An implementation method and device of an internet of things service based on a large language model

Abstract

This invention discloses a method and apparatus for implementing IoT services based on a large language model. The method involves collecting multi-source data, preprocessing the data, and normalizing the indicators to obtain standardized data. A large language model is used to receive natural language input from users, and contextual semantic parsing is performed to generate structured task intents. Signaling data packets conforming to the corresponding communication protocol specifications are generated using a preset protocol template. The signaling data packets are verified, and upon successful verification, they are distributed to the target IoT device for execution via service orchestration and interaction layer. Execution feedback data from the target IoT device is collected in real time, parsed, and optimized in a closed-loop manner based on the execution results. This invention enables intelligent analysis and interaction of IoT services across multiple industries and scenarios.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and in particular to a method and apparatus for implementing IoT services based on a large language model. Background Technology

[0002] The large-scale application of IoT technology is driving the digital transformation of many industries, such as smart factories and smart energy. Industry users have increasingly urgent needs for advanced intelligence such as natural language interaction, cross-scenario decision-making, and automatic equipment control. Traditional IoT service systems are no longer able to adapt to this development trend and urgently need to break through the bottleneck through the deep integration of artificial intelligence and communication technology.

[0003] Current mainstream IoT smart service solutions have significant limitations: rule engine-based solutions rely on manually preset logic, resulting in insufficient flexibility and maintainability; lightweight machine learning solutions can only handle single tasks, with weak generalization capabilities and support for natural language interaction; knowledge base solutions are limited to shallow semantic matching and lack deep reasoning capabilities; edge computing solutions are constrained by computing power, resulting in a low level of intelligence. Furthermore, existing LLM-IoT convergence solutions lack a secure semantic-to-signaling mapping link and lack cross-protocol adaptation and physical constraint verification capabilities.

[0004] The common shortcomings of these solutions result in insufficient accuracy of natural language interaction, weak cross-scenario comprehensive analysis capabilities, limited depth of knowledge utilization, disconnect between data analysis and device execution, and lack of system scalability and security. They cannot meet the comprehensive needs of industry users for intelligence, adaptability, and security, thus hindering the deep implementation of IoT technology and industrial upgrading. Summary of the Invention

[0005] This invention provides a method and apparatus for implementing Internet of Things (IoT) services based on a large language model, enabling intelligent analysis and interaction of IoT services across multiple industries and scenarios.

[0006] According to one aspect of the present invention, a method for implementing Internet of Things (IoT) services based on a large language model is provided, comprising: Collect multi-source data, including environmental parameters, equipment operating status data, user operation commands, and network performance indicators; preprocess the multi-source data and perform indicator normalization to obtain standardized data. The system receives natural language requests from users through a large language model, and performs contextual semantic parsing on the natural language requests based on an industry knowledge base and protocol knowledge graph to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints and priorities. Using a preset protocol template, signaling data packets conforming to the corresponding communication protocol specifications are generated through semantic alignment and parameter mapping; the signaling data packets are verified, and after successful verification, they are sent to the target IoT device for execution through service orchestration and interaction layer; The system collects execution receipt data from the target IoT device in real time. The receipt data includes ACK responses, status codes, and exception messages. The system then parses the receipt data and performs closed-loop optimization based on the execution results.

[0007] Optionally, the preprocessing of the multi-source data includes: Extreme abnormal data that exceeds the preset reasonable value range in the multivariate data are filtered and removed, and redundant data that are collected repeatedly are deduplicated to obtain preliminary purified data. The preliminary cleaned search data is categorized by data type. Numerical data is converted to a uniform value range using extreme value normalization, while status data is formatted using standardized coding. Abnormal data is identified, and high-frequency fluctuation data caused by sensor jitter is compensated by moving average. For missing data, linear interpolation or prediction based on historical data of the same period is used to reconstruct the data according to its time series characteristics.

[0008] Optionally, the step of performing contextual semantic parsing on the natural language requirement to generate a structured task intent containing the target IoT device, operation type, execution parameters, constraints, and priorities includes: The natural language requirements are processed by word segmentation, part-of-speech tagging and semantic normalization, eliminating pre-set meaningless interjections and redundant expressions, and extracting semantic keywords; Based on a pre-set industry knowledge base and protocol knowledge graph, and combined with the current real-time operating status of IoT devices, historical interaction records, and scenario-based constraint rules, a semantic parsing context is constructed. Based on contextual association analysis, the task objectives in the natural language requirements are identified, the main task and related sub-tasks are distinguished, and the unique identifier of the target IoT device and the device group to which it belongs are determined. From the semantic keywords and context information, extract the operation type, execution parameters, constraints, and task priority; The extracted target IoT device, operation type, execution parameters, constraints, and priority information are combined and encapsulated according to a preset structured format to generate a structured task intent in JSON format.

[0009] Optionally, the step of generating signaling data packets conforming to the corresponding communication protocol specifications through semantic alignment and parameter mapping using a preset protocol template includes: Based on the target device type in the structured task intent, the appropriate communication protocol type is selected from the preset protocol template library, and the corresponding signaling template under the communication protocol is matched according to the operation type. Extract the execution parameters, constraints, and priority information from the structured task intent and match them one-to-one with the preset fields in the signaling template for semantic alignment. The aligned execution parameters are standardized to unify the parameter units, data types, and value formats; the standardized execution parameters are then filled into the signaling template one by one to obtain the initial signaling data packet. The initial signaling data packet is subjected to protocol format compliance verification, which checks whether the field integrity, syntax correctness, and parameter value range conform to the protocol specifications. After the verification is passed, the signaling data packet is generated.

