Automatic optimization method for regulating and controlling key influence factors of gastrodia elata germination bacterium fermentation liquor

Abstract

The invention provides an automatic optimization method for regulating and controlling key influence factors of gastrodia elata germination bacterium fermentation liquor, which comprises the following steps: designing an online experimental platform to obtain an original experimental data set; a plurality of micro sensors are integrated on the reaction unit, and data acquisition, image analysis, feature extraction, physical and chemical parameter inference and microorganism multi-component detection and activity online analysis are performed to obtain multi-dimensional biological index data; the method comprises the following steps: collecting a fermentation liquor sample, and treating and analyzing the collected sample to obtain a microorganism-environment-metabolism collaborative network; a multi-target comprehensive evaluation function and a multi-target prediction model are constructed, influences of different factors are predicted, and an optimal parameter solution set and a factor and result interpretable report are obtained; and carrying out data storage and management, and carrying out model iterative optimization. Accurate control and efficient optimization of key influence factors of the fermentation liquor are achieved, experiments and fields are integrated, the production efficiency is improved, and large-scale expansion of gastrodia elata regulation and control is promoted.

Description

Technical Field

[0001] This invention relates to the field of Gastrodia elata growth analysis technology, and in particular to an automated optimization method for regulating key influencing factors in the fermentation broth of Gastrodia elata germination bacteria. Background Technology

[0002] Gastrodia elata is a perennial herb belonging to the genus Gastrodia in the Orchidaceae family. Its unique medicinal value has earned it the reputation of a precious traditional Chinese medicine. The rhizome is primarily used medicinally, possessing effects such as calming the liver and extinguishing wind, dispelling wind and unblocking meridians, and relieving pain. It is widely used in the treatment of headaches, dizziness, stroke, and epilepsy. The most distinctive feature of Gastrodia elata is that it has no green leaves, roots, or chloroplasts, resulting in an extremely low reproduction rate. Its seeds are very small and have incomplete endosperm development, relying on germination bacteria for germination and development. The fermentation broth of these germination bacteria, as an essential external factor for its growth, plays a fundamental role in the artificial propagation of Gastrodia elata. By identifying and controlling the key influencing factors of the fermentation broth, the germination rate and seedling survival rate can be improved, providing a technical foundation and guarantee for large-scale, standardized, and pollution-free production of Gastrodia elata.

[0003] Current mainstream control methods include single-factor experiments, orthogonal experiments, and molecular biology-assisted screening using machine learning modeling. However, existing methods have the following limitations: 1) Limited experimental scale and complexity: Single-factor and traditional orthogonal methods cannot fully reveal the complex relationships between multiple factors, resulting in insufficient optimization interactions and easy omission of optimal conditions. 2) Unstable results or difficulty in industrialization: Experimental conditions and results often vary greatly due to differences between laboratory and field environments, making it difficult to directly extend to large-scale field production. 3) Neglecting the effect of microbial communities: Focusing only on a single germination bacterium without considering the synergistic effects with other soil bacteria and environmental factors. 4) Single optimization objective or incomplete indicator system: Often focusing solely on germination rate, rarely considering multiple dimensions such as seedling health, stress resistance, and cost. 5) High cost of manual control and limited level of information management: Traditional manual optimization is time-consuming and labor-intensive, with low automation in data collection and analysis. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide an automated optimization method for regulating key influencing factors of the fermentation broth of Gastrodia elata germination bacteria. Through a high-throughput automated experimental platform, not only is precise control and efficient optimization of key influencing factors of the fermentation broth of Gastrodia elata germination bacteria achieved, but also integration of experimental field and experimental settings is realized, improving production efficiency and promoting the scalable regulation of Gastrodia elata.

[0005] To achieve the above objectives, the present invention provides the following solution: an automated optimization method for regulating key influencing factors in the fermentation broth of Gastrodia elata germination bacteria, comprising:

[0006] Design an online experimental platform to automatically add and mix fermentation broth, automatically control temperature, provide online pH feedback, perform AI imaging analysis, and preprocess samples to obtain the original experimental dataset;

[0007] Multiple micro-sensors are integrated on the reaction unit to perform data acquisition, image analysis, feature extraction, physicochemical parameter inference, and multi-component detection and online activity analysis of microorganisms, thereby obtaining multi-dimensional biological indicator data;

[0008] Fermentation broth samples were collected and subjected to standardization, multi-omics sequencing, metabolic analysis, data processing, and microbial community association analysis to obtain a microbial-environment-metabolic synergistic network.

[0009] By leveraging big data and machine learning, we construct a multi-objective comprehensive evaluation function and a multi-objective prediction model to predict the impact of different factors, obtain the optimal parameter solution set, and provide an interpretable report of factors and results.

[0010] A cloud-based distributed data center is constructed for data storage and management, and the multi-objective prediction model is iteratively optimized based on a knowledge graph.

[0011] Optionally, the online experimental platform includes:

[0012] The parameter setting and experiment design module is used to set fermentation broth culture parameters and experimental schemes. The parameter setting and experiment design module includes an AI experiment design tool and an experiment parameter input interface.

