Intelligent management and control system based on large model and integrated multi-dimensional spatio-temporal data fusion

Abstract

The application discloses an intelligent management and control system based on a large model and integrated multi-dimensional space-time data fusion, and belongs to the fields of artificial intelligence and engineering application. The system comprises a space-time large model base, an engineering object three-dimensional or four-dimensional integrated model, an intelligent agent cooperation module and an autonomous analysis, decision and control module. Based on embedding engineering field knowledge class public domain data into a large model, a space-time large model base is formed, engineering object three-dimensional or four-dimensional geometric and attribute data, monitoring and monitoring and equipment automation and other dynamic business data are fused, an integrated expression, storage and management private domain data source is constructed, autonomous analysis and decision are made by using the reasoning capability of the space-time large model, intelligent agent cooperation is dispatched to complete task planning arrangement, safety production analysis and decision support and management and control, and various engineering graphic and text reports are automatically generated. The application can solve the intelligent management and control problems in the fields of mines, factories, traffic, water conservancy, buildings and machinery, and significantly improves the practicability of the artificial intelligence system in the engineering field.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and engineering, and more specifically, to an intelligent control system based on large models and integrated multidimensional spatiotemporal data fusion. Background Technology

[0002] With the rapid development of large-scale artificial intelligence (AI) modeling technology, its application in various industries is becoming increasingly widespread. However, in engineering fields (such as mining, chemical industry, water conservancy, urban construction, and environmental governance), the in-depth application and value realization of large-scale modeling technology still face many challenges.

[0003] Existing engineering management platforms or systems often have the following problems:

[0004] (1) Weak spatiotemporal correlation processing capability. Engineering objects (such as geological bodies, mine roadways, buildings, and pipelines) and their environment have complex three-dimensional spatial structures and temporal evolution characteristics. Existing systems have difficulty effectively processing and understanding this inherent spatiotemporal correlation. In particular, they lack the ability to process and analyze multidimensional vector graphics commonly used in the engineering field (such as CAD, BIM, and GIS data), making it impossible to perform in-depth spatiotemporal reasoning and limiting the application of large models.

[0005] (2) Difficulty in data fusion and utilization. The entire lifecycle of engineering management (covering planning, design, construction, operation, maintenance and even abandonment) generates massive amounts of multi-source heterogeneous data, including 3D geometry and attribute models in the design phase, progress and material data in the construction phase, monitoring and operation data in the operation phase, and management data such as safety, production and operation. These dynamic data streams have different standards, diverse formats, non-standardized storage directory systems and inconsistent spatiotemporal benchmarks, making it difficult to effectively integrate them to form a globally consistent cognitive foundation, resulting in serious data silos.

[0006] (3) Insufficient representativeness and completeness of the data hinders the improvement of model training and control application accuracy. Spatiotemporal data are often two-dimensional or three-dimensional graphics or images of a single object or type, lacking an integrated representation of the geometry of the engineering object and its associated attribute data as completely as possible. The representation and retrieval of data are discrete, rather than derived from a unified data base.

[0007] (4) Delayed business response and insufficient intelligent decision-making. Engineering management faces a dynamically changing environment and complex business needs. Existing systems often respond based on fixed rules or simple models, lacking a deep understanding of complex business logic and the ability to accurately predict future trends, resulting in delayed business response. The decision-making process still relies heavily on human experience and lacks optimization decision support based on global information and intelligent algorithms, making it difficult to cope with complexity and uncertainty.

[0008] (5) Difficulty in integrating AI technology with business operations. While general-purpose AI models are powerful, they lack professional knowledge in the engineering field and understanding of business processes, resulting in limited effectiveness when directly applied to engineering scenarios. How to deeply integrate the capabilities of large-scale AI models with core engineering businesses (such as planning and design, risk assessment, schedule optimization, and safety management) to form an end-to-end intelligent solution is a technical challenge that urgently needs to be addressed.

[0009] Therefore, there is an urgent need for a new technical solution that can overcome the above-mentioned shortcomings, make full use of the capabilities of AI large models, and combine the spatiotemporal data characteristics of the engineering field to achieve intelligent analysis, decision-making, and adaptive management and control of the entire life cycle of engineering fields such as smart mines, smart factories, water conservancy projects, urban engineering, and environmental engineering. Summary of the Invention

[0010] In view of the above problems, the present invention aims to overcome the shortcomings of the prior art and provide an intelligent control system based on large model and integrated multidimensional spatiotemporal data fusion.

