Marine ecological multi-agent construction method and interaction system thereof

Abstract

The invention relates to the technical field of marine ecology, in particular to a marine ecology multi-agent construction method and an interaction system thereof. The construction method comprises the following steps: a marine ecology large model construction step: constructing a fine adjustment data set, and training based on the fine adjustment data set to obtain a marine ecology large model; a parallel application module construction step: constructing a domain knowledge base and a marine ecological multi-intelligent body, constructing a multi-modal large model, and deploying the domain knowledge base, the marine ecological multi-intelligent body, the multi-modal large model and the marine ecological large model of the parallel application module in a GPU server, and the marine ecology large model is used for analyzing the user request and distributing the user request to the parallel application module, triggering the parallel application module to generate a response result, feeding the response result back to the marine ecology large model for semantic integration, and outputting the result. According to the invention, the intelligent level in the field of marine ecology is improved, the prediction capability of the model is enhanced, and the prediction result is more accurate and reliable.

Description

Technical Field

[0001] The present application relates to the field of marine ecological technology, and in particular to a method for constructing a marine ecological multi-agent and its interactive system. Background Art

[0002] As a global issue, marine ecological protection has attracted more and more attention, especially in the research and application of AI technology. However, the existing "AI + marine ecology" technology still has many shortcomings and fails to fully solve the key problems in the field.

[0003] In the field of "AI + marine ecology", traditional machine learning methods such as support vector machines (SVM), random forests (RF), and artificial neural networks (ANN) have been applied to the study of species distribution prediction, helping scientists predict the potential distribution of species, assess species richness, and explore the relationship between species habitat selection and landscape pattern changes. At the same time, image recognition technology has also been applied in species identification and quantity statistics, marine pollutant monitoring, etc. These AI technologies provide decision-making support for marine ecological resource management, ecosystem prediction, and marine environmental protection, showing increasing application potential.

[0004] However, although AI technology has made some progress in the field of marine ecological protection, its application and development still faces a series of problems and challenges: For example, data resources in the field of marine ecology are limited, and data islands are serious. The effective training of AI models often relies on large-scale, high-quality data. However, since the acquisition of marine ecological data requires a lot of field surveys, which is not only time-consuming and costly, the data scale is difficult to meet the training needs of AI models, which in turn affects the prediction accuracy and generalization ability of the model.

[0005] For example, the existing "AI+marine ecology" models mostly use traditional machine learning, convolutional neural networks (CNN) and long short-term memory networks (LSTM) and other methods. The technology is relatively backward and cannot cope with the complex and changing marine ecological environment. Especially when dealing with interference factors such as typhoons, storm surges and human activities, the accuracy and stability of the model are poor, and it fails to make full use of business databases and Internet data.

[0006] In addition, at present, AI models in sub-sectors such as marine biological distribution prediction, species identification, population analysis and pollutant monitoring are often implemented by multiple independent small models. These models lack data sharing and coupling, and cannot form a unified analysis framework, resulting in the inability to systematically develop "AI + marine ecology" applications, which greatly limits the progress of scientific research and industrial applications. Summary of the invention

[0007] The embodiments of the present application provide a method for constructing a marine ecological multi-intelligence body and its interactive system, aiming to solve key problems such as data scarcity, poor model accuracy and stability, and application fragmentation, and provide more efficient and accurate technical support for marine ecological protection.

[0008] In a first aspect, the present application provides a method for constructing a marine ecological multi-agent, including: The steps of constructing a large marine ecological model include constructing a fine-tuning dataset, and training the selected large language model based on the fine-tuning dataset to obtain a large marine ecological model; The parallel application module construction step includes constructing a domain knowledge base, a marine ecological multi-agent, and a multimodal large model. The domain knowledge base, the marine ecological multi-agent, the multimodal large model of the parallel application module and the marine ecological large model are deployed on the GPU server in a diversified parallel deployment mode. Among them, the knowledge data set of the domain knowledge base includes: marine environment data, human activity pressure data, the human activity pressure data includes: socio-economic data, human activity data, the marine ecological multi-intelligent body includes: pre-packaged basic functional intelligent body, Internet functional intelligent body, marine ecological functional intelligent body, among which the basic functional intelligent body is used to realize business database information retrieval, mathematical calculation, server infrastructure settings, operation and maintenance, etc. The Internet functional intelligent body is used to realize language translation, weather query, search engine, map service, paper retrieval, Wikipedia, etc. The marine ecological functional intelligent body is used to realize the functions in the field of fishery resources and marine ecological protection, and can support quantitative statistics, species identification, habitat simulation, species distribution prediction, pollutant monitoring, extreme weather prediction, ecological disaster warning and other functions; the multimodal large model includes a multimodal model, a text-based graph model, a text-speech model, and a text-based video model, which is used to support the multimodal large model to realize the mutual conversion and generation of visual question answering, text to picture, speech, and video; The marine ecological big model is used to parse user requests and distribute them to the parallel application module, triggering any one of the domain knowledge base, marine ecological multi-agent, and multimodal big model in the parallel application module or any combination thereof to generate response results that are fed back to the marine ecological big model for semantic integration and output.