[0013] A high-throughput reaction module is used for the cultivation of fermentation broth of Gastrodia elata germination bacteria. The high-throughput reaction module includes a microporous array reactor and a QR code for identification. Multiple reaction units are arranged inside the microporous array reactor.

[0014] An automatic sample dispensing and mixing module is used for the automated dispensing and uniform mixing of reagents and bacterial solutions. The automatic sample dispensing and mixing module includes a multi-channel robotic arm, a high-precision pipetting system, and a micro-oscillator.

[0015] A temperature and pH control module is used for automated regulation of fermentation broth and stabilization of environmental parameters. The temperature and pH control module includes a temperature monitoring probe and a pH monitoring probe embedded in the reaction unit, as well as a temperature controller and an acid-base pump connected to the reaction unit.

[0016] An oxygen supply and stirring module is used to independently supply oxygen and magnetically stir each of the reaction units.

[0017] An imaging processing module is used to acquire biological features and perform automated sample preparation. The imaging processing module includes an imaging device, an AI image analysis server, an automatic liquid dispenser, a centrifuge, an automatic filtration device, and a refrigeration device.

[0018] The central control module is used for data acquisition and management. The central control module includes a multi-channel data acquisition card, a central control and data management server, a database server, and storage devices.

[0019] Optionally, the reaction unit includes an upper reaction hole, a gas-permeable but liquid-impermeable gas diffusion membrane disposed at the bottom of the upper reaction hole, and a gas cavity disposed at the bottom of the gas-permeable but liquid-impermeable gas diffusion membrane. A micro electromagnet array is disposed at the bottom of the reaction unit, and a micro gas passage interface is disposed in the gas cavity.

[0020] Optionally, an online experimental platform can be designed to automatically add and mix the fermentation broth, automatically control the temperature, provide online pH feedback, perform AI imaging analysis, and preprocess the samples to obtain the original experimental dataset, including:

[0021] Using the parameter setting and experimental design module, the selectable range and step size of the culture medium ratio and environmental parameters are input, and an experimental scheme is generated by combining orthogonal design, full factorial design, response surface methodology and AI prediction.

[0022] According to the experimental protocol, each reaction unit is numbered and managed, and various culture media, bacterial groups and additives are automatically aspirated and dispensed using the multi-channel robotic arm and the high-precision pipetting system, and then mixed and stirred evenly using the micro shaker.

[0023] The temperature and pH control modules are used to monitor and control the temperature and pH of each reaction unit in real time, and the imaging processing module is used to collect biological features and process samples to obtain real-time images and fermentation broth samples.

[0024] Based on the real-time images and the fermentation broth samples, the central control module is used to perform online data acquisition and uploading to obtain the original experimental dataset.

[0025] Optionally, multiple microsensors can be integrated into the reaction unit for data acquisition, image analysis, feature extraction, physicochemical parameter inference, and online analysis of multi-component microbial activity to obtain multi-dimensional bioindicator data, including:

[0026] Based on the aforementioned reaction unit, data is collected using a variety of micro sensors to obtain a multimodal raw time-series dataset. Based on the multimodal raw time-series dataset, the germination rate, vigor, proliferation rate, spatial distribution pattern, color, and health level of each reaction unit are extracted to obtain morphological physiological indicators.

[0027] Based on the multimodal original time-series dataset and the morphological physiological indicators, fermentation and growth curves are automatically fitted, stress responses and metabolic inflection points are identified, and stress resistance parameters, growth activity dynamics, fermentation metabolic state, stress occurrence time and response type are analyzed to obtain physicochemical and functional state parameters.

[0028] Based on the physicochemical and functional state parameters, multi-component detection and online activity analysis of microorganisms are performed to obtain comprehensive data on microbial diversity and activity. The comprehensive data on microbial diversity and activity are then aggregated and structured to obtain multi-dimensional biological indicator data.

[0029] Optionally, the microsensor includes an image sensor, a micro gas sensor array, a conductivity electrode, a pH microelectrode, a microbial component fluorescence analyzer, and a microfluidic biosampling nozzle. The multimodal raw time-series dataset includes real-time germination rate images, gas concentrations, and electrochemical parameters.

[0030] Optionally, fermentation broth samples are collected, and the collected samples are standardized, subjected to multi-omics sequencing, metabolic analysis, data processing, and microbial community association analysis to obtain the microbial-environment-metabolic synergistic network, including:

[0031] The fermentation broth sample was collected, and the time, experimental parameters, spatial location and batch information were collected to obtain the collected sample and spatiotemporal tag. The collected sample was then subjected to automated centrifugation, impurity removal, quantitative dilution and preservation and aliquoting to obtain standardized sample. The information of the standardized sample was then entered into the laboratory information management system.

[0032] For the standardized samples, 16S rRNA / ITS amplicon sequencing, metagenomic sequencing and metabolomics analysis were performed to obtain molecular-level raw data. Then, the molecular-level raw data was cleaned, fused and feature extracted to obtain a standard omics and phenotypic dataset.

