Method and device for constructing central air conditioning management and control system based on mcp protocol, and medium

Abstract

The application discloses a construction method and equipment of a central air conditioner management and control system based on an MCP protocol and a medium, and relates to the technical field of central air conditioner management and control. The method comprises the following steps: constructing a central air conditioner equipment interconnection control MCP Server module on an AI edge workstation, uniformly accessing multiple types of equipment, and encapsulating the equipment as an MCP equipment control tool for calling by cloud LLMs; querying and analyzing the MCP Server module through data, integrating multiple source data, and encapsulating the data as metadata service resources to support NL2SQL query and visualization driven by natural language; calling an MCP energy-saving regulation and control tool to realize autonomous optimization of edge side group intelligence by cloud LLMs, generate global optimal energy-saving parameters, and issue and execute energy-saving strategies; and realizing fault diagnosis, parameter query, and maintenance guidance by combining RAG and a knowledge base through an intelligent operation and maintenance question and answer MCP Server module. Finally, fine and intelligent management and control of a central air conditioner system are realized.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of central air conditioning management and control, and particularly relates to a construction method of a central air conditioning management and control system based on an MCP protocol, equipment and a medium. BACKGROUND

[0002] With the development of artificial intelligence, the Internet of Things and other technologies, the management and control of central air conditioning systems are moving towards integration and intelligentization. However, there are still some problems in the centralized intelligent management and control of the currently used central air conditioning systems. For example, there are often barriers to interconnection and intercommunication between devices connected in the central air conditioning system. Specifically, the host, water chiller, cooling tower and water pump integrated in the system come from different manufacturers and use different communication protocols. The differences in communication protocols between devices often lead to an increase in the integration cost between devices. In addition, since each communication protocol corresponds to an independent human-machine interface, the property needs to use multiple human-machine interfaces to manage and control the devices in the central air conditioning system, which increases the management difficulty and integration cost. Secondly, the data stored in the current central air conditioning system is often scattered in the databases of different subsystems. The operation and maintenance personnel need to manually input instructions to query and analyze data across systems and databases, which causes the problem of low efficiency in querying the central air conditioning system log data. Therefore, it is an urgent problem to be solved to provide a construction method of a central air conditioning management and control system based on an MCP protocol to construct a central air conditioning management and control system to improve the management and operation and maintenance efficiency of the central air conditioning system. SUMMARY

[0003] The embodiments of the application provide a construction method of a central air conditioning management and control system based on an MCP protocol, equipment and a medium, and the main purpose is to construct a central air conditioning management and control system to improve the management and operation and maintenance efficiency of the central air conditioning system.

[0004] In a first aspect, the embodiments of the application provide a construction method of a central air conditioning management and control system based on an MCP protocol, and the method comprises the following steps.

[0005] The device interconnection control MCP Server module encapsulates the device control function as an MCP device control tool for unified calling of the LLMs to control different heterogeneous devices of different protocols.

[0006] defining a metadata service resource based on the MCP Resource decorator on the data query and analysis MCP Server module deployed on the cloud server, and constructing a table header extraction tool for metadata extraction of different types of databases to encapsulate the metadata of different databases extracted by the table header extraction tool into the metadata service resource in a unified format, so that the NL2SQL generates SQL statements for different types of databases according to the metadata, and queries data of different databases;

[0007] encapsulating different chart generation functions as an MCP table generation tool on the data query and analysis MCP Server module deployed on the cloud server, so that the LLMs uniformly call the MCP generation tool to generate different charts for database query data for chart analysis and display;

[0008] encapsulating a function of calling a swarm intelligent AI autonomous optimization algorithm as an MCP energy-saving regulation tool on the energy-saving regulation MCP Server module of the AI edge workstation, so that the LLMs call the MCP energy-saving regulation tool to control the swarm intelligent AI autonomous optimization algorithm to adjust the energy-saving state of the central air conditioning system;

[0009] encapsulating an MCP intelligent operation and maintenance question answering tool on the intelligent operation and maintenance question answering MCP Server module deployed on the cloud server, so that the LLMs call the MCP intelligent operation and maintenance question answering tool to assist operation and maintenance work;

[0010] connecting the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy-saving regulation MCP Server module, and the intelligent operation and maintenance question answering MCP Server module to the MCP client deployed on the cloud server to obtain a central air conditioning management and control system, and the MCP client is used for interacting with users and calling LLMs and the NL2SQL to manage and control the central air conditioning system.

[0011] In an implementation manner of the present application, the encapsulating a device control function as an MCP device control tool on the device interconnection control MCP Server module of the AI edge workstation comprises:

[0012] constructing an initial device interconnection control MCP Server module in the AI edge workstation, and accessing a device control function in the initial device interconnection control MCP Server module, and the AI edge workstation is used for connecting each heterogeneous device in the central air conditioning system at a physical layer;

[0013] The device control function is encapsulated based on the MCP protocol specification to obtain the MCP device control tool, and the device control function is a bottom layer control function corresponding to each protocol of the heterogeneous devices.

[0014] In an implementation manner of the present application, the metadata service resource is defined based on the MCP Resource decorator on the data query and analysis MCP Server module deployed in the cloud server, and the table header extraction tool for metadata extraction of different types of databases is constructed, including:

[0015] The metadata service resource is defined based on the MCP Resource decorator, and the path of the metadata service resource is identified using a URI conforming to the data query and analysis MCP Server module standard, so as to search the metadata service resource according to the URI;

[0016] The table header extraction tool for metadata extraction of different types of databases is constructed, so that the table header extraction tool encapsulates the metadata of different types of databases in a unified format in the corresponding metadata service resource according to the URI.

[0017] In an implementation manner of the present application, the MCP intelligent operation and maintenance Q&A tool encapsulated on the intelligent operation and maintenance Q&A MCP Server module deployed in the cloud server includes:

[0018] The operation and maintenance document data is collected and preprocessed;

[0019] The processed operation and maintenance document data is segmented into knowledge fragments, and the segmented knowledge fragment vectors are stored in the Milvus vector knowledge base;

[0020] The Q&A system is built based on the framework RAGFLOW, and the MCP intelligent operation and maintenance Q&A tool for calling the Q&A system is encapsulated on the intelligent operation and maintenance Q&A MCP Server module, so as to complete the construction of the intelligent operation and maintenance Q&A MCP Server module.