[0010] Optionally, verifying the signaling data packet includes one or more of the following steps: Perform protocol legality checks to verify the compliance of signaling data packet format, field integrity, and QoS level. Obtain the real-time operating status of the target IoT device and verify whether the execution parameters exceed the physical constraint boundaries of the target IoT device; Analyze network load and device processing capacity to determine whether the signaling frequency is within the carrying capacity range; Verify the authorization level of the operation's identity; high-risk operations trigger a multi-factor authentication process. In multi-device collaboration scenarios, a graph neural network is used to construct a topology graph of device dependencies to detect conflicts between execution logic and physical processes.

[0011] Optionally, parsing the receipt data includes: The receipt data is classified according to data format and content characteristics, distinguishing four types: ACK success response, status code feedback, fault alarm message and network timeout receipt; the receipt data of different types are parsed separately to extract the execution completion time, actual execution parameters, status code identifier, fault occurrence node and timeout duration, and the digital status code is mapped to a semantic execution result description to obtain a structured parsing report; The structured parsing report is compared with the preset execution target to determine the execution result, which includes success, parameter deviation, equipment failure, network anomaly, and logical conflict. If the execution result is successful, the structured parsing report, the corresponding signaling data packet, the original natural language requirements and scene feature data will be associated and stored in the experience database as sample data for model optimization. If the execution result is a parameter deviation, extract the deviation parameter type and deviation magnitude, combine the equipment constraint model and historical optimization cases to generate parameter adjustment suggestions, and update the parameter mapping rules; If the execution result is a device failure or network anomaly, record the failure type, triggering conditions and scope of impact, add it to the fault diagnosis knowledge base, and optimize the retry interval and queue priority configuration in the fault tolerance strategy. If the execution result is a logical conflict, analyze the device dependencies or execution timing issues caused by the conflict, adjust the topology inference parameters of the graph neural network, and update the timing rules for multi-device collaborative execution.

[0012] Optionally, the closed-loop optimization based on the execution results includes: By employing reinforcement learning algorithms, the mapping relationship between scenario features and optimal control strategies is extracted by analyzing signaling logs, network feedback, and KPI changes in the execution record, and the inference rules and the large language model Prompt template are updated accordingly. For different application scenarios, cross-scenario adaptive adjustment can be achieved by optimizing the industry knowledge base and constraint verification parameters.

[0013] According to another aspect of the present invention, an apparatus for implementing Internet of Things (IoT) services based on a large language model is provided, comprising: The data acquisition unit is used to collect multi-source data, including environmental parameters, equipment operating status data, user operation commands, and network performance indicators; the multi-source data is preprocessed and the indicators are normalized to obtain standardized data. The task intent construction unit is used to receive natural language requirements input by the user through a large language model, and perform contextual semantic parsing on the natural language requirements based on the industry knowledge base and protocol knowledge graph to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints and priorities. The signaling delivery unit is used to generate signaling data packets that conform to the corresponding communication protocol specifications through semantic alignment and parameter mapping using a preset protocol template; to verify the signaling data packets; and to deliver the signaling data packets to the target IoT device for execution through service orchestration and interaction layer after successful verification. The closed-loop optimization unit is used to collect execution receipt data of the target IoT device in real time. The receipt data includes ACK response, status code and exception message; the unit parses the receipt data and performs closed-loop optimization based on the execution result.

[0014] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to execute the IoT service implementation method based on a large language model as described in any embodiment of the present invention.

[0015] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the Internet of Things service implementation method based on a large language model as described in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the Internet of Things service implementation method based on a large language model as described in any embodiment of the present invention.

[0017] The technical solution of this invention integrates multi-source data preprocessing, natural language semantic parsing, semantic-signaling mapping, multi-layer signaling verification, receipt parsing, and reinforcement learning closed-loop optimization. It also integrates LLM, RAG, GNN, and communication protocol specifications to solve the problems of insufficient interaction accuracy, lack of secure semantic-signaling mapping, lack of cross-protocol adaptation, inconsistent constraints, weak generalization, and lack of self-evolution in existing IoT systems. It achieves beneficial effects such as end-to-end intelligent closed loop, signaling compliance and security, rapid adaptation to multiple scenarios, and self-learning to improve efficiency and reduce costs, thus promoting the evolution of IoT services towards semantic intelligent decision-driven approaches.

[0018] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of an IoT service implementation method based on a large language model provided by an embodiment of the present invention; Figure 2 This is a processing architecture diagram of an IoT service implementation method based on a large language model provided by an embodiment of the present invention; Figure 3This is a schematic diagram of the structure of an IoT service implementation device based on a large language model provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device that implements the IoT service implementation method based on a large language model according to an embodiment of the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] Figure 1 This is a flowchart illustrating a method for implementing IoT services based on a large language model, provided by an embodiment of the present invention. This embodiment is applicable to IoT service implementation scenarios. The method can be executed by an IoT service implementation device based on a large language model. This device can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the method includes: S110. Collect multi-source data, including environmental parameters, equipment operating status data, user operation commands, and network performance indicators; preprocess the multi-source data and perform indicator normalization to obtain standardized data.

