A multi-agent system for developing functional microbial flora for water purification

Abstract

The application provides a functional microorganism flora development multi-agent system for water quality purification, and belongs to the technical field of sewage microorganism purification. The system comprises an engineering microorganism group identification agent, a microorganism agent design agent, a microorganism agent evaluation agent and an implementation scheme generation agent. The system generates an engineering microorganism group for degrading target pollutants in a target sewage plant through multi-agent cooperation, and then constructs a microorganism agent based on the engineering microorganism group and predicts and evaluates the biological purification effect of the microorganism agent. Furthermore, an implementation scheme for achieving the water quality purification target is determined according to the biological purification effect and the microorganism agent. The above process is time-saving and labor-saving, and the above microorganism agent can effectively degrade the target pollutants in the target sewage plant environment, thereby effectively improving the sewage treatment effect of the target sewage plant.

Description

Technical Field

[0001] This invention relates to the field of wastewater microbial purification technology, and in particular to a multi-agent system for developing functional microbial communities for water purification. Background Technology

[0002] Traditional wastewater treatment plants that use activated sludge systems for biological wastewater purification face the dilemma of being unable to meet wastewater treatment needs due to large load fluctuations and complex pollutant composition.

[0003] To address the aforementioned challenges, existing technologies often employ methods such as adjusting wastewater treatment process parameters or cultivating single bacteria that degrade pollutants. However, these methods are time-consuming, labor-intensive, and have limited effectiveness in improving wastewater treatment. Summary of the Invention

[0004] This invention proposes a multi-agent system for developing functional microbial communities for water purification. Through multi-agent collaboration, an engineered microbiome is first generated to degrade target pollutants in a target wastewater treatment plant. Then, microbial agents are constructed based on the engineered microbiome, and the biological purification effect of the microbial agents is predicted and evaluated. Furthermore, based on the biological purification effect and the microbial agents, an implementation plan for achieving the water purification target is determined. The above process is time-saving and labor-saving, and the aforementioned microbial agents can effectively degrade target pollutants in the environment of the target wastewater treatment plant, thereby effectively improving the wastewater treatment effect of the target wastewater treatment plant.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A multi-agent system for developing functional microbial communities for water purification includes an engineered microbiome identification agent, a microbial agent design agent, an agent evaluation agent, and an implementation scheme generation agent. The engineered microbiome identification agent identifies the engineered microbiome based on the water purification goals of the target wastewater treatment plant. These goals include the target water quality treatment objectives and target pollutants. The engineered microbiome comprises various functional microorganisms for degrading the target pollutants, as well as various metabolically complementary microorganisms. The microbial agent design agent identifies microbial agents based on the water purification goals and the engineered microbiome. These agents include various functional microorganisms and metabolically complementary microorganisms from the engineered microbiome that exhibit the best degradation effects on the target pollutants. The agent evaluation agent predicts and evaluates the biological purification effect of the microbial agents on the target wastewater treatment plant. The implementation scheme generation agent determines the implementation scheme for achieving the water purification goals based on the microbial agents and the biological purification effect. The implementation scheme includes the acquisition pathways and dosing strategies for the microbial agents.

[0007] In one implementation, identifying the engineered microbiome includes: acquiring domain knowledge related to water purification goals. The domain knowledge includes information on processes, water quality, pollutants, and microorganisms related to the water purification goals. Based on the water purification goals, the domain knowledge, and matching prompts, an engineered microbiome matching model built on a large language model is used to screen for engineered microbiomes from the domain knowledge. The matching prompts are used to guide the engineered microbiome matching model in screening for engineered microbiomes.

[0008] In one implementation, determining the microbial agent includes: generating the microbial agent from the engineered microbiome using a microbial agent design model constructed based on a large language model, according to the water purification target, the engineered microbiome, and agent design prompts. The agent design prompts are used to guide the microbial agent design model in generating the microbial agent.

[0009] In one implementation, for each functional microorganism among multiple functional microorganisms, the complementarity index between each metabolically complementary microorganism and the functional microorganism in the multiple metabolically complementary microorganisms of the functional microorganism is greater than the competition index. The formula for calculating the competition index is as follows:

[0010] ;

[0011] The formula for calculating the complementarity index is as follows:

[0012] ;

[0013] in, This refers to one type of microorganism among multifunctional microorganisms. This refers to another microorganism in a variety of metabolically complementary microorganisms. Indicates microorganisms With microorganisms Competition index between them Indicates microorganisms The collection of metabolites taken up from the environment Indicates microorganisms The collection of metabolites taken up from the environment; Indicates microorganisms With microorganisms The complementarity index between them Indicates microorganisms A collection of its own synthetic metabolites.

[0014] In one implementation, predicting and evaluating the biological purification effect of a microbial agent on a target wastewater treatment plant includes: determining a predicted value of the biological purification effect of the target wastewater treatment plant over a future period under the action of the microbial agent. The predicted biological purification effect includes the degradation rate of the target pollutant in the water quality purification target by the microbial agent. The formula for calculating the degradation rate of the target pollutant is as follows:

[0015] ;

[0016] in, Indicates the degradation rate of the target pollutant. Indicates the metabolic flux of the target pollutant. This indicates the dosage of microbial inoculant. The predicted biological purification effect is evaluated, and the evaluation result is obtained. When the predicted biological purification effect meets the water quality treatment target, the predicted biological purification effect is determined as the biological purification effect of the microbial inoculant on the target wastewater treatment plant. If the biological purification effect of the target wastewater treatment plant does not meet the water quality treatment target in the future, the matching prompts are updated based on the evaluation results, and the process returns to the step of selecting the engineered microbiome from the domain knowledge using an engineering microbial matching model built based on a large language model, based on the water quality purification target, domain knowledge, and matching prompts.

[0017] In one implementation, the microbial agent evaluation agent is further used to evaluate the ecological characteristics of the microbial community in the target wastewater treatment plant after the microbial agent is added to the plant. The ecological characteristics of the microbial community include: community stability, community robustness, species removal sensitivity, and the ability to recover from key pathway blockades.

[0018] In one implementation method, determining the implementation plan for achieving the water purification target includes: generating an implementation plan using an implementation plan generation model based on a large language model, based on microbial agents, biological purification effects, and implementation plan prompts; the implementation plan prompts are used to guide the implementation plan generation model to generate the implementation plan.