[0021] In an implementation manner of the present application, the operation and maintenance document data is collected and preprocessed, including:

[0022] The operation and maintenance document data covering various devices of the central air conditioning system is collected;

[0023] The operation and maintenance steps involved in the operation and maintenance document data are checked against the device manufacturer technical document and the industry operation and maintenance standard, to ensure the reliability of the operation and maintenance document data;

[0024] The outdated and invalid content in the approved operation and maintenance document data is removed;

[0025] The description logic error in the corpus is corrected through manual review and cross verification, so as to guarantee the accuracy of the operation and maintenance document.

[0026] In an implementation manner of the present application, the knowledge segment division on the processed operation and maintenance document data, and the vector storage of the divided knowledge segment to the Milvus vector knowledge base include:

[0027] The processed operation and maintenance document data is divided according to different knowledge point types, and the knowledge segments of different types are obtained;

[0028] The knowledge segment is converted into vector data with semantic features by using a BGE-m3 word embedding model;

[0029] The vector data is stored in the Milvus vector knowledge base.

[0030] In an implementation manner of the present application, the connection of the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy saving regulation MCP Server module, and the intelligent operation and maintenance question and answer MCP Server module to the MCP client to obtain the central air conditioning management system includes:

[0031] The device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy saving regulation MCP Server module, and the intelligent operation and maintenance question and answer MCP Server module are connected to the MCP client based on the FastMCP architecture;

[0032] The LLMs are constructed for single node deployment or multi node deployment with the MCP client, a session link between the LLMs and the MCP client is established, and the central air conditioning management system is obtained.

[0033] In an implementation manner of the present application, the method includes: integrating a multi-modal input interface in the MCP client, and deploying a modal conversion middleware to receive different forms of user instructions according to the multi-modal input interface, and converting the text prompt decision instruction recognizable by the LLMs based on the modal conversion middleware.

[0034] In a second aspect, the embodiments of the present application also provide a construction device of a central air conditioning management and control system based on an MCP protocol, the device comprising: at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: encapsulate a device interconnection control MCP Server module on an AI edge workstation into a device control function MCP device control tool for unified calling of LLMs to control different heterogeneous devices of different protocols by the MCP device control tool; define a metadata service resource based on an MCP Resource decorator on a data query and analysis MCP Server module deployed in a cloud server, and construct a table header extraction tool for metadata extraction of different types of databases, so as to encapsulate metadata of different databases extracted by the table header extraction tool into the metadata service resource in a unified format, and enable NL2SQL to generate SQL statements of different types of databases according to the metadata to query data of different databases; encapsulate different chart generation functions into a MCP chart generation tool on the data query and analysis MCP Server module deployed in the cloud server, so as to enable LLMs to generate different charts by unified calling of the MCP generation tool to analyze and display data queried from the database in a chart; encapsulate a function of calling a group intelligent AI autonomous optimization algorithm on an energy saving regulation and control MCP Server module of the AI edge workstation into a MCP energy saving regulation and control tool, so as to enable the LLMs to control the group intelligent AI autonomous optimization algorithm to adjust an energy saving state of the central air conditioning system by calling the MCP energy saving regulation and control tool; encapsulate a MCP intelligent operation and maintenance question answering tool on an intelligent operation and maintenance question answering MCP Server module deployed in the cloud server, so as to enable the LLMs to call the MCP intelligent operation and maintenance question answering tool to assist operation and maintenance work; and connect the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy saving regulation and control MCP Server module, and the intelligent operation and maintenance question answering MCP Server module to a MCP client deployed in the cloud server to obtain a central air conditioning management and control system, wherein the MCP client is used to interact with a user and call LLMs and NL2SQL to manage and control the central air conditioning system.

[0035] In a third aspect, the embodiments of the present application also provide a non-volatile computer storage medium corresponding to the method of the first aspect, which stores computer executable instructions configured to: encapsulate device control functions of a device interconnection control MCP Server module of an AI edge workstation into MCP device control tools for unified calling of the MCP device control tools by LLMs to control different heterogeneous devices of different protocols; define metadata service resources based on MCP Resource decorators on a data query and analysis MCP Server module deployed in a cloud server, and build a table header extraction tool for metadata extraction of different types of databases, so as to encapsulate metadata of different databases extracted by the table header extraction tool into the metadata service resources in a unified format, so that NL2SQL generates SQL statements of different types of databases according to the metadata to query data of different databases; encapsulate different chart generation functions into a MCP chart generation tool on the data query and analysis MCP Server module deployed in the cloud server, so that LLMs uniformly call the MCP generation tool to generate different charts to analyze and display data queried from the database; encapsulate functions of calling a swarm intelligent AI autonomous optimization algorithm into a MCP energy saving regulation tool on an energy saving regulation MCP Server module of the AI edge workstation, so that the LLMs call the MCP energy saving regulation tool to control the swarm intelligent AI autonomous optimization algorithm to adjust the energy saving state of the central air conditioning system; encapsulate a MCP intelligent operation and maintenance question answering tool on an intelligent operation and maintenance question answering MCP Server module deployed in the cloud server, so that the LLMs call the MCP intelligent operation and maintenance question answering tool to assist operation and maintenance work; and connect the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy saving regulation MCP Server module, and the intelligent operation and maintenance question answering MCP Server module to a MCP client deployed in the cloud server to obtain a central air conditioning management and control system, wherein the MCP client is configured to interact with a user and call LLMs and the NL2SQL to manage and control the central air conditioning system.

[0036] The method, device and medium provided by the embodiment of the application provide a construction method of a central air conditioner management and control system based on an MCP protocol, encapsulate device control functions of an AI edge workstation device interconnection control MCP Server module as MCP device control tools for unified calling of LLMs to control different heterogeneous devices of different protocols, define metadata service resources based on an MCP Resource decorator on a data query and analysis MCP Server module deployed on a cloud server, and construct a table header extraction tool for metadata extraction of different types of databases to encapsulate metadata of different databases extracted by the table header extraction tool in the metadata service resources in a unified format, so that NL2SQL generates SQL statements of different types of databases according to the metadata to query data of different databases; encapsulate different chart generation functions as an MCP table generation tool on the data query and analysis MCP Server module deployed on the cloud server, so that LLMs uniformly call the MCP generation tool to generate different charts to analyze and display data queried from the database; encapsulate functions of a group intelligent AI autonomous optimization algorithm on an energy-saving regulation and control MCP Server module of the AI edge workstation as an MCP energy-saving regulation and control tool, so that the LLMs call the MCP energy-saving regulation and control tool to control the group intelligent AI autonomous optimization algorithm to adjust the energy-saving state of the central air conditioner system; encapsulate an MCP intelligent operation and maintenance question answering tool on an intelligent operation and maintenance question answering MCP Server module deployed on the cloud server, so that the LLMs call the MCP intelligent operation and maintenance question answering tool to assist operation and maintenance work; connect the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy-saving regulation and control MCP Server module and the intelligent operation and maintenance question answering MCP Server module to an MCP client deployed on the cloud server to obtain a central air conditioner management and control system, and the MCP client is configured to interact with a user and call LLMs and NL2SQL to manage and control the central air conditioner system. The construction method of the central air conditioner management and control system improves the management and operation and maintenance efficiency of the central air conditioner system. BRIEF DESCRIPTION OF DRAWINGS