[0009] Based on the above steps, in order to solve the problems of lagging existing model technology, poor model accuracy and stability, the embodiment of the present application constructs a large marine ecological model and adopts a large model parallel application architecture to improve the level of intelligence in the marine ecological field. The large marine ecological model is used to capture subtle changes and long-term trends that are difficult to capture with traditional models, thereby improving the prediction accuracy of the model. The parallel application architecture can support larger-scale data processing and more complex model training, further enhance the prediction ability of the model, and make the prediction results more accurate and reliable. The parallel application architecture can make full use of the computing resources of multiple computing nodes to realize parallel training and reasoning of the model, thereby significantly shortening the time of model training and reasoning, and improving the response speed of the model.

[0010] In some embodiments, the process of constructing the fine-tuning dataset includes: A data acquisition step, acquiring documents in the field of marine ecology, and parsing and extracting text content in the documents in the field of marine ecology; The document data processing step is to divide the text content into segments and extract the abstract of each segment. In the fine-tuning dataset construction step, the segment content and its summary are stored in Json format to obtain a fine-tuning dataset.

[0011] In some embodiments, the process of constructing the domain knowledge base includes: The knowledge dataset acquisition step is to obtain marine environment data and human activity pressure data through the document loader; A data vectorization step, using a word segmenter to convert the marine environment data and human activity pressure data into text blocks, and performing vectorization processing based on an Embedding model to obtain a text vector; When the domain knowledge base obtains the distributed user request, the user request is vectorized and the text vector in the domain knowledge base is matched by similarity. After sorting the matching results using the Rerank model to obtain text vectors with high similarity, the text is integrated using the prompt word project and sent to the marine ecological model.

[0012] In some of the embodiments, the step of building a large marine ecological model uses the LORA fine-tuning method to perform training based on the Llama_factory framework.

[0013] In some of the embodiments, the workflow of the parallel application module is turned on or off by configuring a start node and an end node, and the main model of the domain knowledge base, marine ecological multi-agent, and multimodal large model is set to the marine ecological large model through an LLM node, and a switch node is configured for the domain knowledge base, marine ecological multi-agent, and multimodal large model, respectively, to control whether the workflow of the domain knowledge base, marine ecological multi-agent, and multimodal large model is turned on.

[0014] In some of the embodiments, the marine ecological multi-agent is also configured with a trigger, and the trigger is configured as a keyword trigger. When the switch node of the marine ecological multi-agent is turned on, the corresponding intelligent agent in the marine ecological multi-agent is turned on through the trigger.

[0015] In some embodiments, the basic function agent includes: a business database processing agent, a mathematical calculation agent and a server basic setting agent, wherein: The business database processing agent is obtained by encapsulating the API interface of the marine ecological business database into an agent, and the marine ecological business database includes: a marine ranch monitoring database, a pollutant monitoring database, and a global biodiversity database; The mathematical calculation skill problem is obtained by encapsulating a calculator as an intelligent agent; The server basic setting agent is obtained by encapsulating the server system query command into an agent.

[0016] In some embodiments, the Internet functional agent includes a language translation agent, a weather query agent, a search engine agent, a map service agent, a paper retrieval agent, and a knowledge encyclopedia agent, wherein: The language translation agent is obtained by encapsulating the translation platform API interface into an agent; The weather query agent is obtained by encapsulating the weather forecast platform API interface into an agent; The search engine agent is obtained by encapsulating the search engine API into an agent; The map service agent is obtained by encapsulating the map platform POI service API into an agent; The paper retrieval agent is obtained by encapsulating the library database API into an agent; The knowledge encyclopedia agent is obtained by encapsulating the Wikipedia database API interface into an agent.

[0017] In some embodiments, the marine ecological functional agent includes: a quantitative statistics agent, a species identification agent, a habitat simulation agent, and a species distribution prediction agent, which are used to achieve the purpose of protecting fishery resources; wherein, The quantitative statistics agent is obtained by encapsulating the fish fine-grained recognition model into an agent; The species identification intelligent agent is obtained by encapsulating the marine species identification model into an intelligent agent; The habitat simulation intelligent agent is obtained by encapsulating the marine habitat prediction simulation analysis model into an intelligent agent; The species distribution prediction agent is obtained by encapsulating the species distribution prediction model into an agent.

[0018] In some embodiments, the marine ecological function intelligent body also includes: a pollutant monitoring intelligent body, an extreme weather prediction intelligent body, and an ecological disaster early warning intelligent body, which are used to achieve the purpose of pollutant monitoring; wherein, The pollutant monitoring intelligent agent is obtained by obtaining the seawater quality monitoring of the marine environment monitoring center and encapsulating the water quality information of the bathing beach into an intelligent agent; The extreme weather prediction intelligent agent is obtained by encapsulating the extreme weather prediction model into an intelligent agent; The ecological disaster early warning intelligent agent is obtained by encapsulating the interface of the ecological disaster monitoring and early warning system into an intelligent agent.