[0019] In one implementation, the system further includes a task coordination agent. This task coordination agent is used to regulate the execution order of the engineered microbiome identification agent, the microbial agent design agent, the agent evaluation agent, and the implementation plan generation agent.

[0020] In one implementation, the system further includes a knowledge management agent. This agent is used to determine domain knowledge relevant to the water purification target.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] In the multi-agent system for developing functional microbial communities for water purification provided by this invention, the engineered microbiome identification agent determines the engineered microbiome based on the water purification target of the target wastewater treatment plant. The microbial agent design agent then uses the water purification target and the engineered microbiome to determine the microbial agent containing the microbial community with the best degradation effect on the target pollutants. The agent evaluation agent predicts and evaluates the biological purification effect after the above-mentioned microbial agent is added to the target wastewater treatment plant. The implementation scheme generation agent, based on the above-mentioned microbial agent and the above-mentioned biological purification effect, derives an implementation scheme to achieve the water purification target. The collaboration among the four agents to derive the implementation scheme for achieving the water purification target is time-saving and labor-saving. Furthermore, the above-mentioned microbial agent can effectively degrade the target pollutants in the environment of the target wastewater treatment plant, thereby effectively improving the wastewater treatment effect of the target wastewater treatment plant. Attached Figure Description

[0023] Figure 1 This is one of the schematic diagrams of a multi-agent system for developing functional microbial communities for water purification provided in the embodiments of this application;

[0024] Figure 2 This is a logical flowchart of the engineered microbiome identification intelligent agent provided in the embodiments of this application;

[0025] Figure 3 This is the second schematic diagram of a multi-agent system for developing functional microbial communities for water purification provided in this application embodiment;

[0026] Figure 4 This is the third schematic diagram of a multi-agent system for developing functional microbial communities for water purification provided in this application embodiment;

[0027] Figure 5 This is the fourth schematic diagram of a multi-agent system for developing functional microbial communities for water purification provided in this application embodiment;

[0028] Figure 6 This is the fifth schematic diagram of a multi-agent system for developing functional microbial communities for water purification provided in this application embodiment;

[0029] Figure 7 This is a flowchart illustrating the logic of the microbial agent design intelligence provided in this application embodiment;

[0030] Figure 8 This is a flowchart of the logic of the microbial agent evaluation intelligent agent provided in the embodiments of this application;

[0031] Figure 9 This is the sixth schematic diagram of a multi-agent system for developing functional microbial communities for water purification provided in this application embodiment;

[0032] Figure 10 This is the seventh schematic diagram of a multi-agent system for developing functional microbial communities for water purification provided in this application embodiment;

[0033] Figure 11 This is the eighth schematic diagram of a multi-agent system for developing functional microbial communities for water purification provided in this application embodiment;

[0034] Figure 12 This is a logical flowchart of the task coordination agent provided in the embodiments of this application. Detailed Implementation

[0035] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0036] In the description of this invention, unless otherwise stated, "multiple" means two or more. For example, multiple microorganisms means two or more types of microorganisms.

[0037] The system provided in this application relates to biological wastewater purification treatment and can be used to generate biological wastewater purification treatment solutions.

[0038] To address the shortcomings of existing technologies, which often employ methods such as adjusting wastewater treatment process parameters or cultivating single bacteria capable of degrading pollutants, this application provides a multi-agent system for developing functional microbial communities for water purification. The system includes an engineered microbiome identification agent that identifies the engineered microbiome based on the water purification goals of the target wastewater treatment plant. A microbial agent design agent then uses the water purification goals and the engineered microbiome to identify microbial agents containing the microbial community with the best degradation effect on the target pollutants. An agent evaluation agent predicts and evaluates the biological purification effect after the microbial agents are added to the target wastewater treatment plant. An implementation scheme generation agent, based on the microbial agents and the biological purification effect, derives an implementation scheme to achieve the water purification goals. The collaboration among these four agents results in a time-saving and labor-saving process for deriving the implementation scheme to achieve the water purification goals. Furthermore, the microbial agents can effectively degrade the target pollutants in the environment of the target wastewater treatment plant, thereby effectively improving the wastewater treatment efficiency.

[0039] To better understand the technical solutions of the embodiments of this application, the technical terms involved in the embodiments of this application will be explained below.

[0040] 1. Intelligent agent (or artificial intelligence agent)

[0041] Artificial intelligence (AI) is a technological system that simulates human intelligence through algorithms and data. The four core technologies of AI are perception, reasoning, learning, and action. An AI agent, as a specific application of AI, is an intelligent entity capable of perceiving its environment, making autonomous decisions, and executing actions. Its goal is to complete specific tasks through interaction with the outside world.

[0042] Specifically, an intelligent agent typically comprises four core modules: a perception module, a decision-making module (also known as a technology module), an action module, and a memory module. The perception module is used to acquire information through data input (such as APIs) and transform that information into a understandable format. The decision-making module (e.g., a large language model) performs logical reasoning, task planning, or strategy generation based on the information acquired by the perception module. The action module executes the decision results (e.g., generating text, calling APIs). The memory module (e.g., a database, a knowledge base) is used to store data, supporting long-term reasoning and contextual understanding.

[0043] It should be noted that, for the intelligent agent in the embodiments of this application, the perception module of the intelligent agent can be either a large language model or other deep learning models, and the embodiments of this application do not limit it.

[0044] 2. Prompt words

[0045] Prompts are instructions or questions that users input into large language models (such as GPT-4, Claude, LLaMA, etc.) to guide the large language model to output specific types of response results.

[0046] As the core commands for user-model interaction, prompts directly influence the output quality and direction of large language models. To ensure a satisfactory output, prompts need to include task instructions, contextual information, input data, output format, and examples. If the output of a large language model is unsatisfactory after prompts are input, adjustments to the prompts can improve its performance.

[0047] The technical solutions of the embodiments of this application will be described in detail below.

[0048] like Figure 1As shown in the embodiments of this application, a multi-agent system for developing functional microbial communities for water purification is provided, including an engineered microbiome identification agent, a microbial agent design agent, an agent evaluation agent, and an implementation scheme generation agent.

[0049] The engineered microbiome identification agent is used to identify engineered microbiomes based on the water quality purification goals of the target wastewater treatment plant.