[0037] The accompanying drawings, which are included to provide a further understanding of the application and constitute a part of this application, illustrate certain illustrative embodiments of the application and together with the description serve to explain the application. In the drawings:

[0038] Figure 1 The construction method of the central air conditioner management and control system based on the MCP protocol provided by the embodiment of the application is shown in a flowchart;

[0039] Figure 2A framework schematic diagram of a central air conditioning management and control system provided by an embodiment of the present application is provided.

[0040] Figure 3 A construction device internal structure schematic diagram of a central air conditioning management and control system based on the MCP protocol provided by an embodiment of the present application is provided. DETAILED DESCRIPTION

[0041] In order to make the purposes, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described in detail below in combination with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0042] The embodiments of the present application provide a construction method, device and medium of a central air conditioning management and control system based on the MCP protocol, to solve the problem that terminal device upgrade based on the setting of prior conditions by relevant management personnel is prone to human configuration errors.

[0043] The technical solutions proposed by the embodiments of the present application will be described in detail below with reference to the drawings.

[0044] Figure 1 A construction method flowchart of a central air conditioning management and control system based on the MCP protocol provided by an embodiment of the present application is provided. As shown in Figure 1 The construction method of a central air conditioning management and control system based on the MCP protocol provided by the embodiments of the present application specifically includes the following steps:

[0045] Step 101: encapsulate the device control function of the device interconnection control MCP Server module of the AI edge workstation into an MCP device control tool for unified calling of the MCP device control tool by the LLMs to control different heterogeneous devices of different protocols.

[0046] Step 102: define a metadata service resource based on the MCPResource decorator on the data query and analysis MCP Server module deployed in the cloud server, and construct a table header extraction tool for metadata extraction of different types of databases, so as to encapsulate the metadata of different databases extracted by the table header extraction tool into the metadata service resource in a unified format, so that the NL2SQL generates SQL statements of different types of databases according to the metadata, and queries data of different databases; it is realized that the user only needs to interact with the NL2SQL to generate the SQL statement corresponding to the database to be queried, and the fast data query across databases is realized.

[0047] Step 103: encapsulate different chart generation functions as an MCP chart generation tool on the data query and analysis MCP Server module deployed on the cloud server, so that the LLMs uniformly call the MCP generation tool to generate different charts to analyze and display the data queried from the database;

[0048] Step 104: encapsulate the function of calling the swarm intelligent AI autonomous optimization algorithm as an MCP energy-saving regulation tool on the energy-saving regulation MCP Server module of the AI edge workstation, so that the LLMs call the MCP energy-saving regulation tool to control the swarm intelligent AI autonomous optimization algorithm to adjust the energy-saving state of the central air conditioning system;

[0049] Step 105: encapsulate an MCP intelligent operation and maintenance question answering tool on the intelligent operation and maintenance question answering MCP Server module deployed on the cloud server, so that the LLMs call the MCP intelligent operation and maintenance question answering tool to assist in operation and maintenance work;

[0050] Step 106: connect the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy-saving regulation MCP Server module, and the intelligent operation and maintenance question answering MCP Server module to an MCP client deployed on the cloud server to obtain a central air conditioning management and control system, and the MCP client is used for interacting with a user and calling the LLMs and the NL2SQL to manage and control the central air conditioning system.

[0051] The application provides a construction method of a central air conditioner management and control system based on an MCP protocol, comprising: encapsulating device control functions of an equipment interconnection control MCP Server module of an AI edge workstation into MCP device control tools for unified calling of the MCP device control tools by LLMs to control different heterogeneous devices of different protocols; defining metadata service resources based on an MCP Resource decorator on a data query and analysis MCP Server module deployed on a cloud server, and constructing a table header extraction tool for metadata extraction of different types of databases to encapsulate metadata of different databases extracted by the table header extraction tool into the metadata service resources in a unified format, so that NL2SQL generates SQL statements of different types of databases according to the metadata to query data of different databases; encapsulating different chart generation functions on the data query and analysis MCP Server module deployed on the cloud server into MCP table generation tools to make LLMs uniformly call the MCP generation tools to generate different charts to analyze and display data queried from the database; encapsulating functions of a group intelligent AI autonomous optimization algorithm on an energy-saving regulation and control MCP Server module of the AI edge workstation into MCP energy-saving regulation and control tools to make the LLMs call the MCP energy-saving regulation and control tools to control the group intelligent AI autonomous optimization algorithm to adjust the energy-saving state of the central air conditioner system; encapsulating an MCP intelligent operation and maintenance question answering tool on an intelligent operation and maintenance question answering MCP Server module deployed on the cloud server to make the LLMs call the MCP intelligent operation and maintenance question answering tool to assist operation and maintenance work; connecting the equipment interconnection control MCP Server module, the data query and analysis MCP Server module, the energy-saving regulation and control MCP Server module and the intelligent operation and maintenance question answering MCP Server module to an MCP client deployed on the cloud server to obtain a central air conditioner management and control system, and the MCP client is used for interacting with a user and calling LLMs and NL2SQL to manage and control the central air conditioner system. The application has the following beneficial effects: a central air conditioner management and control system is constructed to improve the management and operation and maintenance efficiency of the central air conditioner system.

[0052] As a refinement of the embodiments of the present application, in the step of encapsulating the device control function of the MCP Server module of the AI edge workstation into the MCP device control tool in the step 101, the following implementation manners can also be adopted, but are not limited to, for example: constructing an initial device interconnection control MCP Server module in the AI edge workstation, and accessing the device control function in the initial device interconnection control MCP Server module, the AI edge workstation is used to connect each heterogeneous device in the central air conditioning system at the physical layer; encapsulating the device control function based on the MCP protocol specification to obtain the MCP device control tool, and the device control function is a bottom layer control function under the respective corresponding protocol of each heterogeneous device.