[0033] Based on the aforementioned standard omics and phenotypic datasets, AI modeling and network analysis techniques are used to identify microbial communities, functional genes, and metabolites related to the multidimensional biomarker data, in order to automatically identify key regulatory factors and synergistic core microbial communities, thereby obtaining a microbial-environment-metabolic synergistic network.

[0034] Optionally, big data and machine learning can be used to construct a multi-objective comprehensive evaluation function and a multi-objective prediction model to predict the impact of different factors, obtain the optimal parameter solution set, and provide interpretable reports on factors and results, including:

[0035] By integrating the original experimental dataset, the multidimensional bioindicator data, and the standard omics and phenotypic dataset, a multi-factor multi-indicator feature sample set is obtained. Using the multi-factor multi-indicator feature sample set, a multi-objective comprehensive evaluation function and a multi-objective prediction model are designed.

[0036] Maximizing the multi-objective comprehensive evaluation function is set as the optimization objective. Based on the optimization objective, the optimal parameter solution set is automatically searched using the multi-objective prediction model and the multi-objective evolutionary algorithm. The optimal parameter solution set is then comprehensively sorted, and a factor and result interpretable report is generated.

[0037] The factor and result interpretable report includes recommended culture medium and environmental parameter configurations, predicted values ​​of each target performance, a list of key control factors and their sensitivity, supporting process risk warnings, interactive heatmaps, and decision logic flow.

[0038] Optionally, the original experimental dataset, the multi-dimensional biometric data, and the standard omics and phenotypic dataset are integrated to obtain a multi-factor, multi-indicator feature sample set. Using this multi-factor, multi-indicator feature sample set, a multi-objective comprehensive evaluation function and a multi-objective prediction model are designed, including:

[0039] Based on the original experimental dataset, the multidimensional bioindicator data, and the standard omics and phenotypic dataset, feature extraction, normalization, and redundancy removal are performed using statistical methods and cluster analysis to obtain a multi-factor, multi-indicator feature sample set.

[0040] Based on the aforementioned multi-factor, multi-indicator feature sample set, and combining environmental, functional, biological, and economic and microbiological indicators, a multi-objective comprehensive evaluation function is designed; the calculation expression of the multi-objective comprehensive evaluation function is as follows:

[0041] ;

[0042] in, For normalization function, For germination rate, For seedling health index, As a resilience index, This is the cost coefficient. This is an index of gut microbiota homeostasis. All are weighting coefficients;

[0043] A machine learning model is trained using a multi-factor, multi-index feature sample set, and a global sensitivity analysis is performed on the trained model to obtain a multi-objective prediction model.

[0044] Optionally, a cloud-based distributed data center is constructed for data storage and management, and the multi-objective prediction model is iteratively optimized based on a knowledge graph, including:

[0045] Build a cloud-based distributed data center to store data throughout the entire process, and automatically classify, label, trace, and standardize the data throughout the process, and complete data archiving and metadata management;

[0046] Construct a knowledge graph-driven expert knowledge base and causal network to obtain causal information between process formulations, phenotypic responses, omics characteristics, and environmental variables;

[0047] The multi-objective prediction model is iteratively optimized using the expert knowledge base and causal network, and an open API interface is built for multi-terminal interaction.

[0048] This invention discloses the following technical effects by providing an automated optimization method for regulating key influencing factors in the fermentation broth of Gastrodia elata germination bacteria:

[0049] 1. Automated Experiments: Through an automated online experimental platform, 1) functions such as automatic sample addition and mixing, automatic temperature control, online pH feedback, AI imaging analysis, and sample pretreatment are integrated, significantly improving experimental efficiency and automation, reducing human error, and achieving high-throughput, accurate data acquisition. 2) Through high-throughput reactions, real-time control and monitoring of all parameters, and intelligent experimental design, flexible design of experimental schemes and parameters, and fully automated execution of the entire process are achieved, providing high-quality, structured data for subsequent analysis. 3) It solves the problems of cumbersome traditional manual procedures, difficulty in scaling, and non-standard data.

[0050] 2. Multi-dimensional in-situ monitoring and data integration: 1) Integrating multiple micro-sensors into the reaction unit enables real-time, multi-modal data acquisition of all parameters of the fermentation broth, covering multiple indicators such as germination rate, vigor, proliferation rate, gas content, and electrochemistry. This allows for comprehensive, dynamic, and real-time characterization of fermentation and bacterial growth characteristics, yielding rich multi-dimensional biological data. 2) Through AI-assisted biofeedback extraction, physicochemical parameter inference, and online activity analysis, it supports automatic identification of growth and metabolic states and key physiological / chemical inflection points, assisting in high-throughput screening. 3) It breaks through the previous technical bottlenecks of offline, single, or low-throughput detection, greatly improving the depth and breadth of data.