[0024] Figure 2This is a processing architecture diagram of an IoT service implementation method based on a large language model provided by an embodiment of the present invention. The architecture mainly includes five layers. The first is the device and perception layer, which includes various IoT sensing devices (sensors, cameras, actuators, etc.) and is responsible for collecting data such as environmental parameters, device operating status, and user operation commands. It connects with the data acquisition layer through gateway access protocols (such as MQTT, CoAP, Modbus, OPCUA). The next layer is the data acquisition and preprocessing layer, which is responsible for cleaning, formatting, feature extraction, and anomaly detection of the raw acquired data. After preliminary processing by edge computing nodes, the data can be uploaded to cloud storage or local cache to ensure the real-time performance and stability of the system.

[0025] Specifically, key data in IoT scenarios are comprehensively collected through IoT devices and various terminals such as sensors, actuators, and monitoring nodes in the perception layer: environmental parameters include physical environmental status data such as temperature, humidity, vibration, and light; device operation status data includes real-time operation information such as device working mode, load rate, energy consumption, and fault alarm indicators; user operation commands refer to control requirements or query requests issued by users through interactive methods such as APP, Web, and voice; and network performance indicators include key parameters that ensure stable device communication, such as data transmission latency, bandwidth utilization, and packet loss rate.

[0026] In this embodiment of the invention, preprocessing of multi-source data includes: Filter and remove extreme abnormal data that exceed the preset reasonable value range from the multivariate data, and perform deduplication on the redundant data collected repeatedly to obtain preliminary purified data; The preliminary cleaned search data is categorized by data type. Numerical data is converted to a uniform value range using extreme value normalization, while status data is formatted using standardized coding. Abnormal data is identified, and high-frequency fluctuation data caused by sensor jitter is compensated by moving average. For missing data, linear interpolation or prediction based on historical data of the same period is used to reconstruct the data according to its time series characteristics.

[0027] Specifically, based on the characteristics of devices in IoT application scenarios, industry standards, and historical operating data, reasonable value ranges for various types of data are pre-defined. For example, the reasonable range for temperature data can be set to -40℃ to 85℃, and the reasonable range for device load rate is 0% to 100%. Extreme abnormal data exceeding these ranges are filtered out and directly removed. At the same time, for redundant data that is repeatedly collected by multiple terminals or collected multiple times by the same terminal, unique and valid data records are retained through data identification comparison and timestamp deduplication. Finally, preliminary purified data is obtained by removing extreme anomalies and redundant information.

[0028] The preliminary purification data is categorized into numerical data (such as continuous or discrete numerical data like temperature, humidity, energy consumption, and bandwidth utilization) and status data (such as non-numerical data like equipment operation / stop status, fault alarm types, and user operation command categories). For numerical data, an extreme value normalization algorithm is used to map data of different magnitudes and units to a unified value range such as [0,1] or [-1,1], eliminating the interference of data scale differences on subsequent analysis. For status data, standardized coding rules (such as binary coding and classification coding) are used to convert textual descriptions of status into a standardized format that can be recognized by computers, achieving format unification for different types of data.

[0029] Statistical analysis and trend detection methods are used to identify high-frequency fluctuations in the preliminary cleaned data caused by sensor jitter, such as irregular and rapid temperature fluctuations within a short period. A moving average method is used to smooth this type of data, replacing fluctuating data points with the average value calculated within a fixed window to reduce jitter interference and restore the true trend of the data. For missing data due to network interruptions or temporary equipment failures during data acquisition, its time series characteristics are analyzed: if the data trend is stable, linear interpolation is used to fit and fill in the missing data based on valid data points before and after it; if the data has a periodic pattern, a prediction method based on historical data from the same period is used, either by calling historical valid data within the same period or by generating supplementary data through a trend prediction model, to achieve accurate reconstruction of the missing data and ensure data integrity and continuity.

[0030] S120. Receive natural language requirements input by the user through a large language model. Based on the industry knowledge base and protocol knowledge graph, perform contextual semantic parsing on the natural language requirements to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints, and priorities.

[0031] like Figure 2 As shown, the large language model intelligent engine layer performs intelligent reasoning and natural language understanding based on the large language model, providing the following functions: multimodal input understanding: processing multi-source inputs such as natural language, sensor data, and images; contextual semantic parsing: combining device data and user instructions to generate high-semantic-level task objectives; decision and reasoning generation: generating operational suggestions or control strategies that conform to industry rules based on knowledge base constraints. The knowledge enhancement layer (RAG) constructs an industry knowledge base, including IoT device operation specifications, fault diagnosis rules, energy consumption optimization strategies, and safety standards. It adopts graph databases (such as Neo4j) or knowledge graph structures to achieve rapid retrieval and reasoning of structured knowledge, while combining with the large language model to provide factual support and industry constraints, avoiding "illusion" output.

[0032] In this embodiment of the invention, natural language requirements undergo contextual semantic parsing to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints, and priorities, including: The natural language processing performs word segmentation, part-of-speech tagging, and semantic normalization to remove meaningless interjections and redundant expressions, and extract semantic keywords. Based on a pre-set industry knowledge base and protocol knowledge graph, and combined with the current real-time operating status of IoT devices, historical interaction records, and scenario-based constraint rules, a semantic parsing context is constructed. Based on contextual association analysis, the task objectives in the natural language requirements are identified, the main task and related sub-tasks are distinguished, and the unique identifier of the target IoT device and the device group to which it belongs are determined. Extract operation type, execution parameters, constraints, and task priority from semantic keywords and contextual information; The extracted target IoT device, operation type, execution parameters, constraints, and priority information are combined and encapsulated according to a preset structured format to generate a structured task intent in JSON format.