[0053] It should be understood that the MCP protocol is a Model Context Protocol (MCP), which is a standardized communication protocol designed to enable large language models (LLMs) to securely and structuredly call external tools, services, and data sources, thereby expanding their capability boundaries. By constructing the device interconnection control MCP Server module of the AI edge workstation, it is realized that the user can interact with the LLMs through text or voice instructions, and can control the central air conditioning system with different types of heterogeneous devices. The operation interface is unified and has high integration degree. The device control function is encapsulated based on the MCP protocol specification, which facilitates the access of new heterogeneous devices in the central air conditioning system and reduces the integration cost.

[0054] In some embodiments, the device interconnection control MCP Server module includes but is not limited to the following implementations, such as: the device interconnection control MCP Server module aims to realize the interconnection and intelligent control of central air conditioning unit core devices. This module will deploy the AI edge workstation in the central air conditioning machine room as an intelligent edge computing node, and its core architecture includes two parts of downlink physical device interface and uplink model service interface. It is connected to physical devices downward, serving as an intelligent gateway to uniformly manage multi-protocol heterogeneous devices including water chiller, cooling tower, and water pump. It provides services for LLMs upward, encapsulates a packaged tool (MCP device control tool) through MCP protocol, that is, the MCP device control tool, thereby realizing remote control of on-site physical devices. The device interconnection control MCP Server module specifically includes the following two functions: one is the device data and control access function, which is manifested as a built-in Internet of Things platform in the AI edge workstation, supporting mainstream protocols such as BACnet, Modbus, and MQTT, and uniformly accessing heterogeneous devices. Key operation data such as main machine, water pump, and cooling tower can be obtained in real time, and control instructions can be issued through RS-485 and other interfaces to realize effective control of the device. The other function is to provide a northbound interface for the communication between the AI edge workstation and the LLMs based on the MCP device control tool, which is the key to realizing the "edge-cloud collaborative control". The technical core lies in providing a method of automatically encapsulating the bottom-layer device management and control capability of the edge workstation into a standardized tool (Tool) conforming to the MCP protocol specification.

[0055] The specific implementation of the MCP device control Tool is that the device control function of the AI edge workstation can be declared as an MCP tool through the decorator (@mcp.tool). The developer can provide a clear function description for the tool through description, which is used to explain the purpose of the tool to the LLMs; and the types of function parameters and return values are defined, which helps the LLMs understand the data structure and type constraints required when calling, so as to generate more accurate and safe calling requests. The following is a partial original code example of the MCP device control Tool:

[0056] @mcp_tool()

[0057] def set_device_parameter(device_name: str, parameter: str, value:any) ->dict:

[0058] """

[0059] Description:

[0060] Setting individual operating parameters for specific equipment in a central air conditioning system allows modification of individual status points, such as equipment start / stop and frequency settings.

[0061] Args (input parameters):

[0062] device_name (str): The name of the target device, which is a string, such as "chiller-01", "pump-cw-03", etc.

[0063] parameter (str): The name of the parameter to be set. The data type is string. It is defined in an enumeration set of strings, such as ["tempeature", "frequency", "power_state"], to prevent setting non-existent parameters.

[0064] value (any): The parameter value to be set. The data type, format, and value range of this parameter are subject to dynamic and strongly dependent constraints.

[0065] For example, when `parameter` is "power_state", `value` must be a boolean; when `parameter` is "frequency", `value` must be a floating-point number and the value must be within the safe range [40.0, 50.0].

[0066] Returns (return value):

[0067] `dict`: A dictionary object containing the results of operations performed. Its structure is fixed as: `{"status": str, / / "success" or "error" "message": str / / Provides confirmation information for success or a detailed error description for failure}`

[0068] """

[0069] The aforementioned source code example of the MCP device control tool illustrates the implementation logic of the MCP device control tool calling local device control functions.

[0070] Subsequently, the MCP device control Tool is hosted by an MCP server, i.e., the device interconnection control MCP Server module, on the AI edge workstation, and listens to ToolCall requests from the cloud LLMs. When receiving a request, the MCP device control Tool is responsible for parsing the function name and parameters carried in the request, and calling a local device control function to perform a specific task. After the task is completed, the MCP device control Tool encapsulates the execution result (such as a success status or error information) in a predefined format and returns it to the cloud LLMs through the MCP protocol.

[0071] As a refinement of the above embodiment, in the step of executing 102, the metadata service resource is defined based on the MCP Resource decorator, and the table header extraction tool for metadata extraction of different types of databases is constructed, which can be implemented in the following manner, but is not limited thereto, for example: the metadata service resource is defined based on the MCP Resource decorator, and the path of the metadata service resource is identified using a URI conforming to the data query and analysis MCP Server module standard, so as to facilitate the search of the metadata service resource according to the URI; the table header extraction tool for metadata extraction of different types of databases is constructed, so that the table header extraction tool encapsulates the metadata of different types of databases in a unified format in the corresponding metadata service resource.

[0072] In some embodiments, the construction and use of the data query and analysis MCP Server module includes, but is not limited to, the following implementation manners, for example: the module serves as a "data hub" for central air conditioning system operation data, which is responsible for integrating multi-dimensional data of the central air conditioning system and encapsulating it as a standardized resource conforming to the MCP protocol for calling by upper-layer applications. The module specifically includes the following functions:

[0073] Firstly, the multi-source data fusion function, the module serves as an entrance for data access, responsible for connecting and integrating multi-dimensional data sources required for the operation of the central air conditioning system. Including: a time series database (such as TDengine) for storing device operation data, a relational database (such as MySQL) for platform power management system, and an API interface of a third-party weather service, etc. The data query and analysis MCP Server module first completes the connection and integration of all subsystem databases through the MCP protocol as a logical whole. The operation and maintenance personnel do not need to pay attention to the actual location of the data, and can query all categories of data through the unified entrance provided by the module, solving the pain point problem of "multi-system switching query".

[0074] Table 1 shows the integrated central air conditioning operation data, and the specific content of Table 1 is as follows:

[0075] Table 1

[0076]

[0077] Second, the unified call function of heterogeneous databases, the data query and analysis MCP Server module converts the scattered database resources into callable metadata service resources conforming to the MCP protocol through the MCPResource encapsulation mechanism. The specific implementation method is:

[0078] Based on the@mcp.resource decorator, a standardized metadata service resource (such as db: / / metadata / get_schema / {db_name}) is defined, which is identified by a URI conforming to the MCP Server standard and provides information such as database table structure, field annotations, and permission rules. After receiving the call request, the service will use the table header extraction tool to obtain the metadata (such as table structure, field annotations) and access permissions (query permissions, data filtering rules, etc.) of the database.