[0050] The aforementioned water quality purification targets include the water quality treatment targets and target pollutants of the target wastewater treatment plant (also known as the target sewage treatment plant). For example, the aforementioned water quality treatment targets may include target water quality parameters and target pollutant concentrations. The water quality treatment targets refer to the target water quality parameters and target pollutant concentrations of the target wastewater treatment plant under the action of the implementation scheme generated by the multi-agent system. The aforementioned target pollutants may include conventional pollutants, recalcitrant organic matter, and novel pollutants. The aforementioned conventional pollutants may include ammonia nitrogen, nitrate nitrogen, and total phosphorus. The aforementioned recalcitrant organic matter may include polycyclic aromatic hydrocarbons, phenols, dyes, and phthalates. The aforementioned novel pollutants may include personal care products, antibiotics, and polychlorinated biphenyls (PCBs). This application does not limit the type of the aforementioned target pollutants.

[0051] The aforementioned engineered microbiome includes multiple functional microorganisms for degrading target pollutants in the target wastewater treatment plant, as well as multiple metabolically complementary microorganisms. These multiple metabolically complementary microorganisms refer to multiple microorganisms whose metabolism is complementary to that of the multiple functional microorganisms in the engineered microbiome.

[0052] It should be noted that the subject of water purification using water quality microbial purification technology in this application embodiment is not limited to the above-mentioned sewage treatment plant, but may also be other equipment or factories. This application embodiment does not limit the above-mentioned subject.

[0053] In one implementation, such as Figure 2 and Figure 3 As shown, the engineered microbiome identified above includes S101-S102:

[0054] S101. Acquire domain knowledge related to water purification goals;

[0055] The domain knowledge includes information on processes, water quality, pollutants, and microorganisms related to water purification targets; optionally, the water quality information may include annual average chemical oxygen demand and annual average ammonia nitrogen concentration, etc.; the pollutant information may include the type and concentration of the pollutants; and the microbial information may include the name of the microorganism, degradation enzymes, degradation genes, and degradation reaction pathways, etc.

[0056] In one application scenario, such as Figure 4As shown, the information on processes, water quality, and pollutants in the above-mentioned domain knowledge can be retrieved from public databases related to wastewater treatment in wastewater treatment plants, or from literature data related to biological treatment of water purification; the information on microorganisms in the above-mentioned domain knowledge can be retrieved from public databases related to microorganisms, or from the whole genome data of various microorganisms contained in the actual obtained metagenomics of activated sludge.

[0057] For example, the literature data related to biological water purification mentioned above can be based on a knowledge base of 100,000 articles in the field of biological treatment of industrial wastewater and 2,000 articles on the biodegradation of pollutants; it can also be a large number of academic journal articles, academic conference reports and patents and other types of literature. The specific content of the literature data mentioned above is not limited in this application embodiment.

[0058] The aforementioned publicly available databases related to wastewater treatment in wastewater treatment plants may include the Web of Science website, the HydroWASTE database, the China Wastewater Treatment Plant Database, and CnOpenData, etc.; the aforementioned publicly available databases related to microorganisms may include the KEGG database, the enviPath database, the BIOCYC database, the MetOrigin database, the BiGG database, the Uniprot database, and the NCBI database, etc.; among them, the KEGG database is used to obtain pollutant degradation pathways and degradation genes, etc.; the enviPath database is used to obtain pollutant degradation pathways, degradation genes, and functional microorganisms, etc.; the BIOCYC database is used to obtain pollutant degradation pathways and degradation genes, etc.; the MetOrigin database is used to obtain pollutant degradation pathways and degradation functional microorganisms, etc.; the BiGG database is used to obtain microbial metabolic models, etc.; the Uniprot database is used to obtain amino acid sequences of degradation enzymes, etc.; the NCBI database is used to obtain microbial genome sequences, etc.; since the above databases are all publicly available databases, and obtaining data from the databases is a common technical means in this field, the process of obtaining the above data in this application embodiment will not be described in detail;

[0059] In another application scenario, the aforementioned domain knowledge can be acquired by searching for information on processes, water quality, pollutants, and microorganisms related to water purification targets in a biological domain knowledge base. For example, the aforementioned biological domain knowledge base includes data such as pollutant types, pollutant biodegradability, pollutant toxicity, pollutant degradation pathways, pollutant degradation chemical reaction formulas, pollutant degradation enzymes, pollutant degradation enzyme sequences, pollutant degradation functional genes, taxonomic information of pollutant degradation functional microorganisms, growth habits of degradation functional microorganisms, microbial genome-scale metabolic models, microbial metabolic complementarity and competition indices, microbial community assembly principles, and synthetic microbial community construction methods.

[0060] Understandably, the aforementioned biological knowledge base can be constructed by integrating relevant data from the aforementioned public databases related to wastewater treatment in wastewater treatment plants and the aforementioned public databases related to microorganisms through a mature large language model guided by prompt words (such as the Tongyi Qianwen-Omni-Turbo model); it can also be constructed by directly retrieving corresponding information from the aforementioned databases through manual or programming methods; in some cases, the information on the aforementioned microorganisms can also be obtained from the whole genome data of various microorganisms contained in the actual activated sludge metagenomics using a large language model; this application embodiment does not limit the construction process of the aforementioned biological knowledge base; the aforementioned data may include academic journal articles, academic conference reports, and patents, etc., and this application embodiment does not limit them;

[0061] Another application scenario, such as Figure 5 As shown, the above system also includes a knowledge management intelligent agent;

[0062] The aforementioned knowledge management agent is used to determine domain knowledge related to water purification goals;

[0063] In the above application scenarios, the above-mentioned domain knowledge can be acquired through the above-mentioned knowledge management intelligent agent; then the process of determining the domain knowledge related to the water purification target is as follows;

[0064] Based on the water purification goals and domain knowledge filtering prompts, a domain knowledge filtering model based on a large language model is used to determine the domain knowledge related to the water purification goals from the biological domain knowledge base.

[0065] The aforementioned domain knowledge screening prompts are used to guide the domain knowledge screening model in identifying domain knowledge relevant to water purification goals;

[0066] In some implementations, the domain knowledge filtering model invokes tools to determine the aforementioned domain knowledge, guided by domain knowledge filtering prompts; the tools invoked in the above process are described below.