[0033] Specifically, the natural language input from the user is first preprocessed. The complete sentence is broken down into independent lexical units using a word segmentation algorithm. Then, the grammatical attributes of each word are clarified through part-of-speech tagging. Semantic normalization is then performed, mapping synonyms and near-synonyms to standardized expressions. For example, "lower temperature" and "adjust temperature" are normalized to "lower temperature". At the same time, referring to a pre-set meaningless vocabulary library containing modifiers, auxiliary words, and repetitive modifiers, expressions that do not affect the core semantics, such as "ah", "oh", and "very", are removed. Semantic keywords related to device operation, task objectives, and parameter requirements are then selected from the processed text.

[0034] It calls upon a pre-defined industry knowledge base (including equipment operation specifications, fault handling rules, industry safety standards, etc.) and a protocol knowledge graph (including device associations, communication protocol adaptation rules, etc.), and simultaneously obtains the current operating status of IoT devices in real time (such as whether they are online, load status, current parameter values), user historical interaction records (such as past operating habits, commonly used parameter settings), and scenario-based constraint rules (such as production period constraints in smart factories and energy consumption threshold limits in smart energy). It integrates multi-dimensional information into a semantic parsing context environment, providing industry background, equipment status, and historical reference support for accurately understanding user needs.

[0035] Based on the constructed semantic parsing context, the extracted semantic keywords are analyzed for correlation. The task objectives of user needs (such as "reducing workshop energy consumption") are identified, and related sub-tasks (such as "adjusting air conditioning temperature" and "turning off idle equipment") are mined to clarify the hierarchical relationship between the main task and sub-tasks. Combining the unique identifier library of equipment (such as equipment number and MAC address) and equipment group classification rules, the unique identifier of the target IoT device is locked from the context information, and its equipment group (such as "workshop refrigeration equipment group" and "power pump group") is determined.

[0036] From the preprocessed semantic keywords and the integrated context, four types of core information are accurately extracted: operation type, which is the specific action that the device needs to perform (such as start, stop, adjust, detect, alarm); execution parameters, which are the specific values ​​or attributes required for the operation (such as temperature adjustment range "-2℃", delay time "600 seconds"); constraints, including time window (such as "18:00-22:00"), device operating boundary (such as "power not exceeding 5kW"), safety requirements (such as "confirm that the device is fault-free before operation"), and other restrictions; task priority is divided into three levels: high, medium, and low according to the urgency of the scenario and the importance of user needs (such as setting device fault handling as high level and routine parameter adjustment as medium level).

[0037] According to the pre-set unified data structure specifications, the core information extracted from the target IoT device, such as unique identifier, operation type, execution parameters, constraints and task priority, is combined in an orderly manner and encapsulated in JSON format. The structured task intent generated after encapsulation realizes the transformation from natural language semantics to standardized data, that is, the transformation from user needs to device execution.

[0038] S130. Using a preset protocol template, signaling data packets conforming to the corresponding communication protocol specifications are generated through semantic alignment and parameter mapping. The signaling data packets are verified, and after successful verification, they are sent to the target IoT device for execution through service orchestration and interaction layer.

[0039] like Figure 2 As shown, the design goal of the communication signaling inference layer is to achieve intelligent mapping and closed-loop verification between "semantic layer intent" and "communication layer execution signaling", and to have the ability to generate, optimize and verify communication signaling while understanding language.

[0040] In this embodiment of the invention, a signaling data packet conforming to the corresponding communication protocol specification is generated through a preset protocol template, semantic alignment, and parameter mapping, including: Based on the target device type in the structured task intent, the appropriate communication protocol type is selected from the preset protocol template library, and the corresponding signaling template under the communication protocol is matched according to the operation type. Extract the execution parameters, constraints, and priority information from the structured task intent and match them one-to-one with the preset fields in the signaling template for semantic alignment. The aligned execution parameters are standardized to unify the parameter units, data types, and value formats; the standardized execution parameters are then filled into the signaling template one by one to obtain the initial signaling data packet. The initial signaling data packet is validated for protocol format compliance, checking the integrity of fields, the correctness of syntax, and whether the range of parameter values ​​conforms to the protocol specifications. Once the validation is successful, the signaling data packet is generated.

[0041] Specifically, the system first retrieves a pre-defined protocol template library, which stores a full set of template resources corresponding to mainstream communication protocols in IoT scenarios. Based on the structured task intent, the target device type is extracted. Using the device type and protocol adaptation mapping rules, communication protocol types compatible with the device type are selected from the template library. Then, based on the operation type in the structured task intent, the corresponding signaling template is matched under the selected communication protocol to ensure that the template is compatible with the device type and operation requirements.

[0042] The system extracts key information from the structured task intent, including specific execution parameters and constraints. Simultaneously, it parses the selected signaling template to determine its preset field system (such as parameter fields, constraint fields, priority identifier fields, etc.). The extracted key information is then matched one-to-one with the template fields according to semantic association, achieving precise alignment between the requirement semantics and the template fields, thus avoiding invalid signaling caused by information misalignment.