[0079] Part of the original code implemented in this process is shown below:

[0080] @mcp.resource("db: / / metadata / get_schema / {db_name}")

[0081] def get_schema(db_name: str) ->dict:

[0082] """

[0083] Description: According to the database name, obtain the metadata information (table structure, field annotation, permission rule, etc.) of the database.

[0084] Args (input parameter): db_name (str): database name.

[0085] Returns (return value): dict: dictionary containing database metadata, including table structure, field annotation and permission rule.

[0086] """

[0087] The foregoing code example shows the process of calling the table header extraction tool function and encapsulating the returned result. Specifically, the table header extraction tool is connected to the database, the database table structure, field annotations, and permission rules are obtained, the metadata is organized in a unified format, and a dictionary containing the metadata is returned. The table header extraction tool is implemented to extract the metadata.

[0088] After returning the metadata to the dictionary, the NL2SQL service does not need to directly interface with the underlying database, but only needs to generate the correct sql statement according to the user's intention and the metadata service resource calling the MCP Resource interface, to realize accurate and fast query, and greatly reduce the technical complexity and response time of cross-database query.

[0089] For different database query languages (such as MySQL, TDengine, and other database syntaxes), the application develops corresponding "table header extraction tools". The tool can automatically read the table name, field name, and semantic annotations of the database (for example, mapping the field electricity_consumption to "electricity consumption"), and store these metadata in a unified format, providing structured semantic support for subsequent NL2SQL services.

[0090] In some embodiments, the workflow of the data query and analysis MCP Server module providing natural language data query services to the outside is as follows: when the LLM receives a user's data query request ("Query the running efficiency of No. 2 chiller yesterday afternoon"), it will first call the currently registered MCP Resource to obtain the relevant table structure and field description information. Then, NL2SQL generates an executable SQL statement according to the user's intention and metadata, and LLM is the implementation framework of NL2SQL. The executable SQL statement code example is as follows:

[0091] SELECT timestamp, cop FROM chiller_runtime_data WHERE (device_id = 'chiller-2') AND timestamp BETWEEN 'YYYY-MM-DD 12:00:00' AND 'YYYY-MM-DD 18:00:00'.

[0092] Wherein, timestmap, cop are the timestamp and chiller running COP value of the query, chiller_runtime is the chiller table, device_id = 'chiller-2' indicates No. 2 chiller, and BETWEEN 'YYYY-MM-DD 12:00:00' AND 'YYYY-MM-DD 18:00:00' is the selected relevant time period.

[0093] Subsequently, the generated SQL statement is called as a parameter to another independent MCP Resource query tool (e.g., db: / / execute_query) to perform a query using the SQL statement. The query result is then returned to the LLMs in a structured format. It should be noted that the MCP Resource query tool is optional in this step, and the corresponding database can also be directly queried using the SQL statement.

[0094] In some embodiments, the MCP table generation tool and the process of generating different charts based on the MCP table generation tool can adopt, but are not limited to, the following implementation manners, for example: in order to make the query result more intuitive, the data query and analysis MCP Server module encapsulates the chart generation function as a standard MCP TOOL, namely the MCP table generation tool. After the LLM obtains the original data, the chart generation tool can be called according to the user's instruction (such as "use a line chart to display"), to generate a corresponding chart, so as to facilitate the user to more intuitively understand the data. The specific implementation manner is shown in the following partial original code:

[0095] @mcp.tool()

[0096] def generate_chart(chart_type: str, data: list[dict], x_axis: str, y_axis: list[str], title: str) ->dict:

[0097] """

[0098] Description:

[0099] According to the given structured data, generate a chart of the specified type.

[0100] Args (input parameters):

[0101] chart_type (str): chart type, for example "line", "bar" or "pie".

[0102] data (list[dict]): data used for drawing, data type is a list of dictionaries.

[0103] x_axis (str): specify the field name used as the X-axis of the chart.

[0104] y_axis (list[str]): specify one or more field names used as the Y-axis of the chart.

[0105] title (str): The title of the chart.

[0106] Returns (return value):

[0107] dict: Returns a dictionary containing execution status and graph data (Base64 encoded).

[0108] For example: {"status": "success", "chart_data": "iVBORw0KG...", ...}

[0109] """

[0110] The code example above illustrates the process of calling the chart generation function and encapsulating the returned result.

[0111] In some embodiments, the construction and operation of the energy-saving control MCP Server module includes, but is not limited to, the following implementation methods. For example, the energy-saving control MCP Server module provides a cloud-edge collaborative energy-saving control triggering and execution architecture based on the MCP protocol, used to realize standardized remote calls from cloud-based LLMs to edge-side energy-saving optimization units (AI edge workstations). The specific implementation process is as follows:

[0112] Once the cloud-based LLMs recognize the user's energy-saving intention through natural language interaction (such as "Please adjust the central air conditioning system to its most energy-efficient state"), it does not directly participate in complex numerical calculations. Instead, it generates a goal-oriented MCP energy-saving control tool invocation request, pointing to the swarm intelligence autonomous optimization module of the AI ​​edge workstation, requesting it to calculate the optimal operating strategy. The source code of the MCP energy-saving control tool encapsulation is shown below:

[0113] @mcp.tool()

[0114] def calculate_optimal_strategy(comfort_requirements: dict = None) ->dict:

[0115] """

[0116] Description: The edge workstation invokes the optimal energy-saving operation strategy generated by the swarm intelligence AI autonomous optimization module based on the current real-time operating conditions and given comfort constraints.

[0117] Args (input parameters):

[0118] `comfort_requirements` (dict, optional): Defines the constraints for user comfort. This is a key-value dictionary used to set boundaries such as temperature and humidity for specific areas. For example, to require the indoor temperature to be no higher than 25°C: `{"temp_max": 25}`. If not provided, the system will use the default comfort standard for optimization.

[0119] Returns: dict: A dictionary object containing the computation results or task status. On success, it returns the optimal parameter combination obtained through optimization. For example: {"status": "success", "optimal_parameters": {"chiller_temp": 7.5, "pump_freq": 45.0, ...}}

[0120] """

[0121] The aforementioned code example demonstrates the process of invoking subsequent edge data collection and swarm intelligence AI for autonomous optimization and returning results.

[0122] The energy-saving control MCP Server module deployed on the AI ​​edge workstation continuously listens for tool call requests from cloud-based LLMs. Upon receiving such a request, the energy-saving control MCP Server module directly invokes the device data and control access functions of the device interconnection control MCP Server module also on the edge workstation to collect the latest and most accurate operating parameters (chimney load rate, water pump frequency, temperature and pressure at various points, etc.) of all relevant devices in real time, and combines them with historical data. The collected data is then transmitted to the swarm intelligence autonomous optimization module as input for the optimization algorithm.