[0019] In a second aspect, an embodiment of the present application provides a marine ecological multi-agent interaction system, based on the marine ecological multi-agent construction method described in the first aspect above, including: The data layer is used to generate fine-tuning data, domain knowledge base and business database based on data sources, including the data sources of the fine-tuning data set, knowledge data set and marine ecological multi-agent; The model layer is used to build the marine ecological big model, domain knowledge base, multimodal big model and encapsulate the sub-agents required by the basic functional agent, Internet functional agent and marine ecological functional agent. The sub-agents include: business database processing agent, mathematical calculation agent and server basic setting agent, language translation agent, weather query agent, search engine agent, map service agent, paper retrieval agent, knowledge encyclopedia agent, quantitative statistics agent, species identification agent, habitat simulation agent and species distribution prediction agent; The agent layer is used to integrate the sub-agents to build the basic function agent, Internet function agent, and marine ecological function agent based on the model layer; The application layer includes at least a fishery resource protection application module and an ecological protection application module. The fishery resource protection application module includes: a marine biological quantity counting unit, a species identification unit, a habitat simulation unit, and a species distribution prediction unit. The ecological protection application module includes: a pollutant monitoring unit, an extreme weather prediction unit, and an ecological disaster early warning unit. The interaction layer provides model configuration, multimodal dialogue, domain knowledge dialogue and knowledge base management through a front-end interactive interface.

[0020] In some embodiments, the front-end interactive interface includes at least: Model configuration unit, used to select large marine ecological models; A multimodal dialogue configuration unit, used to configure a multimodal large model; A domain knowledge base dialogue configuration unit is used to configure at least one knowledge base and perform question-answering services based on the configured knowledge base. Customization of new knowledge bases and editing of existing knowledge bases can be achieved through a domain knowledge base management unit; The multi-agent configuration unit obtains a multi-agent activation request through a selection box and provides a multi-agent list for selecting at least one agent; The dialogue interaction unit uploads files through the text input box or the file transfer button and communicates with the marine ecological model, obtains user requests, and sends them to the marine ecological model to obtain the output results of the marine ecological model after integrating semantics.

[0021] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 is a flow chart of a method for constructing a marine ecological multi-agent according to an embodiment of the present application; Figure 2 is a schematic diagram of the construction process of a fine-tuning dataset according to the method for constructing a marine ecological multi-agent according to an embodiment of the present application; Figure 3 1 is a schematic diagram of the process of constructing a domain knowledge base according to the method for constructing a marine ecological multi-agent according to an embodiment of the present application; Figure 4 It is a schematic diagram of the parallel application principle of the large marine ecological model according to an embodiment of the present application; Figure 5 is a schematic diagram of the construction principle of a domain knowledge base according to an embodiment of the present application; Figure 6 is a schematic diagram of the working principle of the domain knowledge base according to an embodiment of the present application; Figure 7 is a schematic diagram of the workflow control principle of a parallel application module according to an embodiment of the present application; Figure 8 is a schematic diagram of the principle of the marine ecological multi-agent interaction system according to an embodiment of the present application; Fig. 9 It is a schematic diagram of the interactive interface of the marine ecological multi-agent interactive system according to an embodiment of the present application.

[0023] In the figure: 1. Data layer; 2. Model layer; 3. Agent layer; 4. Application layer; 5. Interaction layer; 501, model configuration unit; 502, multimodal dialogue configuration unit; 503, domain knowledge base dialogue configuration unit; 504, multi-agent configuration unit; 505, dialogue interaction unit. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0025] Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.

[0026] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0027] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the usual meanings understood by persons with ordinary skills in the technical field to which this application belongs. The words "one", "a", "the" and the like involved in this application do not indicate a quantitative limitation and may indicate the singular or plural. The terms "include", "comprises", "has" and any variations thereof involved in this application are intended to cover non-exclusive inclusions.

[0028] This embodiment provides a method for constructing a marine ecological multi-agent. Figure 1 is a flow chart of a method for constructing a marine ecological multi-agent according to an embodiment of the present application. Figure 1 As shown, the process includes the following steps: Marine ecological large model construction step S1, construct a fine-tuning data set, and train the selected large language model based on the fine-tuning data set to obtain a marine ecological large model. The large language model selected in this embodiment is Qwen2-7B to meet the requirements of performance, deployment framework compatibility, and intelligent agent compatibility, specifically Qwen-vl-chat, glm-4v, etc.; The parallel application module constructs step S2, constructs a domain knowledge base, a marine ecological multi-agent, and builds a multimodal large model. The domain knowledge base, the marine ecological multi-agent, the multimodal large model of the parallel application module and the marine ecological large model are deployed on the GPU server in a diversified parallel deployment manner. Optionally, the GPU server includes at least 8 64G video memory H800 cards, and each H800 GPU card can run about 3-5 large models of different magnitudes.

[0029] Among them, the knowledge data sets of the domain knowledge base include: marine environment data, human activity pressure data, human activity pressure data include: socio-economic data, human activity data, and the marine ecological multi-intelligent body includes: pre-packaged basic functional intelligent body, Internet functional intelligent body, and marine ecological functional intelligent body. Among them, the basic functional intelligent body is used to realize business database information retrieval, mathematical calculations, server basic settings, operation and maintenance, etc. Internet functional agents are used to implement language translation, weather query, search engine, map service, paper retrieval, Wikipedia, etc. Marine ecological functional agents are used to implement functions in the field of fishery resources and marine ecological protection, and can support quantitative statistics, species identification, habitat simulation, species distribution prediction, pollutant monitoring, extreme weather prediction, ecological disaster warning, etc. Multimodal large models include multimodal models, Wensheng graph models, text speech models, and Wensheng video models, which are used to support multimodal large models to achieve visual question answering, text to picture, speech, and video conversion and generation; Optionally, the supported Wensheng graph models are sd-turbo, sdxl-turbo, and sd-medium, the supported text-speech models are ChatTTS and whisper-base, the supported Reranker models are Bge-reranker-base and jina-reranker-v2, the supported Embedding models are Bge-m3, bge-base-zh, and gte-base, and the supported Wensheng video models are Keling and Vidu. In addition, small models and other interfaces are also supported.