[0051] 3. Precise Regulation and Standardized Multi-omics Analysis: 1) Enables automated / standardized preprocessing of collected samples and high-throughput integrated analysis of multiple omics (amplicon sequencing, metagenomics, metabolomics). This not only standardizes sample processing and ensures data standardization and traceability but also improves the accuracy of big data analysis. 2) Through automatic data entry into LIMS management and AI-driven analysis of the microbiome-environment-metabolism synergistic network, it enables automatic modeling of synergistic relationships between microbial diversity, function, and environmental parameters, directly driving the identification of regulatory mechanisms and key factors. 3) It addresses the shortcomings of traditional physiological and functional omics data fusion, effectively bridging the multi-layered mechanistic connections between phenotype, omics, and environment.

[0052] 4. Scientific and Reliable Solution: 1) Utilizing big data and machine learning methods, a multi-objective comprehensive evaluation function and predictive model are established to conduct factor sensitivity analysis, optimize parameter selection, and interpret results. This enables automatic balancing of multiple objectives and finding the global optimal solution under complex influencing factors, significantly improving design reliability and interpretability. 2) A multi-objective evolutionary algorithm is used to automatically search for the optimal parameter solution set and generate an interpretable report, efficiently completing adaptive parameter optimization and reducing reliance on manual experience. 3) Breaking through the limitations of traditional single-index optimization or static optimization, it achieves multi-objective, global, and dynamic system-level intelligent optimization.

[0053] 5. Intelligent Optimization Closed Loop of Experimental Results: 1) Through automated field trials, integrated feedback between the laboratory and the field is achieved, ensuring that the experimental optimization results can be truly transferred to field production, and improving the model's generalization and robustness. 2) By using algorithms such as transfer learning to dynamically correct and improve multi-objective prediction models, potential deviations or mismatches between the laboratory and the field can be quickly identified and corrected, forming a continuously adaptive intelligent optimization closed loop.

[0054] 6. Promote industrialization and large-scale development: By building a cloud-based distributed data center and knowledge graph, it is possible to achieve full-link, structured data management and knowledge accumulation, support continuous model upgrades and iterations and multi-terminal interoperability, and facilitate experience reuse, transparent decision-making and industrial collaboration.

[0055] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1This is a schematic diagram of the method flow provided in an embodiment of the present invention;

[0058] Figure 2 This is a schematic diagram of the online experimental platform architecture provided in an embodiment of the present invention;

[0059] Figure 3 This is a schematic diagram of the factor regulation process provided in an embodiment of the present invention. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] like Figure 1 As shown, this invention provides an automated optimization method for regulating key influencing factors in the fermentation broth of Gastrodia elata germination bacteria, including:

[0063] Step 1: Design an online experimental platform to automatically add and mix fermentation broth, automatically control temperature, provide online pH feedback, perform AI imaging analysis, and preprocess samples to obtain the original experimental dataset.

[0064] like Figure 2 As shown, the online experimental platform includes:

[0065] The parameter setting and experimental design module is used to set fermentation broth culture parameters and experimental schemes. This module includes an AI experimental design tool and an experimental parameter input interface. The AI ​​experimental design tool is software or a platform that uses artificial intelligence technology to assist researchers or related practitioners in designing experiments, improving the efficiency and accuracy of experimental design and reducing human error and time costs. Examples include InsCode AI IDE, Agent Laboratory, and scientific experiment assistants. The experimental parameter input interface can be provided via dedicated PC and mobile software or a web page.

[0066] A high-throughput reaction module is used for cultivating the fermentation broth of Gastrodia elata germination fungus. The module includes a microporous array reactor and a QR code identifier. The microporous array reactor contains multiple reaction units. Specifically, the microporous array reactor is a miniature chemical reaction device fabricated using microfabrication technology. Its core structure consists of a microporous array, primarily used for mixing and controlling chemical reactions. Through its advantages of high throughput, miniaturization, and precise control, the microporous array reactor can meet the research needs of optimizing conditions and analyzing components in the Gastrodia elata germination fermentation broth, significantly shortening the experimental cycle, reducing costs, and improving data reliability.

[0067] The automated sample dispensing and mixing module is used for the automated dispensing and uniform mixing of reagents and bacterial solutions. This module includes a multi-channel robotic arm, a high-precision pipetting system, and a micro-oscillator. The multi-channel robotic arm is an automated mechanical device integrating multiple operating terminals, such as robotic grippers, probes, and pipetting needles. Its core function is to improve operational efficiency through parallel operation of multiple channels, and it is widely used in laboratory automation. The high-precision pipetting system is an automated device specifically designed for the precise transfer of trace or ultra-trace volumes of liquids. Its core function is to reduce liquid transfer errors through precise structure and control technology, and it is a core tool in biochemistry and molecular biology experiments. The micro-oscillator is an instrument used for mixing laboratory samples, primarily for the rapid and uniform agitation of trace liquid samples, such as liquids in petri dishes, test tubes, centrifuge tubes, etc., and is suitable for clinical analysis, routine testing, and research.