[0067] Tool_Carveme: Tool_Carveme is used to construct genome-scale metabolic models. This tool uses the Carveme tool as input, taking a microbial whole genome .fa file as input, and outputting a genome-scale metabolic model .xml file for that species. It should be noted that Carveme is a commonly used tool in bioinformatics, jointly developed by the European Molecular Biology Laboratory and the Norwegian University of Science and Technology, and the software language is Python. In this embodiment of the application, the code after setting the usage parameters and command format of this tool is encapsulated into a backend web service and deployed on a server for the multi-agent system to call in the form of an application programming interface (API).

[0068] Tool_api: The Tool_api tool is mainly used to query information in target public data; the domain knowledge filtering model can automatically call the database API connection to query data as needed;

[0069] For example, the domain knowledge filtering prompts mentioned above are shown below;

[0070] Role Definition: You are an expert in wastewater biological treatment. You specialize in the microbial degradation of pollutants and bioinformatics; you are adept at locating, identifying, extracting, and organizing information related to pollutant degradation from massive amounts of literature, specifically but not limited to pollutant degradation pathways, degradation reaction formulas, degradation genes, functional microorganisms, and enzymes involved in pollutant degradation. You can use the Tool_api to query required information from public databases.

[0071] Key information extraction: Extraction based solely on original literature: All information must come directly from the original text, without adding any speculative content; Accuracy and completeness: Ensure that details such as microbial species, names, pollutant degradation pathways, and degradation genes are completely consistent with the original text;

[0072] Output requirements:

[0073] Please output the extracted information in the following JSON format:

[0074] {

[0075] "pollutants": "target pollutants"

[0076] "degradation_reaction": "degradation reaction pathway",

[0077] "degradation_gene": "degradation gene",

[0078] "degradation_enzyme": "degradation enzyme",

[0079] "Microbial species": "Microbial name" ...

[0080] }

[0081] It should be noted that the domain knowledge filtering model constructed based on the above-mentioned large language model refers to the domain knowledge filtering model that is trained based on the large language model. Therefore, the domain knowledge filtering model can be considered as a kind of large language model. Specifically, the domain knowledge filtering model is obtained by fine-tuning the pre-trained large language model with knowledge of this technical field. In the embodiments of this application, the large language model is a transformer-based large language model. The large language model can be the Tongyi Qianwen-Omni-Turbo model, the GPT series of large language models (GPT-3, GPT-4, and GPT-5, etc.), or the LLaMA series of large language models (LLaMA-1 and LLaMA-2, etc.). The embodiments of this application do not limit the specific type of the large language model.

[0082] S102. Based on water purification goals, domain knowledge, and matching prompts, an engineering microbiome is selected from the domain knowledge using an engineering microbiome matching model built on a large language model.

[0083] The aforementioned engineering microbial matching model based on a large language model refers to the engineering microbial matching model that is trained based on a large language model. Therefore, the engineering microbial matching model can also be considered a type of large language model. Specifically, the engineering microbial matching model is obtained by fine-tuning a pre-trained large language model using knowledge from this technical field. In this embodiment, the large language model is a transformer-based large language model. The large language model can be the Tongyi Qianwen-Omni-Turbo model, the GPT series of large language models (GPT-3, GPT-4, and GPT-5, etc.), or the LLaMA series of large language models (LLaMA-1 and LLaMA-2, etc.). This embodiment does not limit the specific type of the large language model.

[0084] The aforementioned engineered microbiome includes multiple functional microorganisms for degrading target pollutants in the target wastewater treatment plant, as well as multiple metabolic complementary microorganisms among these functional microorganisms. Specifically, for each functional microorganism among these functional microorganisms, the complementarity index between each metabolic complementary microorganism and the functional microorganism is greater than the competition index.

[0085] The formula for calculating the competition index is as follows:

[0086] ;

[0087] The formula for calculating the complementarity index is as follows:

[0088] ;

[0089] in, This refers to one type of microorganism among multifunctional microorganisms. This refers to another microorganism in a variety of metabolically complementary microorganisms. Indicates microorganisms With microorganisms Competition index between them Indicates microorganisms The collection of metabolites taken up from the environment Indicates microorganisms The collection of metabolites taken up from the environment; Indicates microorganisms With microorganisms The complementarity index between them Indicates microorganisms A collection of its own synthetic metabolites;

[0090] Understandably, the aforementioned matching prompts are used to guide the engineered microbial matching model in screening engineered microbiomes;

[0091] In some implementations, the engineered microbiome matching model, guided by matching prompts, invokes tools to filter engineered microbiomes from domain knowledge. The tools invoked in this process are described below.

[0092] Tool_blast: The Tool_blast tool is used to compare whether input microorganisms have the function of degrading target pollutants. Tool_blast takes microbial genome information as input and the comparison result as output. The specific implementation steps of the above tool are given below:

[0093] Step 1: First, use Prodigal software to identify the coding DNA sequence (CDS) of the microbial genome information. Based on the identified CDS sequence, obtain a collection file of all proteins (the amino acid sequence after CDS translation) in .aa format.

[0094] Step 2: Then, using the Blastp tool with certain sequence identity and alignment threshold E-value parameters, compare the amino acid sequence of the .aa file with that of the degrading enzyme and output the alignment results; an alignment result of 1 indicates that the microorganism has the target degradation function; an alignment result of 0 indicates that the microorganism does not have the target degradation function.

[0095] Specifically, the Tool_blast tool has built-in Prodigal and Blastp software, and encapsulates the Python code for the two steps of Prodigal to obtain CDS (Coding DNA Sequence) and Blastp to align CDS with degradation functional enzyme sequences into a backend web service, which is then deployed on a server and made available for multi-agent systems to call in the form of an application programming interface (API).

[0096] It should be noted that Prodigal and Blastp are widely used software in bioinformatics. Specifically, Prodigal is a software for predicting prokaryotic microbial genes, developed by Doug Hyatt's team at Oak Ridge National Laboratory and the University of Tennessee. This software can accurately identify the CDS sequences of gene fragments that can encode proteins from the massive base pairs in microbial genome files, and translate the CDS sequences into protein sequences, representing the set of protein sequences that the microorganism can compile. Blastp software was developed by the National Center for Biotechnology Information (NCBI) and is mainly used for comparing protein amino acid sequences.