[0123] Next, the AI ​​edge workstation loads and executes a swarm intelligence algorithm (AI autonomous optimization algorithm) on its local computing unit. This algorithm uses swarm intelligence algorithms (such as genetic algorithms, artificial bee colony algorithms, and particle swarm optimization) to search for the globally optimal energy-saving operating parameters of the central air conditioning system under complex and multi-constraint conditions. The algorithm first constructs an algorithm library containing various swarm intelligence algorithms and enhances its optimization capabilities by employing multi-algorithm parallelism and random initialization of multiple sets of parameters to avoid getting trapped in local optima. Simultaneously, a physical model constraint module is introduced to ensure that the optimization process conforms to physical laws and safety limits. Finally, through an autonomous optimization mechanism, it combines current environmental data with cosine similarity matching of historical experience. If the current environmental data has a high degree of matching with historical experience, the historical optimization results are directly reused to perform global optimization, solving for the optimal combination of operating parameters that minimizes the energy consumption of the entire central air conditioning system, thus obtaining the optimal energy-saving strategy (such as chilled water outlet temperature setting, optimal frequency of water pumps and air towers, etc.).

[0124] The AI edge workstation pushes the energy-saving optimal strategy to the cloud server, the cloud server LLM interacts with the user, and decides whether to issue according to the user's intention. If issued, the device interconnection control MCP Server module will accept the tool call of the cloud LLM and complete the control of the device.

[0125] As a refinement of the above embodiment, when performing the step 103 of encapsulating the MCP intelligent operation and maintenance question and answer tool on the intelligent operation and maintenance question and answer MCP Server module deployed in the cloud server, the following implementation manners can also be adopted, but are not limited to, for example: collecting operation and maintenance document data and pre-processing the operation and maintenance document data; performing knowledge segment segmentation on the processed operation and maintenance document data, and vectorizing the segmented knowledge segments and storing them in the Milvus vector knowledge base; building a question and answer system based on the framework RAGFLOW, and encapsulating the MCP intelligent operation and maintenance question and answer tool calling the question and answer system on the intelligent operation and maintenance question and answer MCP Server module to complete the building of the intelligent operation and maintenance question and answer MCP Server module.

[0126] As a refinement of the above embodiment, when performing the step of collecting operation and maintenance document data and pre-processing the operation and maintenance document data, the following implementation manners can also be adopted, but are not limited to, for example: collecting the operation and maintenance document data covering various types of devices of the central air conditioning system; checking the operation steps involved in the operation and maintenance document data against the device manufacturer's technical documents and industry operation and maintenance standards to ensure the reliability of the operation and maintenance document data; removing outdated and invalid content in the approved operation and maintenance document data; correcting description logic errors in the corpus through manual review and cross-validation to ensure the accuracy of the operation and maintenance document.

[0127] As a refinement of the above embodiment, when performing the step of performing knowledge segment segmentation on the processed operation and maintenance document data, and vectorizing the segmented knowledge segments and storing them in the Milvus vector knowledge base, the following implementation manners can also be adopted, but are not limited to, for example: segmenting the processed operation and maintenance document data according to different knowledge point types to obtain different types of knowledge segments; using the BGE-m3 word embedding model to convert the knowledge segments into vector data with semantic features; storing the vectorized data in the Milvus vector knowledge base.

[0128] In some embodiments, the building and use process of the intelligent operation and maintenance question and answer MCP Server module can also adopt the following implementation manners, but are not limited to, for example: the intelligent operation and maintenance question and answer MCP Server module integrates a knowledge base, RAG technology and question and answer logic, cooperates with LLM to provide intelligent operation and maintenance services, and provides intelligent support for central air conditioning operation and maintenance, and the specific implementation process is as follows:

[0129] Collect maintenance guidebooks (such as daily maintenance steps, regular maintenance cycles, and replacement methods for vulnerable components) for various equipment in central air conditioning systems (such as air conditioners, fresh air, lighting, elevators, etc.), equipment troubleshooting guides (such as fault phenomenon and reason comparison table, maintenance operation specification), and equipment parameter configuration instructions (such as refrigeration chilled water temperature adjustment range, water pump frequency setting range) as operation and maintenance document data to support the solution to problems such as "equipment maintenance steps" and "how to troubleshoot faults".

[0130] To avoid deviations in the results of the question and answer due to corpus errors and obsolescence, the collected operation and maintenance document data needs to be processed in multiple dimensions, including operation and maintenance step proofreading, obsolete content removal, and error correction. Specifically, by comparing the official technical documents provided by equipment manufacturers and industry operation and maintenance standards, the operation and maintenance steps involved in the corpus are checked to ensure the accuracy and compliance of the operations. Remove the content that is no longer valid due to equipment updates, policy adjustments, and technology upgrades to prevent outdated information from adversely affecting the results of the question and answer. Through manual review and cross-validation, the errors in the corpus are corrected to ensure the reliability of the corpus.

[0131] The intelligent operation and maintenance question and answer MCP Server module is based on the Milvus vector knowledge base for document segmentation and vector storage. Specifically, the standardized PDF format maintenance manuals, Excel format inspection records, and other documents are split into knowledge segments according to "topic relevance", such as "air conditioner daily maintenance" and "air conditioner troubleshooting". Use BGE-m3 and other word embedding models to convert knowledge segments into vector data with semantic features and store them in the Milvus vector knowledge base. Milvus, as a high-performance vector database, supports efficient storage and retrieval of massive vector data.

[0132] To quickly implement the landing and service deployment of intelligent question and answer functions, the intelligent operation and maintenance question and answer MCP Server module is based on the open-source framework RAGFLOW, which natively supports Retrieval Augmented Generation (RAG), is easy to use and efficient to deploy, can significantly reduce the engineering complexity from knowledge base to intelligent question and answer, and can build a question and answer system. Combined with the MCP protocol, the MCP Server service is built and released.

[0133] For equipment parameters, operation, and maintenance of central air conditioning systems, the intelligent operation and maintenance question and answer MCP module generates appropriate answers by analyzing the semantics of the question and calling relevant knowledge sources. For example, when the user asks "how to reduce air conditioner energy consumption", the module will combine air conditioner operation data and energy saving strategy knowledge to provide specific operation suggestions for the user. The specific implementation is to encapsulate the MCP intelligent operation and maintenance question and answer tool, as shown in the original code:

[0134] @mcp.tool()

[0135] def query_om_knowledge_base(question: str) ->dict:

[0136] """

[0137] Description:

[0138] We provide intelligent Q&A services based on a professional knowledge base to address issues related to equipment parameters, operation, and maintenance of central air conditioning systems.