[0030] refer to Figure 4 As shown, the marine ecological big model is used to parse user requests and distribute them to the parallel application module, triggering any one of the domain knowledge base, marine ecological multi-agent, and multimodal big model in the parallel application module or any combination thereof to generate response results that are fed back to the marine ecological big model for semantic integration and output.

[0031] Based on the above steps, in order to solve the problems of lagging existing model technology, poor model accuracy and stability, the embodiment of the present application constructs a large marine ecological model and adopts a large model parallel application architecture to improve the level of intelligence in the marine ecological field. The large marine ecological model is used to capture subtle changes and long-term trends that are difficult to capture with traditional models, thereby improving the prediction accuracy of the model. The parallel application architecture can support larger-scale data processing and more complex model training, further enhance the prediction ability of the model, and make the prediction results more accurate and reliable. The parallel application architecture can make full use of the computing resources of multiple computing nodes to realize parallel training and reasoning of the model, thereby significantly shortening the time of model training and reasoning, and improving the response speed of the model.

[0032] refer to Figure 2 As shown in Figure 2, the construction process of the fine-tuning dataset includes: Data acquisition step S201, write a python crawler to obtain marine ecology literature in .pdf format, including Chinese papers related to marine ecology in HowNet, and parse and extract text content in marine ecology literature; optionally, use the Nougat model to extract text content in the pdf file, remove other content, and save the text content in makedown format.

[0033] The document data processing step S202 divides the text content into segments and extracts the summary of each segment accordingly. Optionally, a summary generation model under the PaddleNLP framework is used to perform batch summary extraction.

[0034] In the fine-tuning dataset construction step S203, the segment content and its summary are stored in Json format to obtain a fine-tuning dataset. Optionally, the fine-tuning dataset content is saved as three fields in the Json file, where the instruction field is the summary content, the input field is empty, and the output field is the detailed content corresponding to the summary.

[0035] refer to Figure 3 , Figure 5 As shown in the figure, the construction process of the domain knowledge base includes: In the step S301 of acquiring the knowledge data set, the marine environment data and human activity pressure data in the formats of csf and pdf are acquired through the document loader; wherein the marine environment data are divided into four types of data: seabed topography, marine physics, marine chemistry, and nutrients. The seabed topography data includes information such as the distance to the coastline and elevation; the marine physics data includes information such as the mean sea surface temperature, the range of sea surface temperature, the sea surface temperature in the warmest month, the sea surface temperature in the coldest month, the mean sea surface salinity, the range of sea surface salinity, the water flow speed, the mean bottom illumination, the range of bottom illumination, and the height of waves; the marine chemistry data includes information such as chlorophyll a, primary productivity, PH, and total suspended solids; the nutrient data includes information such as nitrate, phosphate, dissolved oxygen, and silicate. The specific data source examples of the marine environment data are shown in Table 1. The data formats are makedown and pdf. Data in other data formats can be saved in makedown format after format conversion to optimize the data structure.

[0036] Table 1 Marine environmental data

[0037] The specific data source examples of human activity pressure data are shown in Table 2. Human activity pressure data include socioeconomic data and human activity data. Socioeconomic data include information such as GDP, coastal population density, and fishery economy. Human activity data include information such as aquaculture, shipping density, distance to ports, coastal land use, aquaculture pond expansion, and marine protected areas. The data formats are makedown and pdf.

[0038] Table 2 Human activity pressure data

[0039] In the data vectorization step S302, a word segmenter is used to convert the marine environment data and human activity pressure data into text blocks, and vectorization processing is performed based on the Embedding model to obtain text vectors, which are stored in a data vector database, such as the Chroma vector database and the Faiss vector database. The combination of the vector database with the Embedding model and the word segmenter will directly affect the accuracy of the domain knowledge base question and answer. In practice, the Chroma vector database with the highest matching degree with the Embedding model and the word segmenter and the best actual effect is adopted; in order to take into account the effectiveness of the vectorization of Chinese and English, the Embedding model adopts the Bge-m3 model; in addition, the UnstructuredMarkdownLoader loader is used for the unique tree structure of the document data in the makedown format to reduce the tree structure logic errors during the document loading process, and the markdown_header_metadata_splitter word segmenter is used to ensure the integrity and context logic of the text blocks after segmentation; Among them, reference Figure 6 As shown, when the domain knowledge base obtains the distributed user request, the user request is vectorized and the text vectors in the domain knowledge base are matched by similarity. After sorting the matching results using the Rerank model to obtain text vectors with high similarity, the text is integrated using the prompt word project and sent to the marine ecological model. Optionally, the question-answer parameter of the domain knowledge base is set to 5, and the top 5 high-similarity text vectors are selected as candidates for the final answer. The context is set to 1024, and a maximum of 1024 characters or words before and after the user query are considered as context.

[0040] In another embodiment, data such as marine ecological news information, laws and regulations, papers, books, etc. can also be used in the construction of the domain knowledge base.