[0097] In one application scenario, the knowledge management agent supplements the knowledge database with a metabolic model by calling the Tool_api and Tool_carveme tools; the knowledge management agent calls the Tool_api to obtain the microbial genome sequence from the NCBI database, and then calls the Tool_carveme tool to generate the corresponding metabolic model from the obtained microbial genome information, and then supplements the knowledge database with the metabolic model;

[0098] In one application scenario, an engineered microbiome identification agent calls the Tool_api and Tool_blast tools to identify functional microorganisms;

[0099] Specifically, the engineered microbiome identification agent calls the Tool_api to obtain the genome sequence of the microorganisms with degradation functions; the engineered microbiome identification agent calls the Tool_api to obtain the amino acid sequence of the degradation enzyme encoded by the degradation function gene;

[0100] Furthermore, the engineered microbiome identification agent calls Tool_blast to compare the microbial genome sequence and the amino acid sequence of the degradation enzyme to determine whether the compared microorganism has the function of degrading specific pollutants;

[0101] For example, the matching suggestions are shown below;

[0102] You are an expert in biological water purification. You are proficient in the degradation pathways, degradation genes, and functional microorganisms of pollutants. You can summarize and integrate fragmented information to construct degradation maps for different pollutants. You can choose to use the Tool_api tool to query required information from public databases. You can choose to use the Tool_blast tool for sequence alignment.

[0103] The microbial agent design intelligence is used to determine the microbial agent based on the water purification target and the engineered microbiome;

[0104] Optionally, such as Figure 6 and Figure 7 As shown, the aforementioned identified microbial inoculants include S201;

[0105] S201. Based on the water purification target, engineered microbiome, and microbial agent design prompts, a microbial agent design model based on a large language model is used to generate microbial agents from the engineered microbiome.

[0106] Among them, the above-mentioned microbial agents include multiple functional microorganisms and multiple metabolically complementary microorganisms in the engineered microbiome that have the best effect on degrading target pollutants; it can be understood that the multiple metabolically complementary microorganisms included in the above-mentioned microbial agents refer to multiple microorganisms that are metabolically complementary to the multiple functional microorganisms in the above-mentioned microbial agents.

[0107] It should be noted that the microbial agent design model constructed based on the large language model mentioned above refers to the microbial agent design model that was trained based on the large language model. Therefore, the microbial agent design model can also be considered as a kind of large language model. Specifically, the microbial agent design model was obtained by fine-tuning the pre-trained large language model using knowledge from this technical field.

[0108] Guided by the aforementioned microbial agent design prompts, the microbial agent design model employs a traversal approach to select multiple functional microorganisms and multiple metabolic microorganisms from the aforementioned engineered microbiome, forming multiple candidate microbial colonies. For each candidate microbial colony, the cooperative tradeoff flux balance analysis (ctFBA) method is first used to simulate the degradation performance of the candidate microbial community on the target pollutant under culture conditions where the target pollutant is the sole carbon source. Finally, the multiple functional microorganisms and multiple metabolically complementary microorganisms in the candidate microbial community with the best degradation performance are identified as the microbial agent.

[0109] It should be noted that the aforementioned cooperative tradeoff flux balance analysis (ctFBA) is a commonly used method in bioinformatics for simulating microbial community growth and metabolite flux. Similar methods include simplified metabolic flux balance analysis (pFBA) and dynamic metabolic flux balance analysis (dFBA). The cfFBA method allows for setting the tradeoff value to adjust whether the growth direction of the microbial community is optimized based on total biomass or the biomass of individual species. By adjusting different thresholds, it is possible to achieve favorable growth conditions for both degradation-functional microorganisms and metabolically complementary microorganisms in the microbial community.

[0110] For example, the culture medium used by the above-mentioned microbial agent design intelligence to simulate the microbial community is set with reference to the M9 culture medium. The carbon source can be set to use the target pollutant as the sole carbon source, or it can be set to conventional carbon sources such as glucose, maltose, or starch. This application does not specify the specific components and proportions of the culture medium.

[0111] It should be noted that the tradeoff value can be set to any value between 0 and 1. This application does not impose specific requirements on the tradeoff value setting. It is understood that the tradeoff value is an important parameter for simulating the performance of microbial communities in degrading target pollutants. It controls the direction of microbial community simulation and the growth of each species in the community. For example, when the threshold is close to 0, the optimization objective tends to maximize the individual biomass of each species. At this time, the growth rate of different species in the community tends to be balanced, which is conducive to maintaining high species diversity and metabolic functional complementarity.

[0112] The above-mentioned microbial agent design prompts are used to guide the microbial agent design model to generate microbial agents; for example, the above-mentioned microbial agent design prompts are shown below;

[0113] Role Definition: You are an expert in microbial community assembly with extensive experience in constructing synthetic microbial communities. You are proficient in constructing microbial communities using various advanced methods, with a particular focus on constructing microbial communities from a metabolic perspective.

[0114] Output requirements: Please output the simulation results in the following JSON format:

[0115] {

[0116] "pollutants": "pollutants",

[0117] "degradation_rate":"pollutant metabolic flux",

[0118] "microbial_inoculant":"

[0119] {

[0120] "sece_1":["Species 1"]

[0121] {

[0122] Kingdom:["boundary"],

[0123] Phylum:["door"],

[0124] Class:["Outline"],

[0125] " Order ":["order"],

[0126] " Family ":["family"],

[0127] "Genus":["genus"],

[0128] "Species": ["Species"]

[0129] }

[0130] "sepecie_2"::["Species 2"]

[0131] {

[0132] Kingdom:["boundary"],

[0133] Phylum:["door"],

[0134] Class:["Outline"],

[0135] " Order ":["order"],

[0136] " Family ":["family"],

[0137] "Genus":["genus"],

[0138] "Species": ["Species"]

[0139] } ...

[0140] }

[0141] The microbial agent evaluation agent is used to predict and evaluate the biological purification effect of the microbial agent on the target wastewater treatment plant.