[0068] The temperature and pH control module is used for automated regulation of the fermentation broth and stabilization of environmental parameters. This module includes a temperature monitoring probe and a pH monitoring probe embedded within the reaction unit, as well as a temperature controller and an acid / base pump connected to the reaction unit. The core functions of the temperature and pH monitoring probes are to accurately sense and report the temperature and pH values ​​in the environment, ensuring precise capture of these parameters. The temperature controller and acid / base pump are responsible for regulation, providing a stable temperature and pH environment for scenarios such as Gastrodia elata fermentation and chemical reactions, which is crucial for ensuring experimental accuracy.

[0069] An oxygen supply and stirring module is used to independently supply oxygen and magnetically stir each reaction unit. Each reaction unit includes an upper reaction orifice, a gas-permeable but liquid-impermeable gas diffusion membrane located at the bottom of the upper reaction orifice, and a gas chamber located at the bottom of the gas-permeable but liquid-impermeable gas diffusion membrane. A miniature electromagnet array is located at the bottom of each reaction unit, and a miniature gas passage interface is located within the gas chamber. Oxygen supply and stirring are achieved through the reaction units, as detailed below:

[0070] Gas diffusion membrane that is permeable to air but impermeable to liquid: High-efficiency PTFE / silicone rubber microporous gas diffusion membrane, 0.1~0.5mm thick, unidirectional gas permeability and liquid barrier, allowing oxygen to permeate evenly from the lower gas chamber without the disturbance of traditional bubbles, avoiding bubble shear damage and interface interference.

[0071] Gas chamber: It has a built-in micro gas interface, which intelligently distributes gas through a micro pump to achieve independent pressurized air intake through separate ports.

[0072] Miniature electromagnet array: 1-2mm miniature magnetic stirrers, each hole can be independently controlled for stirring by managing the magnetic field strength and rotation speed.

[0073] The imaging processing module is used to acquire biometric data and perform automated sample preparation. This module includes an imaging device, an AI image analysis server, an automated liquid dispenser, a centrifuge, an automated filtration device, and a refrigeration unit. Specifically, the imaging device is a high-resolution optical imaging element, such as a CSG14K image sensor or a CSG8K image sensor; the AI ​​image analysis server is an AI server specifically designed for image analysis tasks, typically equipped with acceleration hardware such as GPUs and FPGAs, and integrates optimization algorithms to support image processing tasks such as computer vision.

[0074] The central control module is used for data acquisition and management. The central control module includes a multi-channel data acquisition card, a central control and data management server, a database server, and storage devices.

[0075] Step 1 specifically includes:

[0076] 1.1 Using the parameter setting and experimental design module, input the selectable range and step size of the culture medium ratio and environmental parameters, and generate an experimental scheme by combining orthogonal design, full factorial design, response surface methodology, and AI prediction.

[0077] 1.2 According to the experimental scheme, each reaction unit is numbered and managed, and various culture media, bacterial groups and additives are automatically aspirated and dispensed using the multi-channel robotic arm and the high-precision pipetting system, and then mixed and stirred evenly using the micro shaker.

[0078] 1.3 The temperature and pH control modules are used to monitor and control the temperature and pH of each reaction unit in real time, and the imaging processing module is used to collect biological features and process samples to obtain real-time images and fermentation broth samples.

[0079] 1.4 Based on the real-time images and the fermentation broth samples, the central control module is used to perform online data acquisition and uploading to obtain the original experimental dataset.

[0080] Step 2, as follows Figure 3As shown, multiple micro-sensors are integrated into the reaction unit for data acquisition, image analysis, feature extraction, physicochemical parameter inference, and online analysis of multi-component microbial activity, yielding multi-dimensional bioindicator data. Specifically, this includes:

[0081] 2.1 Based on the reaction unit, data is collected using multiple micro sensors to obtain a multimodal raw time-series dataset. Based on the multimodal raw time-series dataset, the germination rate, vigor, proliferation rate, spatial distribution pattern, color, and health level of each reaction unit are extracted to obtain morphological physiological indicators.

[0082] The microsensors include an image sensor, a micro gas sensor array, a conductivity electrode, a pH microelectrode, a microbial component fluorescence analyzer, and a microfluidic biological sampling nozzle. The multimodal raw time-series dataset includes real-time germination rate images, gas concentrations, and electrochemical parameters.

[0083] 2.2 Based on the multimodal original time-series dataset and the morphological physiological indicators, fermentation and growth curves are automatically fitted, stress responses and metabolic inflection points are identified, and stress resistance parameters, growth activity dynamics, fermentation metabolic state, stress occurrence time and response type are analyzed to obtain physicochemical and functional state parameters.

[0084] 2.3 Based on the physicochemical and functional state parameters, multi-component detection and online activity analysis of microorganisms are performed to obtain comprehensive data on microbial diversity and activity. The comprehensive data on microbial diversity and activity are then aggregated and structured to obtain multi-dimensional biological indicator data.

[0085] Step 3, as follows Figure 3 As shown, fermentation broth samples were collected, and the samples underwent standardization, multi-omics sequencing, metabolic analysis, data processing, and microbial community association analysis to obtain a microbial-environment-metabolic synergistic network. Specifically, this included:

[0086] 3.1 Collect the fermentation broth sample and collect the time, experimental parameters, spatial location and batch information to obtain the collected sample and spatiotemporal tag. Then, perform automated centrifugation, impurity removal, quantitative dilution and preservation and packaging of the collected sample to obtain standardized sample. Finally, enter the information of the standardized sample into the laboratory information management system.