[0041] Based on the above steps, the embodiment of the present application constructs a fine-tuning data set of a large marine ecological model covering a variety of marine environmental elements and human activity elements, builds and optimizes a knowledge base in the field of marine ecology, solves the problem of lack of marine ecological data, and enables the model to be calculated based on more comprehensive and accurate data, thereby improving the accuracy and reliability of the model.

[0042] In some of the embodiments, the LORA fine-tuning method is used in the step of building a large marine ecological model based on the Llama_factory framework for training. The fine-tuning parameter settings of the training process are shown in Table 3 below. In this embodiment, considering that 3.0 training rounds will cause overfitting and affect the basic performance of Qwen2-7B, in order to avoid overfitting, it is necessary to screen the LORA weights of each training round. Qwen2-7B loads each LORA weight in turn, and conducts a question-and-answer test to eliminate overfitting and logically incorrect weights. Select the LORA weight with the smallest number of training rounds under the premise of fine-tuning function.

[0043] Table 3 Fine-tuning parameter settings

[0044] Furthermore, the weights of Qwen2-7B were merged with the LORA weights. The maximum block size was set to 2GB and the export quantization level was set to none. No quantization was performed to ensure model performance. The merged weights were subjected to inference tests. The accuracy of the large model question and answer was actually verified using manual testing methods. The construction of the marine ecological large model was completed after the accuracy reached the standard. The inference parameter settings are shown in Table 4 below.

[0045] Table 4 Inference parameter settings

[0046] In some of these embodiments, reference Figure 7 As shown, the workflow of the parallel application module is turned on or off through the start node and end node configuration, and the main model of the domain knowledge base, marine ecological multi-agent, and multi-modal large model is set as the marine ecological large model through an LLM node, and a switch node is configured for the domain knowledge base, marine ecological multi-agent, and multi-modal large model respectively, which is used to control whether the workflow of the domain knowledge base, marine ecological multi-agent, and multi-modal large model is turned on. When the domain knowledge base, marine ecological multi-agent, and multi-modal large model are all turned off, the marine ecological large model will directly output the results according to the user's request, and will no longer trigger the domain knowledge base, marine ecological multi-agent, and multi-modal large model.

[0047] In some embodiments, the marine ecological multi-agent is further configured with a trigger, which is configured as a keyword trigger. When the switch node of the marine ecological multi-agent is turned on, the corresponding agent in the marine ecological multi-agent is turned on through the trigger. In another embodiment, the trigger can also be set as a timed trigger or an event trigger.

[0048] In some embodiments, the basic function agent includes: a business database processing agent, a mathematical calculation agent, and a server basic setting agent, wherein: The business database processing agent is obtained by encapsulating the API interface of the marine ecological business database into an agent. The marine ecological business database includes: the marine ranch monitoring database and the pollutant monitoring database based on MySQL, and the global biodiversity database based on Neo4J. The agent is started, and the corresponding agent can be triggered when the user request involves keywords such as "marine ranch", "pollutants", "monitoring" and "biodiversity"; Mathematical calculation skill questions are obtained by encapsulating a calculator into an intelligent agent. When the intelligent agent is started and an operator is input, the intelligent agent is triggered and run, giving the marine ecological large model computing capabilities, serving the marine ecological refined data analysis scenario, and making up for the shortcomings of the large model's mathematical calculation performance; The server basic setting agent is obtained by encapsulating the server system query command into an agent, and performs operation performance detection, CPU usage query, video memory and memory usage analysis, disk usage analysis, and network performance analysis through the server system query command. When the agent is started and keywords such as "CPU usage", "video memory", "memory", and "disk usage" are entered, this agent is triggered and run.

[0049] In some embodiments, the Internet functional agent includes a language translation agent, a weather query agent, a search engine agent, a map service agent, a paper retrieval agent, and a knowledge encyclopedia agent, wherein: The language translation agent is obtained by encapsulating the translation platform API interface into an agent. When the agent is started and keywords such as "translation" are input, the agent is triggered and run to obtain the translation result of the specified translation content, giving the marine ecological large model the ability to translate multiple languages, especially the translation effect of minority languages. The weather query agent is obtained by encapsulating the weather forecast platform API interface into an agent. For example, if you know the weather, start the agent and enter keywords such as "city" and "weather", the agent will be triggered and run to obtain the weather conditions of the specified city, giving the marine ecological model the ability to forecast weather. The search engine agent is obtained by encapsulating the search engine API into an agent, such as Bing search and Baidu. When the agent is started and keywords such as "Internet search" are input, the agent is triggered and run to obtain Internet search results, thus giving the marine ecological model the ability to search for Internet information. The map service agent is obtained by encapsulating the map platform POI service API into an agent, such as Amap. When the agent is started and location-based service keywords such as "place", "hotel", and "restaurant" are entered, the agent is triggered and run to obtain the corresponding location service results, giving the marine ecological large model the ability to provide location services. The paper retrieval agent is obtained by encapsulating the library database API into an agent, such as the Arxiv database. When the agent is started and information such as "paper name" and "keywords" is input, the agent is triggered and run to obtain the corresponding paper title, abstract and other information; The knowledge encyclopedia agent is obtained by encapsulating the Wikipedia database API interface into an agent. Here, the Wikipedia database can also be replaced by other knowledge query databases, such as the Duxiu knowledge base. Start the agent and enter keywords such as "noun to be queried" and "Wikipedia". It will be triggered and run, and the encyclopedia explanation content corresponding to the noun to be queried will be returned.