[0142] In some implementations, such as Figure 8 As shown, the above prediction and evaluation of the biological purification effect of microbial agents on the target wastewater treatment plant includes S301-S302;

[0143] S301. Determine the predicted value of the biological purification effect of the target wastewater treatment plant under the action of microbial agents over a period of time in the future;

[0144] The above-mentioned predicted values ​​for biological purification effects include the degradation rate of target pollutants in water purification targets by microbial agents;

[0145] In one application scenario, the formula for calculating the degradation rate of the target pollutant is as follows:

[0146] ;

[0147] in, Indicates the degradation rate of the target pollutant. This represents the metabolic flux of the target pollutant, expressed in mmol / gDW / h. The dosage of microbial inoculant is expressed in gDW / L. The above formula can be used to calculate the rate at which a certain dosage of microbial inoculant degrades the target pollutant per unit time, expressed in mmol / L / h. It is understood that the above dosage of microbial inoculant... The value can be set according to the actual situation, based on past experience, or obtained through experiments. This application embodiment specifies the dosage of the aforementioned microbial agent. The value of is not limited;

[0148] Optionally, the predicted value of the above-mentioned biological purification effect may also include the change in the concentration of the target pollutant; the formula for calculating the change in concentration is as follows:

[0149] ;

[0150] in, This represents the change in concentration, expressed in mmol / L. t Degradation time, in hours (h).

[0151] In one implementation, the unit of the above concentration change can be converted from molar concentration to mass units; the calculation formula for the unit conversion is as follows:

[0152] ;

[0153] in, This represents the change in concentration, expressed in g / L. MW The molar mass of the pollutant is expressed in g / mol.

[0154] S302. Evaluate the predicted value of biological purification effect and obtain the evaluation result;

[0155] When the predicted value of biological purification effect meets the water quality treatment target, the predicted value of biological purification effect is determined as the biological purification effect of microbial agents on the target sewage treatment plant.

[0156] If the biological purification effect of the target wastewater treatment plant does not meet the water quality treatment target in the future, update the matching prompt words according to the assessment results and return to S102;

[0157] In one implementation, when the predicted value of the biological purification effect is the water quality parameter of the target wastewater treatment plant over a future period and the water quality treatment target is the target water quality parameter:

[0158] If the water quality parameters of the target wastewater treatment plant are basically equivalent to the target water quality parameters in the predicted future period, the assessment result is that the water quality purification target can be achieved. Then the predicted value of the biological purification effect is determined as the biological purification effect of the microbial agent on the target wastewater treatment plant.

[0159] If the predicted water quality parameters of the target wastewater treatment plant are greater than the target water quality parameters in the future, the assessment result is that the microbial agent has too great an impact on the water quality parameters. In this case, the phrase "the impact of the microbial agent on the water quality parameters needs to be smaller" will be updated in the matching prompt and the process will return to S102.

[0160] If the predicted water quality parameters of the target wastewater treatment plant are less than the target water quality parameters in the future, the assessment result is that the impact of microbial agents on water quality parameters is too small. In this case, the phrase "the impact of microbial agents on water quality parameters needs to be larger" will be updated in the matching prompts and returned to S102.

[0161] In one implementation, the microbial agent evaluation agent is also used to evaluate the ecological characteristics of the microbial community in the target wastewater treatment plant after the microbial agent is added to the target wastewater treatment plant (also known as agent stability); the microbial community ecological characteristics include: community stability, community robustness, species removal sensitivity, and key pathway blockade recovery capacity.

[0162] In one application scenario of the above implementation, the above evaluation process includes S3021-S3024;

[0163] S3021. Assess the community stability of microbial agents;

[0164] In one implementation, the stability of the microbial community is quantitatively assessed using ecological methods, including but not limited to indicators such as niche overlap.

[0165] In some implementations, a resource utilization profile is used to describe the intensity of different substrates used by various species in the community. The niche overlap index (such as the Pianka index) between species pairs is calculated using the following formula:

[0166] ;

[0167] in, Indicates species i With species j The degree of overlap in resource utilization; Indicates species i Resources r Utilization intensity; This represents the summation over all resources;

[0168] Furthermore, the average overlap of the communities is calculated. and complementarity ;

[0169] Understandably, lower overlap (higher complementarity) indicates a clear division of resources, which helps reduce competition and improves the stability of the community under environmental disturbances.

[0170] The aforementioned resource utilization profile can be obtained through metagenomic functional annotation or metabolic flux simulation, and this application does not limit it in this regard;

[0171] S3022. Evaluate the community robustness of microbial agents;

[0172] In one implementation, the robustness is assessed by simulating the community's response to species loss, key metabolic pathway blockage, and functional redundancy.

[0173] S3023, Species Removal Sensitivity (Unit Knockout);

[0174] The rate of change of pollutants was calculated by knocking out single species one by one in the community model. The robustness index is obtained; the formula for calculating the robustness index is as follows:

[0175] ;

[0176] in, Indicates the number of species removed; A sensitivity index indicating a single point of a species; This represents the relative change rate of pollutant degradation efficiency after removing the kth species; it should be understood that... The closer a value is to 1, the lower its dependence on a single species and the stronger the robustness of the community structure.

[0177] S3024, Key pathway blockade and recovery capability;

[0178] Flow constraint calculations were performed on key metabolic pathways identified in the community to determine the recovery rate of pollutant degradation efficiency after perturbation. R and recovery time Recovery rate R and recovery time The calculation formula is as follows:

[0179] ;

[0180] in, This indicates the pollutant degradation efficiency during the recovery phase after disturbance. This indicates the pollutant degradation efficiency before the disturbance; R Indicates the recovery rate; This indicates the time required for pollutant degradation efficiency to recover to 90% of its pre-disturbance level; Indicates time t The pollutant degradation efficiency under the given conditions; understandably, the recovery rate. R The higher the temperature, the longer the recovery time. The shorter the length, the stronger the metabolic compensation capacity and system robustness of the community when key functions are impaired;

[0181] It should be noted that the assessment of the ecological characteristics of microbial communities is not limited to the assessment methods described in S3021 to S3024, and this application does not impose any specific restrictions here.

[0182] It should be understood that in the above application scenarios, the water purification target also includes the expected value of the microbial community ecological characteristics; after obtaining the evaluation result of the microbial community ecological characteristics, if the evaluation result of the microbial community ecological characteristics does not meet the expected value of the microbial community ecological characteristics, then the current microbial agent does not meet the expected value of the microbial community ecological characteristics, so it is updated to the matching prompt words and returns to S102.

[0183] The implementation plan generates an intelligent agent to determine the implementation plan for achieving the water purification target based on the effects of microbial agents and biological purification.

[0184] Understandably, the intelligent agent that generates the above implementation scheme can also transmit the implementation scheme and the corresponding microbial agent combination to the knowledge management intelligent agent after generating the implementation scheme, and the knowledge management intelligent agent will store it in the biological domain knowledge base;

[0185] In one implementation, such as Figure 9As shown, the process of determining the implementation plan to achieve the water purification goal is as follows;

[0186] Based on the microbial agent, the biological purification effect, and the implementation scheme prompts, the implementation scheme is generated using an implementation scheme generation model built on a large language model.