[0087] 3.2 For the standardized samples, 16S rRNA / ITS amplicon sequencing was performed to characterize the bacterial / fungal community structure, metagenomic sequencing was performed to explore all microbial genes and ecological functions, and metabolomics analysis was performed to determine the types and contents of metabolites to obtain molecular-level raw data. Then, the molecular-level raw data was cleaned, fused, and feature extracted to obtain a standard omics and phenotypic dataset.

[0088] 3.3 Based on the aforementioned standard omics and phenotypic datasets, AI modeling and network analysis techniques are used to identify microbial communities, functional genes, and metabolites related to the multidimensional biomarker data. This allows for the automatic identification of key regulatory factors and synergistic core microbial communities, resulting in a microbial-environment-metabolism synergistic network. Essentially, this synergistic network is a relationship map, showing the microorganisms, their environmental conditions, the metabolites they produce, and how they collectively influence the growth of *Gastrodia elata*. For example:

[0089] When the ethylene concentration in the environment is 0.5 ppm, *Pseudomonas* (a key microorganism) activates its acid-producing gene (functional gene), producing salicylic acid (a key metabolite), ultimately increasing the germination rate of *Gastrodia elata* by 20% and enhancing its drought resistance. This establishes the relationship: "ethylene → *Pseudomonas* → salicylic acid → *Gastrodia elata* growth".

[0090] Step 4, as follows Figure 3 As shown, a multi-objective comprehensive evaluation function and a multi-objective prediction model are constructed using big data and machine learning to predict the impact of different factors, obtain the optimal parameter solution set, and generate an interpretable report of factors and results. Specifically, this includes:

[0091] 4.1 Integrate the original experimental dataset, the multi-dimensional bioindicator data, and the standard omics and phenotypic dataset to obtain a multi-factor, multi-indicator feature sample set. Using this multi-factor, multi-indicator feature sample set, design a multi-objective comprehensive evaluation function and a multi-objective prediction model; specifically including:

[0092] 4.1.1 Based on the original experimental dataset, the multidimensional bioindicator data, and the standard omics and phenotypic dataset, feature extraction, normalization, and redundancy removal are performed using statistical methods and cluster analysis to obtain a multi-factor, multi-indicator feature sample set.

[0093] 4.1.2 Based on the aforementioned multi-factor, multi-indicator feature sample set, and combining environmental indicators, functional indicators, biological indicators, and economic and microbiological indicators across all dimensions, a multi-objective comprehensive evaluation function is designed to avoid the one-sidedness of using a single germination rate as the final objective. The calculation expression of the multi-objective comprehensive evaluation function is as follows:

[0094] ;

[0095] in, For normalization function, For germination rate, For seedling health index, As a resilience index, This is the cost coefficient. This is an index of gut microbiota homeostasis. All are weighting coefficients.

[0096] 4.2 A machine learning model is trained using a multi-factor, multi-index feature sample set, and a global sensitivity analysis is performed on the trained model to obtain a multi-objective prediction model. Cross-validation and multi-index regression evaluation are used to optimize the model's hyperparameters to ensure generalization performance. Global sensitivity analyses include SHAP, principal component analysis, and Sobol index.

[0097] 4.3 Maximize the multi-objective comprehensive evaluation function as the optimization objective, such as ensuring a germination rate > 90%, a health index > 0.85, and minimizing costs, among other constraints. Based on the optimization objective, automatically search for the optimal parameter solution set using the multi-objective prediction model and the multi-objective evolutionary algorithm. Perform a comprehensive ranking of the optimal parameter solution set and generate an interpretable report of factors and results.

[0098] The factor and result interpretable report includes recommended culture medium and environmental parameter configurations, predicted values ​​of each target performance, a list of key control factors and their sensitivity, supporting process risk warnings, interactive heatmaps, and decision logic flow.

[0099] Step 5: Construct a cloud-based distributed data center for data storage and management, and iteratively optimize the multi-objective prediction model based on a knowledge graph. Specifically, this includes:

[0100] 5.1 Construct a cloud-based distributed data center to store data throughout the entire process, and automatically classify, label, trace, and standardize the data throughout the process, and complete data archiving and metadata management.

[0101] 5.2 Construct a knowledge graph-driven expert knowledge base and causal network to obtain causal information between process formulations, phenotypic responses, omics characteristics, and environmental variables.

[0102] 5.3 The multi-objective prediction model is iteratively optimized using the expert knowledge base and causal network, and an open API interface is built to enable interaction between different clients.

[0103] Therefore, this invention provides an automated optimization method for regulating key influencing factors of the fermentation broth of Gastrodia elata germination bacteria. Through a high-throughput automated experimental platform, it not only achieves precise control and efficient optimization of key influencing factors of the fermentation broth of Gastrodia elata germination bacteria, but also realizes integrated experimental field operation, improves production efficiency, and promotes the scalable expansion of Gastrodia elata regulation.