[0050] In some of the embodiments, the marine ecological functional agent includes: a quantitative statistics agent, a species identification agent, a habitat simulation agent, and a species distribution prediction agent, which are used to achieve the purpose of protecting fishery resources; wherein, The quantity counting agent is obtained by encapsulating the fine-grained fish identification model into an agent. Start the agent and input keywords such as "fish quantity counting" to trigger and run it, and return fish quantity counting information such as "fish species", "quantity", and "spatial coordinates", giving the large marine ecological model the ability to identify and count fish in fine grains, so as to serve scenarios such as marine ranching and economic fish population protection; The species identification agent is obtained by encapsulating the marine species identification model into an agent. The agent is started and keywords such as "species name" and "marine ranch name" are input to trigger and run the agent. The agent returns information such as species name and spatial coordinates, giving the marine ecological large model marine ranch the ability to identify common species, serving scenarios such as marine ranches. The habitat simulation agent is obtained by encapsulating the marine habitat prediction simulation analysis model into an agent. The agent is started and keywords such as "marine ranch", "sea bathing beach" or the name of the station are input to trigger and run, and the temperature, salinity, current, depth and water quality indicators of the specified marine area are returned. The marine ecological model is endowed with the ability to predict and analyze the temperature, salinity, current, depth and water quality indicators (including dissolved oxygen, total phosphorus, total nitrogen, ammonia nitrogen, chemical oxygen demand, chlorophyll a) of the selected marine ranch, sea bathing beach, station and other marine areas; The species distribution prediction agent is obtained by encapsulating the species distribution prediction model into an agent, such as the Maxent model. Start the agent and input keywords such as "Maxent", "species distribution prediction", "species name", "latitude and longitude" and "environmental information", trigger and run it, and return information such as species suitable habitats, giving the marine ecological model the ability to predict the distribution and suitable habitats of specific species.

[0051] In some embodiments, the marine ecological function intelligent body also includes: a pollutant monitoring intelligent body, an extreme weather prediction intelligent body, and an ecological disaster early warning intelligent body, which are used to achieve the purpose of pollutant monitoring; wherein, The pollutant monitoring agent is obtained by obtaining the seawater quality monitoring of the marine environment monitoring center and encapsulating the water quality information of the beach into an agent. The agent is started and keywords such as "bathing beach name" and "water quality" are input to trigger and run the agent, and the information such as the water quality of the beach is returned, giving the marine ecological large model the ability to monitor seawater quality. The extreme weather prediction agent is obtained by encapsulating the extreme weather prediction model into an agent. The agent is started and keywords such as "sea area", "typhoon", and "storm surge" are input to trigger and run the agent. The information such as typhoon and storm surge in the specified sea area is returned, giving the marine ecological model the ability to predict typhoons and storm surges. The ecological disaster early warning agent is obtained by encapsulating the interface of the ecological disaster monitoring and early warning system into an agent. The agent is started and keywords such as "sea area", "red tide", "green tide" and "entropy" are input to trigger and run the agent, and the information of red tide, green tide or enteromorpha in the specified sea area is returned, giving the marine ecological large model the ability to predict red tide, green tide and enteromorpha. The fish fine-grained identification model, marine species identification model, marine habitat prediction simulation analysis model, species distribution prediction model, extreme weather prediction model, and ecological disaster monitoring and early warning system of the above-mentioned embodiments can be pre-trained or introduced based on existing platforms. As the functions of each model are enriched, the marine ecological large model of this application is also further expanded accordingly; Based on the above steps, the embodiment of the present application integrates basic functional intelligent bodies, Internet functional intelligent bodies, and marine ecological functional intelligent bodies by constructing a marine ecological multi-intelligent body, which is conducive to serving multiple scenarios in the fields of fishery resource protection and ecological protection, and provides solutions for the systematic application of marine ecology.

[0052] The domain knowledge base, marine ecological multi-agent, multi-modal large model and the marine ecological large model of the above embodiment are deployed on the GPU server in a diversified parallel deployment mode, specifically including: Configuration of the production environment, where the production environment is divided into five modules: large model server, small model server, database, API, and application; The large model server is used to deploy large language models, multimodal large models, etc., and uses the Xinference framework for parallel deployment. Specific models include: marine ecological large model, Qwen-vl-chat, sd-turbo, ChatTTS, Bge-reranker-base, Bge-m3, etc. The small model server includes various marine ecological artificial intelligence small models, mostly using Paddle and PyTorch frameworks, including: marine habitat prediction simulation analysis model, marine species identification model, marine habitat prediction simulation analysis model, species distribution prediction model, extreme weather prediction model, etc., and configures its production environment.

[0053] The database side is used to deploy various marine ecology-related relational databases and graph databases, including: marine ranch monitoring database and pollutant monitoring database, global biodiversity database, ecological disaster monitoring and early warning system, and configure its production environment.

[0054] The API end is used to uniformly call various API interfaces to ensure normal interface access, including: calculator, server system query commands, Keling API, seawater quality monitoring API, beach water quality weekly report API, Baidu translation API, Xinzhi Weather API, Bing search API, Amap API, Wikipedia database API, Arxiv API, etc., and configure its production environment.