[0187] It should be noted that the above-mentioned implementation scheme generation model based on the large language model refers to the implementation scheme generation model being trained based on the large language model. Therefore, the above-mentioned implementation scheme generation model can be considered as a kind of large language model. Specifically, the above-mentioned implementation scheme generation model is obtained by fine-tuning the pre-trained large language model using knowledge from this technical field.

[0188] The aforementioned implementation scheme prompts are used to guide the implementation scheme generation model to generate the implementation scheme; the aforementioned implementation scheme includes the acquisition methods and dosing strategies of microbial agents; specifically, the aforementioned dosing strategies include the dosage and frequency of microbial agents at different stages, and the aforementioned dosing strategies can be adjusted according to periodic operation monitoring and feedback control;

[0189] In some cases, the above implementation plan includes the acquisition methods of bacterial strains, the scale-up of bacterial agents, the strategy and frequency of bacterial agent dosing, operational monitoring and feedback control, and emergency and risk control measures under abnormal operating conditions. The acquisition methods of bacterial strains include obtaining functional microbial strains from legally qualified bacterial culture institutions, along with proof of origin, identification reports, and safety level descriptions. The bacterial agent dosing strategy includes phased dosing based on the recommended dosage calculated according to the treatment unit volume during the start-up period, maintenance dosing at a set frequency during the stable operation period, and triggering supplementary dosing when process performance declines or load fluctuates. Operational monitoring and feedback control include periodic monitoring of water quality indicators, process operating parameters, and microbial community structure, and dynamic adjustment of dosing and operating conditions based on monitoring results.

[0190] In some implementations, the microorganisms in the functional microbial agent can be obtained through the following means: purchasing from legally qualified strain collection centers (such as CGMCC, CCTCC, DSMZ, ATCC, etc.) with accompanying proof of strain origin and identification report; obtaining the target functional strain through laboratory isolation and purification, and identifying its degradation ability and safety through molecular biological methods; or obtaining proprietary strains authorized for use through cooperation with research institutions or enterprises.

[0191] In some embodiments, the above-mentioned functional microbial agents can be prepared as liquid concentrates (live count ≥10). 9 CFU / mL), lyophilized powder (live bacteria count ≥10¹) 0CFU / g) or carrier-based formulations (using alginate, bentonite, porous carbon, etc. as carriers) to meet the needs of different application scenarios; storage and transportation conditions can be cold chain transportation at 2~8℃, freeze-dried powder can be stored for 6~12 months under normal temperature and light protection conditions, and liquid formulations can be stored for 1~3 months; this application does not limit this.

[0192] For example, the prompts generated in the above decision report are shown below;

[0193] Role: You are a senior technical consultant with a background in microbial ecology, environmental engineering, and wastewater treatment technology. You are familiar with the design, evaluation, and engineering application of functional microbial agents and are skilled at integrating experimental data with ecological indicators to form technical feasibility analysis reports.

[0194] Report content requirements:

[0195] Based on the input information on functional microbial agents, biological purification effect data, and ecological assessment indicators, a complete "Microbial Purification Technology Solution Evaluation Report" is generated.

[0196] The report should include the following three main parts:

[0197] Part 1: Pollutant Degradation Performance

[0198] Describe the performance of the microbial agent in the removal of target pollutants, including degradation rate, degradation efficiency, and time required to reach the target removal rate;

[0199] The performance differences under different dosages and environmental conditions are presented.

[0200] Part Two: Ecological Indicators

[0201] Species diversity (such as Shannon index, number of species, etc.);

[0202] Community stability (e.g., niche overlap / complementarity indicators);

[0203] Community robustness (e.g., sensitivity to species removal, and ability to recover from pathway blockage).

[0204] Community metabolic interactions (complementary and cooperative metabolic functions among species).

[0205] Part Three: Technical Feasibility Analysis

[0206] Species acquisition channels (such as strain preservation centers, laboratory culture);

[0207] Feasibility of formulation and storage / transport (liquid concentrate, lyophilized powder, carrier-based formulation);

[0208] Dosing strategy and operation and maintenance requirements;

[0209] Cost and economic analysis (estimation of costs for microbial agent preparation, storage, transportation, and operation).

[0210] When producing reports, use the formal style of technical reports with clear paragraph structure; avoid overly colloquial language, and appropriately cite data and formulas to illustrate performance; begin each section with a summary conclusion, followed by supporting data or analysis.

[0211] It should be noted that the first part of the above-mentioned purification technology report and the second part of the above-mentioned purification report are mainly derived from the evaluation results of the above-mentioned microbial agent evaluation intelligence.

[0212] Corresponding to the above assessment of the ecological characteristics of microbial communities, such as Figure 10 As shown, the intelligent agent generated by the implementation scheme can also be used to determine the implementation scheme for achieving water purification goals based on microbial agents, biological purification effects, and the ecological characteristics of microbial communities.

[0213] Given the above circumstances, the implementation process of generating an intelligent agent to achieve the water purification goal is basically the same as the above implementation process. Therefore, the implementation process of the above circumstances will not be described in detail in the embodiments of this application.

[0214] In some implementations, such as Figure 11 As shown, the system also includes a task coordination agent;

[0215] The task coordination agent is used to regulate the execution order of the engineered microbiome identification agent, the agent evaluation agent, and the implementation plan generation agent.

[0216] In one application scenario, the aforementioned water quality microbial purification technology can also be used to develop a multi-agent system that redesigns microbial agents based on existing purification technologies to meet the needs of multiple water quality purification objectives, in order to address new water quality purification goals.

[0217] Specifically, after the implementation plan generating agent outputs the implementation plan to the user, the task coordination agent asks the user whether there are any other water quality purification goals. If the user has new water quality purification goals, the task coordination agent analyzes the new water quality purification goals and organizes the structured data to determine the workflow and the specific participating agents.

[0218] In the embodiments of this application, such as Figure 12 As shown, the workflow for the task coordination agent described above includes S401-S405.

[0219] S401. The microbial agent assessment agent assesses whether the current microbial agent can meet the new water purification requirements. If the requirements are met, the microbial agent assessment agent sends the assessment results to the implementation plan generation agent, which then regenerates the implementation plan, which includes the results of multiple water purification requirements.