[0139] Args (input parameters):

[0140] question (str): The user's question in natural language.

[0141] For example: "What are the routine maintenance procedures for Refrigeration Unit No. 1?"

[0142] Returns (return value):

[0143] dict: A dictionary object containing question-and-answer results and source information.

[0144] {

[0145] "status": str, / / "success" or "error"

[0146] "answer": str, / / LLM-generated answer text based on a knowledge base

[0147] source_documents: list / / (optional) A list of source knowledge fragments referenced in the answer, used for tracing and verifying the results.

[0148] }

[0149] """

[0150] The code example above illustrates the internal implementation flow of RAG: Vectorize the 'question'. Retrieve relevant documents in Milvus. Build enhanced hints and call LLM to generate the answer. Encapsulate a dictionary containing the answer and source documents and return it.

[0151] As a refinement of the above-mentioned embodiment, in the implementation of the connection of the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy-saving regulation MCP Server module, and the intelligent operation and maintenance question and answer MCP Server module to the MCP client to obtain the central air conditioning management system, the following implementation modes can be used, but are not limited to, for example: connecting the device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy-saving regulation MCP Server module, and the intelligent operation and maintenance question and answer MCP Server module to the MCP client based on the FastMCP architecture; constructing LLMs for single-node deployment or multi-node deployment with the MCP client, establishing a session link between the LLMs and the MCP client, and obtaining the central air conditioning management system.

[0152] As a refinement of the above-mentioned embodiment, the method can also use the following implementation modes, but is not limited to, for example: integrating a multi-modal input interface into the MCP client and deploying a modal conversion middleware to receive different forms of user instructions according to the multi-modal input interface, and converting them into text prompt decision instructions recognizable by the LLMs based on the modal conversion middleware.

[0153] In some embodiments, a unified and integrated MCP Client, i.e., an MPC client, is provided, which seamlessly aggregates device interconnection control, data query and analysis, energy saving regulation, intelligent operation and maintenance question and answer, and multiple independent MCP Server modules, to provide one-stop service for users. The MCP protocol architecture includes MCP Server, LLMs, and MCP Client three ends, to form a user instruction closed loop. The MCP Client serves as an interface for direct interaction with users, so that users can initiate requirements through a unified portal without needing to care about the underlying modules. The FastMCP architecture is the key to multi-service collaboration, and has the characteristics of low-latency communication, service automatic discovery, load balancing, fault tolerance, etc., solving the efficiency and compatibility problems of multi-MCP Server docking, and providing technical support for Client and Server linkage. Based on the FastMCP architecture, multiple MCP Servers can be efficiently docked and communicated. User session management runs through the three-end loop of MCP Server, MCP Client, and LLMs, and context understanding is realized through a memory unit. The intelligence of the MCP protocol is reflected in the ability to remember the context, thereby completing the tasks specified by the user. The MCP Client locally stores the interactive records of the user's input instructions, the return results of the MCP Server, the generated content of the LLMs, and the session number, etc. The user can interact with the AI assistant through multiple ways such as voice and text, and the AI assistant, based on the MCP protocol, collaborates with the central air conditioning device interconnection MCP module, the data query and analysis MCP module, the energy saving regulation agent calling MCP service module, and the intelligent operation and maintenance question and answer MCP service module, to realize comprehensive response to user instructions.

[0154] In order to more clearly show the central air conditioning management and control system constructed based on the MCP protocol-based central air conditioning management and control method of the present application, Figure 2 The framework diagram of the central air conditioning management and control system provided by the embodiments of the present application is shown, Figure 2 It is shown that the central air conditioning device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy saving regulation MCP Server module, the intelligent operation and maintenance question and answer MCP Server module, and the intelligent assistant Client module are shown. In addition, a model unified management module is mentioned, which is used to access different types of big data models, and the access methods include but are not limited to local deployment or API calling.

[0155] In summary, the embodiments of the present application can achieve the following effects:

[0156] Based on the intelligent operation and maintenance question and answer MCP Server module, an operation and maintenance personnel can call an operation and maintenance data knowledge base through a large language model, assist the operation and maintenance personnel, and solve the problems that the central air conditioning system has various types of equipment and various technical parameters, and the operation and maintenance personnel is difficult to master all operation and maintenance knowledge. In terms of equipment fault troubleshooting and maintenance process, it is difficult to obtain timely and accurate technical guidance, and it is easy to cause maintenance progress delay and low efficiency.

[0157] The above is the method embodiment provided by the present application. Based on the same inventive concept, the present application also provides a construction device of a central air conditioning management and control system based on an MCP protocol, which has a structure as shown in Figure 3 .

[0158] Figure 3 The above is the method embodiment provided by the present application. Based on the same inventive concept, the present application also provides a construction device of a central air conditioning management and control system based on an MCP protocol, which has a structure as shown in Figure 3 .

[0159] at least one processor 201;

[0160] and a memory 202 in communication connection with the at least one processor;

[0161] The memory 202 stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor 201 to enable the at least one processor 201 to perform a construction method of a central air conditioning management and control system based on an MCP protocol.

[0162] Some embodiments of the present application provide a non-volatile computer storage medium corresponding to Figure 1 , which stores computer executable instructions, and the computer executable instructions are set to execute a construction method of a central air conditioning management and control system based on an MCP protocol.

[0163] Each of the embodiments in the present application is described in a progressive manner, and the same or similar parts of each of the embodiments can be referred to each other. Each of the embodiments focuses on the differences from other embodiments. In particular, for the Internet of Things device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0164] The system and medium provided by the embodiments of the present application are one-to-one corresponding to the method, so the system and medium also have the similar beneficial technical effects as the method corresponding to them. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and medium will not be described here.

[0165] Those skilled in the art will appreciate that embodiments of the application can be readily used as software, hardware, or a combination of software and hardware. In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0166] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0167] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0168] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0169] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0170] The memory can include non-persistent memory and / or volatile memory, such as a random access memory (RAM) including a cache area for the temporary storage of data. The memory can also include non-volatile memory, such as a read only memory (ROM), EPROM, EEPROM, flash memory, or other non-volatile memory storage. The memory can be another type of computer-readable media, a magnetic-based memory, such as a magnetic disks, magnetic tapes or cassettes, or cards or other types of memory which can store data.