[0055] The application side is mainly responsible for the construction of the front-end and supporting functions of the marine ecological multi-intelligence. The production environment to be configured includes: Chroma vector database, WebUI front-end, Langchain framework, Paddle framework, PyTorch framework, etc.

[0056] In addition, network and external network mapping port configuration is also required, including configuring the external network access interface of the marine ecological multi-intelligence body and configuring the API call interface of the large model server, small model server, database, API and other modules.

[0057] Based on the above configuration, the model is kept running normally, the production environment is connected to the Internet, and the model interface and API are available.

[0058] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0059] Based on the above embodiments, this embodiment also provides a marine ecological multi-agent interaction system, which has been described and will not be repeated. As used below, the terms "module", "unit", "subunit", etc. can be a combination of software and / or hardware that can implement a predetermined function.

[0060] Figure 8 : is a schematic diagram of the principle of the marine ecological multi-agent interaction system according to an embodiment of the present application, such as Figure 8 As shown, the system includes: Data layer 1 is used to generate fine-tuning data, domain knowledge base and business database based on data sources. The data sources include fine-tuning data sets, knowledge data sets and data sources of marine ecological multi-agents. Model layer 2, used to build the marine ecological big model, domain knowledge base, multimodal big model and encapsulate the basic function agent, Internet function agent, and sub-agents required for the marine ecological function agent. The sub-agents include: business database processing agent, mathematical calculation agent and server basic setting agent, language translation agent, weather query agent, search engine agent, map service agent, paper retrieval agent, knowledge encyclopedia agent, quantitative statistics agent, species identification agent, habitat simulation agent and species distribution prediction agent; Agent layer 3 is used to integrate the sub-agents to build the basic function agent, Internet function agent, and marine ecological function agent based on the model layer; Application layer 4, at least including a fishery resource protection application module and an ecological protection application module, the fishery resource protection application module includes: a marine biological quantity counting unit, a species identification unit, a habitat simulation unit, and a species distribution prediction unit, the ecological protection application module includes: a pollutant monitoring unit, an extreme weather prediction unit, and an ecological disaster early warning unit, the fishery resource protection application module is constructed based on the marine ecological functional intelligent body, the marine biological quantity counting unit, the species identification unit, the habitat simulation unit, and the species distribution prediction unit are respectively connected to the quantity counting intelligent body, the species identification intelligent body, the habitat simulation intelligent body, and the species distribution prediction intelligent body, the ecological protection application module is constructed based on the marine ecological functional intelligent body, the pollutant monitoring unit, the extreme weather prediction unit, and the ecological disaster early warning unit are respectively connected to the pollutant monitoring intelligent body, the extreme weather prediction intelligent body, and the ecological disaster early warning intelligent body; The interaction layer 5 provides model configuration, multimodal dialogue, domain knowledge dialogue and knowledge base management through a front-end interactive interface.

[0061] In some of these embodiments, reference Fig. 9 As shown, the front-end interactive interface at least includes: Model configuration unit 501, used to select a large marine ecological model; The multimodal dialogue configuration unit 502 is used to configure the multimodal large model, wherein the sd-turbo model is selected for Wenshengtu, the ChatTTS model is selected for text-to-speech conversion, the Keling large model is selected for Wenshengvideo, and the Qwen-vl-chat model is selected for image semantic question answering; A domain knowledge base dialogue configuration unit 503, used to configure at least one knowledge base and perform question-answering services based on the configured knowledge base. Customization of new knowledge bases and editing of existing knowledge bases can be achieved through a domain knowledge base management unit; The multi-agent configuration unit 504 obtains a multi-agent activation request through a selection box and provides a multi-agent list for selecting at least one agent; The dialogue interaction unit 505 uploads files through the text input box or the file transfer button and communicates with the marine ecological model, obtains the user request, sends it to the marine ecological model, obtains the output result of the marine ecological model after integrating semantics, and displays it in the dialogue content display area. The dialogue interaction unit also provides a clear button to clear the historical chat content.

[0062] It should be noted that the above modules may be functional modules or program modules. The above modules may be located in the same processor; or the above modules may be located in different processors in any combination.

[0063] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for constructing a marine ecological multi-agent, characterized in that: include: The steps of constructing a large marine ecological model include constructing a fine-tuning dataset, and training the selected large language model based on the fine-tuning dataset to obtain a large marine ecological model; The parallel application module construction step includes constructing a domain knowledge base, a marine ecological multi-agent, and a multimodal large model. The domain knowledge base, the marine ecological multi-agent, the multimodal large model of the parallel application module and the marine ecological large model are deployed on the GPU server in a diversified parallel deployment mode. The knowledge data set of the domain knowledge base includes: marine environment data, human activity pressure data, the human activity pressure data includes: socio-economic data, human activity data, the marine ecological multi-agent includes: pre-packaged basic function agent, Internet function agent, marine ecological function agent, the multimodal large model includes a multimodal model, a text-based graph model, a text-based voice model, and a text-based video model; The marine ecological big model is used to parse user requests and distribute them to the parallel application module, triggering any one of the domain knowledge base, marine ecological multi-agent, and multimodal big model in the parallel application module or any combination thereof to generate response results that are fed back to the marine ecological big model for semantic integration and output.