[0043] The execution parameters after semantic alignment are standardized in format: the parameter units are unified, the data types are standardized, and the value formats are unified to ensure that the parameters meet the format requirements of the communication protocol. After the parameters are standardized, each parameter is filled into the corresponding field one by one according to the field order of the signaling template to complete the full assignment of the template fields and form an initial signaling data packet containing complete communication information.

[0044] The initial signaling data packet undergoes multi-dimensional verification. First, it checks field completeness, confirming that all required fields in the template are filled in without any missing items. Second, it verifies syntax correctness, checking that the data packet's field naming, structural hierarchy, and symbol usage conform to the syntax rules of the corresponding communication protocol. Third, it verifies the parameter value range, determining whether each parameter's value is within the reasonable range allowed by the protocol, with no out-of-range cases. If all verification items meet the protocol specifications, the verification is considered passed, and the initial signaling data packet officially becomes a signaling data packet ready for distribution. If any verification fails, the previous steps are returned for parameter adjustment or template re-matching.

[0045] In this embodiment of the invention, verifying signaling data packets includes one or more of the following steps: Perform protocol legality checks to verify the compliance of signaling data packet format, field integrity, and QoS level. Obtain the real-time operating status of the target IoT device and verify whether the execution parameters exceed the physical constraint boundaries of the target IoT device; Analyze network load and device processing capacity to determine whether the signaling frequency is within the carrying capacity range; Verify the authorization level of the operation's identity; high-risk operations trigger a multi-factor authentication process. In multi-device collaboration scenarios, a graph neural network is used to construct a topology graph of device dependencies to detect conflicts between execution logic and physical processes.

[0046] Specifically, check whether the overall format of the data packet conforms to the syntax structure specified by the protocol, including whether the field hierarchy, symbol usage, and data encapsulation method are correct; check one by one whether the required fields required by the protocol are completely filled without omissions or empty values; verify whether the QoS level is compatible with the target device's support range and business scenario requirements, and ensure that the signaling data packet has the basic conditions to be correctly parsed and processed by the device at the protocol level.

[0047] The system acquires real-time operating data of the target IoT device through the device status acquisition channel, including physical constraint parameters such as rated power, maximum load, operating temperature range, and upper / lower limits for parameter adjustment. The execution parameters in the signaling data packet are compared one by one with the above physical constraint parameters to determine whether the execution parameters are within a reasonable range that the device can bear, thus avoiding device failure, damage, or execution failure due to parameters exceeding the device's hardware capabilities.

[0048] Real-time data collection of network load metrics such as bandwidth utilization, data transmission latency, and packet loss rate; simultaneous acquisition of target device processing capacity data such as task processing queue length and current CPU / memory utilization; combined with signaling packet size and distribution frequency requirements, comprehensive calculation of network transmission pressure and device processing pressure; determination of whether the signaling distribution frequency will exceed the network's real-time transmission capacity or the device's concurrent processing capacity; prevention of network congestion, data loss, or device response timeouts caused by excessively dense signaling distribution.

[0049] The system retrieves the preset permission management database to query the identity of the user or system initiating the operation and the corresponding authorization level; it then determines the risk level based on the operation type; ordinary risky operations only require verification that the authorization level matches the operation permission to pass, while high-risk operations automatically trigger a multi-factor authentication process, and only after successful authentication are subsequent steps allowed to proceed, thus preventing security risks caused by unauthorized operations.

[0050] When signaling involves multiple devices coordinating to perform tasks, a graph neural network algorithm is invoked to construct a device dependency topology graph with devices as nodes and the relationships between devices as edges. Based on this topology graph, the execution order, resource consumption requirements, and action triggering conditions of each device are analyzed to detect whether there are logical or physical conflicts and to identify risks of collaborative execution in advance.

[0051] S140. Real-time collection of execution receipt data from the target IoT device. The receipt data includes ACK response, status code and exception message. The receipt data is parsed and closed-loop optimization is performed based on the execution results.

[0052] like Figure 2 As shown, the service orchestration and interaction layer is responsible for transforming the output of the intelligent engine into executable service processes. It provides multiple interaction methods (APP, Web, voice interaction, API interface). Results can be sent to IoT devices for automated execution.

[0053] In this embodiment of the invention, parsing the receipt data includes: The receipt data is classified according to data format and content characteristics, distinguishing four types: ACK success response, status code feedback, fault alarm message and network timeout receipt; different types of receipt data are parsed separately to extract execution completion time, actual execution parameters, status code identifier, fault occurrence node and timeout duration, and the digital status code is mapped to semantic execution result description to obtain a structured parsing report; The structured parsing report is compared with the preset execution target to determine the execution result, which includes success, parameter deviation, equipment failure, network anomaly, and logical conflict. If the execution result is successful, the structured parsing report, the corresponding signaling data packet, the original natural language requirements and scene feature data will be associated and stored in the experience database as sample data for model optimization. If the execution result is parameter deviation, extract the deviation parameter type and deviation magnitude, combine the equipment constraint model and historical optimization cases to generate parameter adjustment suggestions, and update the parameter mapping rules; If the execution result is a device failure or network anomaly, record the failure type, triggering conditions, and scope of impact, and add it to the fault diagnosis knowledge base. At the same time, optimize the retry interval and queue priority configuration in the fault tolerance strategy.