[0171] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0172] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusions, so that a process, method, article or apparatus that includes a list of elements does not only include those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such a process, method, article or apparatus. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0173] The above only describes the embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the scope of claims of the present application.

Claims

1. A method for constructing a central air conditioning control system based on the MCP protocol, characterized in that, The method includes: The device control functions on the AI ​​edge workstation's device interconnection control MCP Server module are encapsulated into an MCP device control tool, which is then used by LLMs to uniformly call the MCP device control tool to control various heterogeneous devices with different protocols. The device interconnection control MCP Server module uses the AI ​​edge workstation deployed in the central air conditioning room as an intelligent edge computing node. Its core architecture includes two parts: a downlink physical device interface and an uplink model service interface. It connects to physical devices downwards and acts as an intelligent gateway to uniformly manage heterogeneous devices with multiple protocols, including chillers, cooling towers, and water pumps. It provides services to LLMs upwards and is encapsulated into an MCP device control tool through the MCP protocol. On the data query and analysis MCP Server module deployed on the cloud server, a metadata service resource is defined based on the MCP Resource decorator, and a header extraction tool for extracting metadata from different types of databases is built. The metadata of different databases extracted by the header extraction tool is encapsulated in the metadata service resource in a unified format, so that NL2SQL can uniformly generate SQL statements for different types of databases based on the metadata and perform data queries on different databases. On the data query and analysis MCP Server module deployed on the cloud server, different chart generation functions are encapsulated into an MCP table generation tool, so that LLMs can uniformly call the MCP table generation tool to generate different charts and perform chart analysis and display on the data queried from the database. On the energy-saving control MCP Server module of the AI ​​edge workstation, the function that calls the swarm intelligence AI autonomous optimization algorithm is encapsulated as an MCP energy-saving control tool, so that the LLMs can call the MCP energy-saving control tool to control the swarm intelligence AI autonomous optimization algorithm to adjust the energy-saving status of the central air conditioning system. An MCP intelligent operation and maintenance Q&A tool is encapsulated on the intelligent operation and maintenance Q&A MCP Server module deployed on a cloud server, so that the LLMs can call the MCP intelligent operation and maintenance Q&A tool to assist in operation and maintenance work; The device interconnection control MCP Server module, the data query and analysis MCP Server module, the energy-saving regulation MCP Server module, and the intelligent operation and maintenance Q&A MCP Server module are connected to the MCP client deployed on the cloud server based on the FastMCP architecture. This constructs an LLMs (Local Management Modules) deployed in a single-node or multi-node configuration with the MCP client. A session link is established between the LLMs and the MCP client to obtain the central air conditioning management system. The MCP client integrates a multimodal input interface and deploys a modality conversion middleware to receive different forms of user instructions based on the multimodal input interface. These instructions are then converted into text-based decision-making instructions recognizable by the LLMs based on the modality conversion middleware. The MCP client is used to interact with the user and invoke the LLMs and the NL2SQL to manage the central air conditioning system.

2. The method for constructing a central air conditioning control system based on the MCP protocol according to claim 1, characterized in that, The process of encapsulating the device control functions on the AI ​​edge workstation's device interconnection control MCP Server module into an MCP device control tool includes: An initial device interconnection control MCP Server module is built in the AI ​​edge workstation, and device control functions are accessed in the initial device interconnection control MCP Server module. The AI ​​edge workstation is used to connect various heterogeneous devices in the central air conditioning system at the physical layer. The device control function is encapsulated based on the MCP protocol specification to obtain the MCP device control tool. The device control function is the underlying control function under the corresponding protocol of each heterogeneous device.

3. The method for constructing a central air conditioning control system based on the MCP protocol according to claim 1, characterized in that, On the MCP Server module for data querying and analysis deployed on a cloud server, metadata service resources are defined based on the MCP Resource decorator, and header extraction tools for extracting metadata from different types of databases are built, including: Metadata service resources are defined based on the MCP Resource decorator, and the paths of the metadata service resources are identified using URIs that conform to the standards of the MCPServer module for data query and analysis, so as to facilitate the retrieval of the metadata service resources based on the URIs; A header extraction tool is constructed to extract metadata from different types of databases, so that the header extraction tool encapsulates the metadata of different types of databases into the corresponding metadata service resource after unifying the metadata format according to the URI.

4. The method for constructing a central air conditioning control system based on the MCP protocol according to claim 1, characterized in that, The encapsulation of the MCP intelligent operation and maintenance Q&A tool on the intelligent operation and maintenance Q&A MCP Server module deployed on a cloud server includes: Collect operation and maintenance document data, and preprocess the operation and maintenance document data; The processed operation and maintenance document data is segmented into knowledge fragments, and the segmented knowledge fragments are vectorized and stored in the Milvus vector knowledge base; A question-and-answer system is built based on the RAGFLOW framework, and an MCP intelligent operation and maintenance question-and-answer tool that calls the question-and-answer system is encapsulated on the intelligent operation and maintenance question-and-answer MCP Server module to complete the construction of the intelligent operation and maintenance question-and-answer MCP Server module.

5. The method for constructing a central air conditioning control system based on the MCP protocol according to claim 4, characterized in that, The process of collecting and preprocessing operation and maintenance document data includes: Collect the operation and maintenance documentation data covering all types of equipment in the central air conditioning system; By comparing the equipment manufacturer's technical documents with industry operation and maintenance standards, the operation and maintenance steps involved in the operation and maintenance document data are verified to ensure the reliability of the operation and maintenance document data; Remove outdated and invalid content from the approved maintenance documentation data; Through manual review and cross-validation, logical errors in the descriptions in the corpus are corrected to ensure the accuracy of the operation and maintenance documents.

6. The method for constructing a central air conditioning control system based on the MCP protocol according to claim 5, characterized in that, The step of segmenting the processed operation and maintenance document data into knowledge fragments and then vectorizing and storing the segmented knowledge fragments in the Milvus vector knowledge base includes: The processed maintenance document data is segmented according to different knowledge point types to obtain different types of knowledge fragments; The knowledge fragments are converted into vector data with semantic features using the BGE-m3 word embedding model; The vector data is stored in the Milvus vector knowledge base.

7. A device for constructing a central air conditioning control system based on the MCP protocol, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to execute a method for constructing a central air conditioning control system based on the MCP protocol as described in any one of claims 1-6.

8. A computer storage medium storing computer-executable instructions, characterized in that, When the computer-executable instructions are executed, the method for constructing a central air conditioning control system based on the MCP protocol as described in any one of claims 1-6 is implemented.

Images