2. The method for constructing a marine ecological multi-intelligent entity according to claim 1, characterized in that: The process of constructing the fine-tuning dataset includes: A data acquisition step, acquiring documents in the field of marine ecology, and parsing and extracting text content in the documents in the field of marine ecology; The document data processing step is to divide the text content into segments and extract the abstract of each segment; The fine-tuning dataset construction step stores the segment content and its summary to obtain a fine-tuning dataset.

3. The method for constructing a marine ecological multi-agent according to claim 1, characterized in that: The construction process of the domain knowledge base includes: The knowledge dataset acquisition step is to obtain marine environment data and human activity pressure data through the document loader; A data vectorization step, using a word segmenter to convert the marine environment data and human activity pressure data into text blocks, and performing vectorization processing based on an Embedding model to obtain a text vector; When the domain knowledge base obtains the distributed user request, the user request is vectorized and the text vector in the domain knowledge base is matched by similarity. After sorting the matching results using the Rerank model to obtain text vectors with high similarity, the text is integrated using the prompt word project and sent to the marine ecological model.

4. The method for constructing a marine ecological multi-agent according to claim 1, characterized in that: The workflow of the parallel application module is turned on or off through the start node and the end node configuration, and the main model of the domain knowledge base, marine ecological multi-agent, and multimodal large model is set as the marine ecological large model through an LLM node, and a switch node is configured for the domain knowledge base, marine ecological multi-agent, and multimodal large model respectively, which is used to control whether the workflow of the domain knowledge base, marine ecological multi-agent, and multimodal large model is turned on.

5. The method for constructing a marine ecological multi-intelligent entity according to claim 4, characterized in that: The marine ecological multi-agent is also configured with a trigger, and the trigger is configured as a keyword trigger.

6. The method for constructing a marine ecological multi-agent according to claim 1, characterized in that: The basic function agent includes: a business database processing agent, a mathematical calculation agent and a server basic setting agent, wherein: The business database processing agent is obtained by encapsulating the API interface of the marine ecological business database into an agent, and the marine ecological business database includes: a marine ranch monitoring database, a pollutant monitoring database, and a global biodiversity database; The mathematical calculation skill problem is obtained by encapsulating a calculator as an intelligent agent; The server basic setting agent is obtained by encapsulating the server system query command into an agent.

7. The method for constructing a marine ecological multi-intelligent entity according to claim 6, characterized in that: The Internet functional agents include language translation agents, weather query agents, search engine agents, map service agents, paper retrieval agents, and knowledge encyclopedia agents, among which: The language translation agent is obtained by encapsulating the translation platform API interface into an agent; The weather query agent is obtained by encapsulating the weather forecast platform API interface into an agent; The search engine agent is obtained by encapsulating the search engine API into an agent; The map service agent is obtained by encapsulating the map platform POI service API into an agent; The paper retrieval agent is obtained by encapsulating the library database API into an agent; The knowledge encyclopedia agent is obtained by encapsulating the Wikipedia database API interface into an agent.

8. The method for constructing a marine ecological multi-intelligent entity according to claim 7, characterized in that: The marine ecological function agents include: quantity statistics agents, species identification agents, habitat simulation agents, and species distribution prediction agents; The quantitative statistics agent is obtained by encapsulating the fish fine-grained recognition model into an agent; The species identification intelligent agent is obtained by encapsulating the marine species identification model into an intelligent agent; The habitat simulation intelligent agent is obtained by encapsulating the marine habitat prediction simulation analysis model into an intelligent agent; The species distribution prediction agent is obtained by encapsulating the species distribution prediction model into an agent.

9. The method for constructing a marine ecological multi-intelligent entity according to claim 8, characterized in that: The marine ecological function intelligent body also includes: pollutant monitoring intelligent body, extreme weather prediction intelligent body, ecological disaster early warning intelligent body; wherein, The pollutant monitoring intelligent agent is obtained by obtaining seawater quality monitoring and bathing beach water quality information and encapsulating it into an intelligent agent; The extreme weather prediction intelligent agent is obtained by encapsulating the extreme weather prediction model into an intelligent agent; The ecological disaster early warning intelligent agent is obtained by encapsulating the interface of the ecological disaster monitoring and early warning system into an intelligent agent.

10. A marine ecological multi-agent interactive system, based on the marine ecological multi-agent construction method according to any one of claims 1 to 9, characterized in that: include: The data layer is used to generate fine-tuning data, domain knowledge base and business database based on data sources, including the data sources of the fine-tuning data set, knowledge data set and marine ecological multi-agent; The model layer is used to build the marine ecological large model, domain knowledge base, multimodal large model and encapsulate the sub-agents required for the basic functional agent, Internet functional agent, and marine ecological functional agent; The agent layer is used to integrate the sub-agents to build the basic function agent, Internet function agent, and marine ecological function agent based on the model layer; The application layer includes at least a fishery resource protection application module and an ecological protection application module. The fishery resource protection application module includes: a marine biological quantity counting unit, a species identification unit, a habitat simulation unit, and a species distribution prediction unit. The ecological protection application module includes: a pollutant monitoring unit, an extreme weather prediction unit, and an ecological disaster early warning unit. The interaction layer provides model configuration, multimodal dialogue, domain knowledge dialogue and knowledge base management through a front-end interactive interface.

Images