[0054] Specifically, based on the format specifications and content characteristics of the receipt data, the receipt data returned by the device is classified, clearly distinguishing four core types: ACK success response, status code feedback, fault alarm message, and network timeout receipt. Targeted analysis is performed for each type of receipt data: the execution completion time and actual execution parameters are extracted from the ACK success response; status code identifiers are captured from the status code feedback and their corresponding meanings are associated; the fault occurrence node is located from the fault alarm message; and the timeout duration is recorded from the network timeout receipt. Simultaneously, through preset status code mapping rules, abstract numerical status codes are transformed into semantic descriptions such as "execution successful" and "device busy," integrating all extracted information to obtain a structured analysis report.

[0055] The system retrieves the pre-set execution target and compares the actual execution status in the structured parsing report with the preset target item by item, determining the execution result as follows: if the actual execution parameters, completion time, etc., completely match the preset target, it is determined as "success"; if the actual execution parameters differ from the preset target but the equipment is running normally, it is determined as "parameter deviation"; if the parsing report contains fault identifiers and abnormal descriptions, it is determined as "equipment failure"; if the network timeout receipt or feedback reflects network transmission abnormalities, it is determined as "network abnormality"; if the feedback in a multi-device collaborative scenario shows information such as disordered execution order or resource contention, it is determined as "logical conflict".

[0056] When the execution result is determined to be "successful," the key data from the entire interaction process are correlated and integrated: this includes a structured parsing report recording the actual execution status, signaling data packets that triggered device operations, the user's initial natural language input requirements, and scenario feature data during execution. This correlated data is then uniformly stored in an experience database, serving as an effective sample for semantic parsing of large language models and optimization of signaling generation rules.

[0057] If the execution result is "equipment failure" or "network anomaly," key information is recorded in detail from the structured parsing report: fault type, triggering conditions, and scope of impact. This information is then categorized and added to the fault diagnosis knowledge base of the knowledge enhancement layer to increase the reference basis for fault identification and handling. Simultaneously, based on the triggering scenarios of faults or anomalies, the system's fault tolerance strategy is optimized: signaling retry intervals are adjusted, and task queue priorities are optimized to improve the efficiency of handling similar faults or anomalies.

[0058] In this embodiment of the invention, closed-loop optimization is performed based on the execution result, including: By employing reinforcement learning algorithms, the mapping relationship between scenario features and optimal control strategies is extracted by analyzing signaling logs, network feedback, and KPI changes in the execution record, and the inference rules and large language model Prompt template are updated. For different application scenarios, cross-scenario adaptive adjustment can be achieved by optimizing the industry knowledge base and constraint verification parameters.

[0059] Specifically, the entire set of device execution records is retrieved, and signaling logs, network feedback, and KPI changes are integrated to construct a multi-dimensional analysis dataset. Reinforcement learning algorithms are used to deeply mine this dataset, aiming for optimal execution performance. Through multiple rounds of iterative training, the precise mapping relationship between scenario features and optimal control strategies in different IoT application scenarios is extracted. Based on this mapping relationship, the inference rules are iteratively updated, and the Prompt template of the large language model is optimized. The semantic guidance direction, field constraint rules, and scenario-based prompts of the template are adjusted to improve the model's parsing accuracy for natural language needs in different scenarios and the generation quality of structured task intents, making the control strategy more closely match the actual scenario requirements.

[0060] For different application scenarios in the Internet of Things (IoT) field, such as smart factories, smart energy, and smart agriculture, feature clustering and unique identification are performed. Considering the differences in industry characteristics, equipment types, business operation specifications, and security requirements of each scenario, targeted optimization of the industry knowledge base is carried out. This includes supplementing each scenario with scenario-specific equipment operation rules, fault handling solutions, industry security standards, and equipment collaborative execution logic, while removing generalized knowledge that is inconsistent with scenario characteristics, giving the industry knowledge base scenario-adaptability. Simultaneously, constraint verification parameters are adjusted to address the differences in equipment physical constraint boundaries, network carrying capacity, and business priority settings across different scenarios. This includes threshold ranges for equipment physical constraints, network load judgment criteria, signaling frequency limits, high-risk operation classification, and authentication rules. Through scenario-based optimization of the industry knowledge base and personalized adjustment of constraint verification parameters, the entire process, from semantic parsing and signaling generation to verification and distribution, can adapt to the personalized needs of different application scenarios. This allows for cross-scenario adaptive adjustment without manual reconfiguration, improving execution reliability and adaptation efficiency across multiple scenarios.

[0061] The following detailed description of the solution of the present invention is provided through a specific embodiment: In telecommunications networks, service demands often originate from policy directives from operations teams or network optimization centers, such as improving service quality in specific areas during major events, holidays, or sudden traffic surges. The data acquisition and preprocessing layer primarily assumes the role of "network vital signs awareness." It collects KPI data (including PRB utilization, RSRP, uplink call drop rate, latency, packet loss rate, etc.) from base station cells in real time through various sensor sources, monitoring nodes, and service probes. Simultaneously, it accesses performance indicators from the bearer network and core network (such as link bandwidth utilization, latency jitter, and SMF / PCF policy traffic statistics). This data undergoes preprocessing at edge nodes, completing anomaly detection, data cleaning, and feature extraction. For example, it automatically identifies sudden loads, signaling congestion, or power imbalances, transforming them into standardized structured inputs for upper-layer modules to perform semantic reasoning and decision generation.

