0218] Table 5 Iron content in fruiting bodies of strain JAUA628

[0219]

[0220] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the methods.

[0221] It should also be noted that, in the embodiments of the present application, relational terms such as first and second, etc. are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device comprising the elements.

[0222] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined in the embodiments of the present application may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown in the embodiments of the present application, but rather will conform to the widest scope consistent with the principles and novel features disclosed in the embodiments of the present application.

Claims

1. A method for optimizing hybridization and breeding of black fungus high-iron strains based on big data analysis, characterized in that: include: Step S1: Collect basic strain data, hybridization process data, culture environment parameters and external reference data, and perform standardization and structured processing; Step S2: using a biased random bond genetic algorithm to encode and optimize the hybrid combination, and combining the dual elite evolution mechanism and dynamic parameter adjustment strategy to generate the optimal hybridization scheme; Step S3: constructing an environmental dynamic model through a model-based hybrid inverse reinforcement learning algorithm to search for the optimal cultivation strategy that maximizes iron content in the virtual environment; Step S4: Construct a multi-index comprehensive scoring system based on the hierarchical analysis method to identify abnormal samples and trigger model backtracking adjustments, and output high-performance strains for the next round of breeding.

2. The black fungus high iron strain hybridization breeding optimization method based on big data analysis according to claim 1, characterized in that, The implementation of the biased random key genetic algorithm in step S2 includes: Encode the hybrid combination as a random key chromosome vector X represented by n-dimensional continuous real numbers; The random bond chromosome vector X is decoded into a specific combination strategy and fitness score by an encoder, and an objective function is constructed with the iron content super-parent rate and agronomic trait stability as the evaluation core; By introducing parental gene expression data, we simulated the gene recombination behavior when different monokaryotic strains merged, predicted the probability distribution of iron content in the offspring, and formed a dynamic hybridization simulation model. Adopting the dual elite evolution mechanism, each generation retains the top Elite individuals are selected and biased crossover is performed with probability p>0.5: the offspring genes inherit the elite parent 1 gene with probability p and the non-elite parent 2 gene with probability 1-p; As the number of iterations advances, parameters such as the elite ratio and mutation probability are dynamically adjusted to adapt to the search progress.

3. The black fungus high iron strain hybridization breeding optimization method based on big data analysis according to claim 2, characterized in that, in, The biased random key genetic algorithm also includes: Dithering mechanism: Probability for elite individuals Apply disturbance; Reset mechanism: When there is no fitness improvement for the preset number of generations, non-elite individuals are cleared and random chromosomes are regenerated.

4. The black fungus high iron strain hybridization breeding optimization method based on big data analysis according to claim 3 is characterized in that, The objective function is: ； in, : The fitness score corresponding to chromosome X; : Iron content super-parent rate, that is, the ratio of the predicted iron content of the offspring to be higher than the average of the parents; : Agronomic trait stability index; α, β: objective function weight parameters, set according to task requirements.

5. The black fungus high iron strain hybridization breeding optimization method based on big data analysis according to claim 4 is characterized in that, The step S3: constructing an environmental dynamic model by using a model-based hybrid inverse reinforcement learning algorithm, and searching for an optimal cultivation strategy that maximizes iron content in the virtual environment, includes: Construct a hybrid dataset using expert demonstration data and learner online data; Use neural network to fit the environmental state transition probability to obtain the environmental dynamic model , and adopts strategy optimization and reward function iteration to form a closed-loop learning mechanism to efficiently simulate the culture medium formula and environmental condition adjustment behavior in a virtual model environment.

6. The black fungus high iron strain hybridization breeding optimization method based on big data analysis according to claim 5 is characterized in that, The step S3 further comprises: Randomly extract states from the expert demonstration dataset as the initial states for policy optimization; In the environmental dynamic model In the process, a gradient-free optimization algorithm or model predictive control method is used to optimize the policy π to maximize the cumulative reward, and the reward function is updated by gradient ascent to generate a new policy π Further feedback to the real environment to generate new data , forming a data closed loop.

7. The black fungus high iron strain hybridization breeding optimization method based on big data analysis according to claim 6 is characterized in that, Reward function iteration formula: ； : The current reward function, used in the tth iteration; : The reward function after the update at the t+1th iteration; : Learning rate, used to control the update step size of the reward function; : Find the gradient of the reward function f; :In the reward function Next, expert strategy Expected cumulative reward of :In the reward function Next, the current strategy Expected cumulative reward of : Expert strategies, usually derived from human experience or high-quality demonstrations; : The current strategy, that is, the strategy optimized in the tth iteration.

8. The black fungus high iron strain hybridization breeding optimization method based on big data analysis according to claim 1 is characterized in that, Step S4: constructing a multi-index comprehensive scoring system based on the analytic hierarchy process, identifying abnormal samples and triggering model retrospective adjustment, and outputting high-performance strains for the next round of breeding, including: The analytic hierarchy process was used to assign weights to each indicator: iron content accounted for 50%, yield accounted for 30%, and growth period accounted for 20%. The comprehensive score of each strain was calculated as follows: Comprehensive score = 0.5 × iron content score + 0.3 × yield score + 0.2 × growth period score Among them, each score is standardized; When the comprehensive score of the strain is lower than the preset threshold or the iron content does not meet the standard, the system automatically goes back to the parent selection or hybrid design stage and updates the parameters of the genetic algorithm or recommended model to optimize the prediction quality of the mating combination.

9. The method for hybridization and breeding optimization of black fungus high iron strains based on big data analysis according to claim 6, characterized in that: The step S4 further includes: If the strain compliance rate is lower than the threshold, the weight of the high-iron trait is increased or a penalty item is added. The total loss is calculated as follows: Total loss = original loss + penalty coefficient × (average iron content of parents - actual iron content) 2 Among them, the penalty coefficient is dynamically adjusted according to the degree of non-compliance with the iron content.

10. A black fungus high iron strain hybridization breeding optimization system based on big data analysis, characterized in that: include: Data collection and preprocessing module: collects basic strain data, hybridization process data, culture environment parameters and external reference data, and performs standardization and structured processing; Hybrid combination optimization module: uses biased random bond genetic algorithm to encode and optimize hybrid combinations, and combines dual elite evolution mechanism and dynamic parameter adjustment strategy to generate the optimal hybridization scheme; Culture condition prediction module: This module uses a model-based hybrid inverse reinforcement learning algorithm to build an environmental dynamic model and search for the optimal culture strategy that maximizes iron content in the virtual environment. Strain performance evaluation module: Build a multi-index comprehensive scoring system based on the hierarchical analysis method to identify abnormal samples and trigger model backtracking adjustments, outputting high-performance strains for the next round of breeding.

Images