[0220] S402. If the current microbial agent cannot meet the new water purification requirements, the agent evaluation agent will return the evaluation results to the engineered microbiome identification agent. The engineered microbiome identification agent will then combine the evaluation results to re-identify the engineered microbiome in order to supplement the required functional microorganisms and their metabolically complementary microorganisms.

[0221] S403. The engineered microbiome identification agent sends the re-identified engineered microbiome to the microbial agent design agent for redesign.

[0222] S404. The microbial agent design agent sends the design results to the agent evaluation agent to evaluate the purification effect of the agent on multi-target water quality purification needs and the corresponding microbial community ecological characteristics.

[0223] S405. The microbial agent assessment agent sends the regenerated assessment results to the implementation plan generation agent to generate an implementation plan that meets the multi-objective water purification requirements.

[0224] In summary, the multi-agent system for developing functional microbial communities for water purification provided in this application involves an engineered microbiome identification agent that identifies the engineered microbiome based on the water purification target of the target wastewater treatment plant. A microbial agent design agent then uses the water purification target and the engineered microbiome to identify microbial agents containing the microbial community with the best degradation effect on the target pollutants. An agent evaluation agent predicts and evaluates the biological purification effect after the aforementioned microbial agents are added to the target wastewater treatment plant. An implementation scheme generation agent, based on the aforementioned microbial agents and the biological purification effect, derives an implementation scheme for achieving the water purification target. The collaboration among these four agents to derive the implementation scheme for achieving the water purification target is time-saving and labor-saving. Furthermore, the aforementioned microbial agents can effectively degrade the target pollutants in the environment of the target wastewater treatment plant, thereby effectively improving the wastewater treatment effect of the target wastewater treatment plant.

[0225] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0226] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A multi-agent system for developing functional microbial communities for water purification, characterized in that, This includes intelligent agents for identifying engineered microbiomes, intelligent agents for designing microbial agents, intelligent agents for evaluating microbial agents, and intelligent agents for generating implementation plans. The engineered microbiome identification agent is used to determine the engineered microbiome based on the water quality purification goals of the target wastewater treatment plant; wherein, the water quality purification goals include the water quality treatment goals and target pollutants of the target wastewater treatment plant; the engineered microbiome includes multiple functional microorganisms for degrading the target pollutants in the target wastewater treatment plant, and multiple metabolically complementary microorganisms of the multiple functional microorganisms; determining the engineered microbiome includes: Acquire domain knowledge related to the water purification target; the domain knowledge includes information on processes, water quality, pollutants, and microorganisms related to the water purification target; Based on the water purification goals, the domain knowledge, and matching prompts, an engineering microbiome matching model built on a large language model is used to screen out the engineering microbiome from the domain knowledge; the matching prompts are used to guide the engineering microbiome matching model in screening the engineering microbiome. The microbial agent design intelligence is used to determine the microbial agent based on the water purification target and the engineered microbiome; the microbial agent includes multiple functional microorganisms and multiple metabolically complementary microorganisms in the engineered microbiome that have the best effect on degrading the target pollutant; for each functional microorganism, the complementarity index between each metabolically complementary microorganism and the functional microorganism is greater than the competition index. The formula for calculating the competition index is as follows: The formula for calculating the complementarity index is as follows: Wherein, A represents one of the multiple functional microorganisms, B represents another of the multiple metabolically complementary microorganisms, and MI competition (A,B) represents the competition index between microorganism A and microorganism B, SeedSet(A) represents the set of metabolites taken up by microorganism A from its environment, and SeedSet(B) represents the set of metabolites taken up by microorganism B from its environment; MI complementarity (A,B) represents the complementarity index between microorganism A and microorganism B, and NonSeedSet(B) represents the set of autosynthetic metabolites of microorganism B. The microbial agent evaluation agent is used to predict and evaluate the biological purification effect of the microbial agent on the target wastewater treatment plant. The implementation scheme generates an intelligent agent to determine an implementation scheme for achieving the water purification target based on the microbial agent and the biological purification effect; the implementation scheme includes the acquisition method and dosing strategy of the microbial agent.

2. The system as described in claim 1, characterized in that, The determined microbial inoculant includes: Based on the water purification target, the engineered microbiome, and the microbial agent design prompts, a microbial agent design model based on a large language model is used to generate the microbial agent from the engineered microbiome; the microbial agent design prompts are used to guide the microbial agent design model to generate the microbial agent.

3. The system as described in claim 1, characterized in that, The prediction and evaluation of the biological purification effect of the microbial agent on the target wastewater treatment plant includes: Determine the predicted value of the biological purification effect of the target wastewater treatment plant under the action of the microbial agent over a future period of time; the predicted value of the biological purification effect includes the degradation rate of the target pollutants in the water purification target by the microbial agent; The formula for calculating the degradation rate of the target pollutant is as follows: Degradation rate =F take ×X Among them, Degradation rate F represents the degradation rate of the target pollutant. take X represents the metabolic flux of the target pollutant, and X represents the dosage of microbial inoculant. The predicted value of the biological purification effect is evaluated to obtain the evaluation result; When the predicted value of the biological purification effect meets the water quality treatment target, the predicted value of the biological purification effect is determined as the biological purification effect of the microbial agent on the target sewage treatment plant. When the biological purification effect of the target wastewater treatment plant does not meet the water quality treatment target within the future period, the matching prompt words are updated according to the evaluation results, and the process returns to the step of selecting the engineered microbiome from the domain knowledge using an engineered microbiome matching model based on a large language model, based on the water quality purification target, the domain knowledge, and the matching prompt words.

4. The system as described in claim 1, characterized in that, The method for determining the implementation scheme to achieve the water purification target includes: Based on the microbial agent, the biological purification effect, and the implementation scheme prompts, an implementation scheme generation model based on a large language model is used to generate the implementation scheme; the implementation scheme prompts are used to guide the implementation scheme generation model to generate the implementation scheme.

5. The system as described in claim 1, characterized in that, The system also includes a task coordination agent; The task coordination agent is used to regulate the execution order of the engineered microbiome identification agent, the microbial agent design agent, the agent evaluation agent, and the implementation scheme generation agent.

6. The system as described in claim 1, characterized in that, The system also includes a knowledge management intelligent agent; The knowledge management agent is used to determine domain knowledge related to the water purification target.

Images