[0104] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0105] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An automated optimization method for regulating key influencing factors in the fermentation broth of Gastrodia elata germination bacteria, characterized in that, include: Design an online experimental platform to automatically add and mix fermentation broth, automatically control temperature, provide online pH feedback, perform AI imaging analysis, and preprocess samples to obtain the original experimental dataset; Multiple micro-sensors are integrated on the reaction unit to perform data acquisition, image analysis, feature extraction, physicochemical parameter inference, and multi-component detection and online activity analysis of microorganisms, thereby obtaining multi-dimensional biological indicator data; Fermentation broth samples were collected and subjected to standardization, multi-omics sequencing, metabolic analysis, data processing, and microbial community association analysis to obtain a microbial-environment-metabolic synergistic network. By leveraging big data and machine learning, we construct a multi-objective comprehensive evaluation function and a multi-objective prediction model to predict the impact of different factors, obtain the optimal parameter solution set, and provide an interpretable report of factors and results. A cloud-based distributed data center is constructed for data storage and management, and the multi-objective prediction model is iteratively optimized based on a knowledge graph.

2. The automated optimization method for regulating key influencing factors of Gastrodia elata germination broth according to claim 1, characterized in that, The online experimental platform includes: The parameter setting and experiment design module is used to set fermentation broth culture parameters and experimental schemes. The parameter setting and experiment design module includes an AI experiment design tool and an experiment parameter input interface. A high-throughput reaction module is used for the cultivation of fermentation broth of Gastrodia elata germination bacteria. The high-throughput reaction module includes a microporous array reactor and a QR code for identification. Multiple reaction units are arranged inside the microporous array reactor. An automatic sample dispensing and mixing module is used for the automated dispensing and uniform mixing of reagents and bacterial solutions. The automatic sample dispensing and mixing module includes a multi-channel robotic arm, a high-precision pipetting system, and a micro-oscillator. A temperature and pH control module is used for automated regulation of fermentation broth and stabilization of environmental parameters. The temperature and pH control module includes a temperature monitoring probe and a pH monitoring probe embedded in the reaction unit, as well as a temperature controller and an acid-base pump connected to the reaction unit. An oxygen supply and stirring module is used to independently supply oxygen and magnetically stir each of the reaction units. An imaging processing module is used to acquire biological features and perform automated sample preparation. The imaging processing module includes an imaging device, an AI image analysis server, an automatic liquid dispenser, a centrifuge, an automatic filtration device, and a refrigeration device. The central control module is used for data acquisition and management. The central control module includes a multi-channel data acquisition card, a central control and data management server, a database server, and storage devices.

3. The automated optimization method for controlling key influencing factors of Gastrodia elata germination broth according to claim 2, characterized in that, The reaction unit includes an upper reaction hole, a gas-permeable but liquid-impermeable gas diffusion membrane disposed at the bottom of the upper reaction hole, and a gas cavity disposed at the bottom of the gas-permeable but liquid-impermeable gas diffusion membrane. A micro electromagnet array is disposed at the bottom of the reaction unit, and a micro gas passage interface is disposed in the gas cavity.

4. The automated optimization method for controlling key influencing factors of Gastrodia elata germination bacteria fermentation broth according to claim 3, characterized in that, Design an online experimental platform to perform automated sample addition and mixing of fermentation broth, automated temperature control, online pH feedback, AI imaging analysis, and sample pretreatment to obtain the original experimental dataset, including: Using the parameter setting and experimental design module, the selectable range and step size of the culture medium ratio and environmental parameters are input, and an experimental scheme is generated by combining orthogonal design, full factorial design, response surface methodology and AI prediction. According to the experimental protocol, each reaction unit is numbered and managed, and various culture media, bacterial groups and additives are automatically aspirated and dispensed using the multi-channel robotic arm and the high-precision pipetting system, and then mixed and stirred evenly using the micro shaker. The temperature and pH control modules are used to monitor and control the temperature and pH of each reaction unit in real time, and the imaging processing module is used to collect biological features and process samples to obtain real-time images and fermentation broth samples. Based on the real-time images and the fermentation broth samples, the central control module is used to perform online data acquisition and uploading to obtain the original experimental dataset.

5. The automated optimization method for controlling key influencing factors of Gastrodia elata germination broth according to claim 4, characterized in that, Multiple microsensors are integrated into the reaction unit for data acquisition, image analysis, feature extraction, physicochemical parameter inference, and online analysis of multi-component microbial activity, yielding multi-dimensional bioindicator data, including: Based on the aforementioned reaction unit, data is collected using a variety of micro sensors to obtain a multimodal raw time-series dataset. Based on the multimodal raw time-series dataset, the germination rate, vigor, proliferation rate, spatial distribution pattern, color, and health level of each reaction unit are extracted to obtain morphological physiological indicators. Based on the multimodal original time-series dataset and the morphological physiological indicators, fermentation and growth curves are automatically fitted, stress responses and metabolic inflection points are identified, and stress resistance parameters, growth activity dynamics, fermentation metabolic state, stress occurrence time and response type are analyzed to obtain physicochemical and functional state parameters. Based on the physicochemical and functional state parameters, multi-component detection and online activity analysis of microorganisms are performed to obtain comprehensive data on microbial diversity and activity. The comprehensive data on microbial diversity and activity are then aggregated and structured to obtain multi-dimensional biological indicator data.