[0062] Upon entering the Knowledge Enhancement Aggregator (RAG) layer, the operator's internal technical knowledge and network configuration rules are introduced into the inference context, including 3GPP protocol specifications, vendor equipment parameters, PCC policy templates, and QoS level definitions. When receiving a service request, the RAG module proactively retrieves relevant knowledge fragments and injects them into the large model hints, thereby ensuring that subsequent inference results comply with operational security boundaries and policy constraints. For example, when the marketing department requests to "improve the video uplink experience in the stadium area between 18:00 and 22:30," the RAG module automatically provides slice configuration templates associated with this goal, AMBR and 5QI level rules, power adjustment limits, and QoE assurance experience data, providing the model with the contextual basis needed for decision-making.

[0063] At the large model engine layer, a language model fine-tuned with telecommunications knowledge (such as GPT or LLaMA variants) is used for semantic understanding and task parsing. The model transforms natural language requirements into structured task descriptions, clearly defining key elements such as "regional scope, time window, target KPI, service type, and priority." For example, the statement "Improve the uplink experience of sports center live streaming apps" is parsed as: target area is sports venues, service type is video uplink, performance target is uplink P95 rate ≥ 8Mbps, and strategy type is temporary enhancement. The output at this stage is an intermediate-level task plan, but it has not yet reached the executable level.

[0064] The true core technology resides at the communication signaling layer. This layer transforms high-level semantic tasks into cross-domain protocol signaling plans through a continuous mechanism of "semantic mapping—reasoning—verification." The module first invokes the protocol templates provided by the knowledge enhancement layer to determine the signaling control points for the RAN, core network, and bearer layer. For example, the RAN side needs to adjust PRB allocation and power, the core network side involves PCF policy changes and SMF session renegotiation, and the bearer layer uses the SDN controller to implement temporary bandwidth reservation. Based on these multi-domain relationships, the communication signaling reasoning module automatically generates signaling sequences and performs constraint verification and risk assessment for each step. It simulates and checks parameter boundaries, QoS levels, and conflict policies to ensure that new configurations do not disrupt existing services. After successful verification, the module issues the signaling in transactional form, with each signaling message having an acknowledgment and rollback mechanism, achieving secure and reversible configuration execution.

[0065] Figure 3 This is a schematic diagram of the structure of an IoT service implementation device based on a large language model, provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: The data acquisition unit 310 is used to collect multi-source data, including environmental parameters, equipment operating status data, user operation commands, and network performance indicators; the multi-source data is preprocessed and the indicators are normalized to obtain standardized data.

[0066] The task intent construction unit 320 is used to receive natural language requirements input by the user through a large language model, and perform contextual semantic parsing on the natural language requirements based on the industry knowledge base and protocol knowledge graph to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints and priorities.

[0067] The signaling delivery unit 330 is used to generate signaling data packets that conform to the corresponding communication protocol specifications through semantic alignment and parameter mapping using a preset protocol template; to verify the signaling data packets; and to deliver them to the target IoT device for execution through service orchestration and interaction layer after successful verification.

[0068] The closed-loop optimization unit 340 is used to collect execution receipt data of the target IoT device in real time. The receipt data includes ACK response, status code and exception message; the receipt data is parsed and closed-loop optimization is performed based on the execution result.

[0069] The IoT service implementation device based on a large language model provided in this embodiment of the invention can execute the IoT service implementation method based on a large language model provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0070] Figure 4A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0071] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0072] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0073] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as methods for implementing IoT services based on large language models.

[0074] In some embodiments, the large language model-based IoT service implementation method can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the large language model-based IoT service implementation method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to execute the large language model-based IoT service implementation method by any other suitable means (e.g., by means of firmware).

[0075] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0076] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0077] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0078] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0079] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0080] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0081] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0082] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for implementing Internet of Things (IoT) services based on a large language model, characterized in that, include: Collect multi-source data, including environmental parameters, equipment operating status data, user operation commands, and network performance indicators; The multi-source data is preprocessed and the index is normalized to obtain standardized data; The system receives natural language requests from users through a large language model, and performs contextual semantic parsing on the natural language requests based on an industry knowledge base and protocol knowledge graph to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints and priorities. Using a preset protocol template, signaling data packets conforming to the corresponding communication protocol specifications are generated through semantic alignment and parameter mapping; the signaling data packets are verified, and after successful verification, they are sent to the target IoT device for execution through service orchestration and interaction layer; The system collects execution receipt data from the target IoT device in real time. The receipt data includes ACK responses, status codes, and exception messages. The system then parses the receipt data and performs closed-loop optimization based on the execution results.

2. The method according to claim 1, characterized in that, The preprocessing of the multi-source data includes: Extreme abnormal data that exceeds the preset reasonable value range in the multivariate data are filtered and removed, and redundant data that are collected repeatedly are deduplicated to obtain preliminary purified data. The preliminary cleaned search data is categorized by data type. Numerical data is converted to a uniform value range using extreme value normalization, while status data is formatted using standardized coding. Abnormal data is identified, and high-frequency fluctuation data caused by sensor jitter is compensated by moving average. For missing data, linear interpolation or prediction based on historical data of the same period is used to reconstruct the data according to its time series characteristics.

3. The method according to claim 1, characterized in that, The process of performing contextual semantic parsing on the natural language requirements to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints, and priorities includes: The natural language requirements are processed by word segmentation, part-of-speech tagging and semantic normalization, eliminating pre-set meaningless interjections and redundant expressions, and extracting semantic keywords; Based on a pre-set industry knowledge base and protocol knowledge graph, and combined with the current real-time operating status of IoT devices, historical interaction records, and scenario-based constraint rules, a semantic parsing context is constructed. Based on contextual association analysis, the task objectives in the natural language requirements are identified, the main task and related sub-tasks are distinguished, and the unique identifier of the target IoT device and the device group to which it belongs are determined. Extract the operation type, execution parameters, constraints, and task priority from the semantic keywords and context information; The extracted target IoT device, operation type, execution parameters, constraints, and priority information are combined and encapsulated according to a preset structured format to generate a structured task intent in JSON format.

4. The method according to claim 1, characterized in that, The process of generating signaling data packets conforming to the corresponding communication protocol specifications through a preset protocol template, semantic alignment, and parameter mapping includes: Based on the target device type in the structured task intent, the appropriate communication protocol type is selected from the preset protocol template library, and the corresponding signaling template under the communication protocol is matched according to the operation type. Extract the execution parameters, constraints, and priority information from the structured task intent and match them one-to-one with the preset fields in the signaling template for semantic alignment. The aligned execution parameters are standardized to unify the parameter units, data types, and value formats; the standardized execution parameters are then filled into the signaling template one by one to obtain the initial signaling data packet. The initial signaling data packet is subjected to protocol format compliance verification, which checks whether the field integrity, syntax correctness, and parameter value range conform to the protocol specifications. After the verification is passed, the signaling data packet is generated.

5. The method according to claim 1, characterized in that, The verification of the signaling data packet includes one or more of the following steps: Perform protocol legality checks to verify the compliance of signaling data packet format, field integrity, and QoS level. Obtain the real-time operating status of the target IoT device and verify whether the execution parameters exceed the physical constraint boundaries of the target IoT device; Analyze network load and device processing capacity to determine whether the signaling frequency is within the carrying capacity range; Verify the authorization level of the operation's identity; high-risk operations trigger a multi-factor authentication process. In multi-device collaboration scenarios, a graph neural network is used to construct a topology graph of device dependencies to detect conflicts between execution logic and physical processes.

6. The method according to claim 1, characterized in that, The parsing of the receipt data includes: The receipt data is classified according to data format and content characteristics, distinguishing four types: ACK success response, status code feedback, fault alarm message and network timeout receipt; the receipt data of different types are parsed separately to extract the execution completion time, actual execution parameters, status code identifier, fault occurrence node and timeout duration, and the digital status code is mapped to a semantic execution result description to obtain a structured parsing report; The structured parsing report is compared with the preset execution target to determine the execution result, which includes success, parameter deviation, equipment failure, network anomaly, and logical conflict. If the execution result is successful, the structured parsing report, the corresponding signaling data packet, the original natural language requirements and scene feature data will be associated and stored in the experience database as sample data for model optimization. If the execution result is a parameter deviation, extract the deviation parameter type and deviation magnitude, combine the equipment constraint model and historical optimization cases to generate parameter adjustment suggestions, and update the parameter mapping rules; If the execution result is a device failure or network anomaly, record the failure type, triggering conditions and scope of impact, add it to the fault diagnosis knowledge base, and optimize the retry interval and queue priority configuration in the fault tolerance strategy. If the execution result is a logical conflict, analyze the device dependencies or execution timing issues caused by the conflict, adjust the topology inference parameters of the graph neural network, and update the timing rules for multi-device collaborative execution.

7. The method according to claim 1, characterized in that, The closed-loop optimization based on the execution results includes: By employing reinforcement learning algorithms, the mapping relationship between scenario features and optimal control strategies is extracted by analyzing signaling logs, network feedback, and KPI changes in the execution record, and the inference rules and the large language model Prompt template are updated accordingly. For different application scenarios, cross-scenario adaptive adjustment can be achieved by optimizing the industry knowledge base and constraint verification parameters.

8. An Internet of Things (IoT) service implementation device based on a large language model, characterized in that, include: The data acquisition unit is used to collect multi-source data, including environmental parameters, equipment operating status data, user operation commands, and network performance indicators. The multi-source data is preprocessed and the index is normalized to obtain standardized data; The task intent construction unit is used to receive natural language requirements input by the user through a large language model, and perform contextual semantic parsing on the natural language requirements based on the industry knowledge base and protocol knowledge graph to generate a structured task intent that includes the target IoT device, operation type, execution parameters, constraints and priorities. The signaling delivery unit is used to generate signaling data packets that conform to the corresponding communication protocol specifications through semantic alignment and parameter mapping using a preset protocol template; The signaling data packet is verified, and after successful verification, it is sent to the target IoT device for execution through the service orchestration and interaction layer; The closed-loop optimization unit is used to collect execution receipt data of the target IoT device in real time. The receipt data includes ACK response, status code and exception message; the unit parses the receipt data and performs closed-loop optimization based on the execution result.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the IoT service implementation method based on a large language model as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the Internet of Things service implementation method based on a large language model as described in any one of claims 1-7.

11. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the Internet of Things service implementation method based on a large language model according to any one of claims 1-7.

Images