6. The automated optimization method for controlling key influencing factors of Gastrodia elata germination broth according to claim 5, characterized in that, The microsensors include an image sensor, a micro gas sensor array, a conductivity electrode, a pH microelectrode, a microbial component fluorescence analyzer, and a microfluidic biological sampling nozzle. The multimodal raw time-series dataset includes real-time germination rate images, gas concentrations, and electrochemical parameters.

7. The automated optimization method for controlling key influencing factors of Gastrodia elata germination bacteria fermentation broth according to claim 6, characterized in that, Fermentation broth samples were collected and subjected to standardization, multi-omics sequencing, metabolic analysis, data processing, and microbial community association analysis to obtain the microbial-environment-metabolic synergistic network, including: The fermentation broth sample was collected, and the time, experimental parameters, spatial location and batch information were collected to obtain the collected sample and spatiotemporal tag. The collected sample was then subjected to automated centrifugation, impurity removal, quantitative dilution and preservation and aliquoting to obtain standardized sample. The information of the standardized sample was then entered into the laboratory information management system. For the standardized samples, 16S rRNA / ITS amplicon sequencing, metagenomic sequencing and metabolomics analysis were performed to obtain molecular-level raw data. Then, the molecular-level raw data was cleaned, fused and feature extracted to obtain a standard omics and phenotypic dataset. Based on the aforementioned standard omics and phenotypic datasets, AI modeling and network analysis techniques are used to identify microbial communities, functional genes, and metabolites related to the multidimensional biomarker data, in order to automatically identify key regulatory factors and synergistic core microbial communities, thereby obtaining a microbial-environment-metabolic synergistic network.

8. The automated optimization method for controlling key influencing factors of Gastrodia elata germination broth according to claim 7, characterized in that, Using big data and machine learning, a multi-objective comprehensive evaluation function and a multi-objective prediction model are constructed to predict the impact of different factors, obtain the optimal parameter solution set, and provide interpretable reports on factors and results, including: By integrating the original experimental dataset, the multidimensional bioindicator data, and the standard omics and phenotypic dataset, a multi-factor multi-indicator feature sample set is obtained. Using the multi-factor multi-indicator feature sample set, a multi-objective comprehensive evaluation function and a multi-objective prediction model are designed. Maximizing the multi-objective comprehensive evaluation function is set as the optimization objective. Based on the optimization objective, the optimal parameter solution set is automatically searched using the multi-objective prediction model and the multi-objective evolutionary algorithm. The optimal parameter solution set is then comprehensively sorted, and a factor and result interpretable report is generated. The factor and result interpretable report includes recommended culture medium and environmental parameter configurations, predicted values ​​of each target performance, a list of key control factors and their sensitivity, supporting process risk warnings, interactive heatmaps, and decision logic flow.

9. The automated optimization method for controlling key influencing factors of Gastrodia elata germination broth according to claim 8, characterized in that, By integrating the original experimental dataset, the multidimensional biometric data, and the standard omics and phenotypic dataset, a multi-factor, multi-indicator feature sample set is obtained. Using this multi-factor, multi-indicator feature sample set, a multi-objective comprehensive evaluation function and a multi-objective prediction model are designed, including: Based on the original experimental dataset, the multidimensional bioindicator data, and the standard omics and phenotypic dataset, feature extraction, normalization, and redundancy removal are performed using statistical methods and cluster analysis to obtain a multi-factor, multi-indicator feature sample set. Based on the aforementioned multi-factor, multi-indicator feature sample set, and combining environmental, functional, biological, and economic and microbiological indicators, a multi-objective comprehensive evaluation function is designed; the calculation expression of the multi-objective comprehensive evaluation function is as follows: ； in, For normalization function, For germination rate, For seedling health index, As a resilience index, This is the cost coefficient. This is an index of gut microbiota homeostasis. All are weighting coefficients; A machine learning model is trained using a multi-factor, multi-index feature sample set, and a global sensitivity analysis is performed on the trained model to obtain a multi-objective prediction model.

10. The automated optimization method for regulating key influencing factors of Gastrodia elata germination broth according to claim 9, characterized in that, Construct a cloud-based distributed data center for data storage and management, and iteratively optimize the multi-objective prediction model based on knowledge graphs, including: Build a cloud-based distributed data center to store data throughout the entire process, and automatically classify, label, trace, and standardize the data throughout the process, and complete data archiving and metadata management; Construct a knowledge graph-driven expert knowledge base and causal network to obtain causal information between process formulations, phenotypic responses, omics characteristics, and environmental variables; The multi-objective prediction model is iteratively optimized using the expert knowledge base and causal network, and an open API interface is built for multi-terminal interaction.

Images