[0011] This invention provides an intelligent control system based on a large model and integrated multidimensional spatiotemporal data fusion, comprising:

[0012] The spatiotemporal large model base is used to receive engineering-related input information and, based on its deep understanding, knowledge association and logical reasoning of the data provided by the engineering domain professional knowledge, the engineering-related input information and the three-dimensional or four-dimensional integrated model of the engineering object, to perform intelligent task planning and arrangement. The input information includes at least business requirements and multimodal data.

[0013] A three-dimensional or four-dimensional integrated model of an engineering object, under a unified spatiotemporal reference, is based on the geometric and attribute data of all elements, three-dimensional or four-dimensional of the engineering object, and integrates dynamic private domain data related to engineering business. Through integrated spatiotemporal expression and processing, a spatiotemporal data body reflecting the true overall picture and full life cycle of the project is constructed and dynamically updated. This spatiotemporal data body serves as the three-dimensional or four-dimensional integrated model of the engineering object, which is used to respond to data requests and is processed, generated or integrated to form global spatiotemporal situation information.

[0014] The intelligent agent collaboration module contains multiple functional intelligent agents and provides a communication and collaboration mechanism between the functional intelligent agents. The intelligent agent collaboration module is used to receive task instructions corresponding to intelligent task planning and orchestration issued by the spatiotemporal large model base, and under the scheduling and monitoring of the autonomous analysis decision and control module, it obtains the data, analysis results or generative content required to execute the task from the three-dimensional or four-dimensional integrated model of the engineering object, and drives the functional intelligent agents to execute.

[0015] The autonomous analysis, decision-making, and control module is used to acquire global spatiotemporal situational information from the three-dimensional or four-dimensional integrated model of the engineering object, receive execution feedback results from the functional intelligent agent collaboration module, and utilize the reasoning capabilities of the spatiotemporal large model base for the engineering domain, combined with the global spatiotemporal situational information, the execution feedback results, and preset or dynamically generated rules and objectives, to perform autonomous analysis and judgment. Based on the results of the autonomous analysis and judgment, it autonomously schedules the intelligent agent collaboration module to execute subsequent tasks and generates intelligent decision results or control instructions.

[0016] Optionally, the spatiotemporal large model base is an intelligent model with the ability to deeply understand and process problems specific to the engineering field, wherein the ability to deeply understand and process problems specific to the engineering field is obtained through an adaptation method;

[0017] The adaptation method includes a combination of one or more of the following technologies:

[0018] Domain fine-tuning technology refers to the use of large-scale text corpora, code corpora, and various public domain data of multimodal data in the engineering field for supervised fine-tuning, instruction fine-tuning, or continuous pre-training of intelligent models.

[0019] Knowledge enhancement technology refers to the following: combining a spatiotemporal data volume constructed from a three-dimensional or four-dimensional integrated model of the engineering object or an external professional knowledge base, using techniques such as retrieval enhancement generation, knowledge graph embedding or arbitrary knowledge injection, and extracting knowledge related to problems specific to the engineering field and providing it as additional background knowledge to the spatiotemporal large model base;

[0020] Spatiotemporal capability enhancement technology refers to training based on the spatiotemporal characteristics of the engineering field. Specifically, it includes designing tasks that incorporate spatiotemporal concepts, constructing datasets with precise spatiotemporal labels, training on a large-scale base model, and enhancing the spatiotemporal base model's ability to understand spatiotemporal queries, reason about spatiotemporal relationships, and predict spatiotemporal evolution patterns.

[0021] Prompt engineering technology refers to designing optimized prompt templates for engineering tasks, which include clear instructions, contextual information, constraints, and expected output formats.

[0022] Optionally, the integrated three-dimensional or four-dimensional model of the engineering object is a high-level data and knowledge set that exists in the form of a knowledge graph or digital twin, is constructed based on fused engineering spatiotemporal data, contains rich semantic relationships and business information, and supports intelligent reasoning and analysis. It is obtained through the following methods:

[0023] Based on the engineering scenario, select either relative time or absolute time as the time reference, and select either local coordinate system or geographic coordinate system as the spatial reference to define a unified spatiotemporal benchmark.

[0024] Under a unified spatiotemporal benchmark, data from different sources and in different formats are accessed, correlated, and integrated to form a consistent and analyzable data set. The data from different sources and in different formats are divided into two main categories of core data: one is the data corresponding to the three-dimensional geometry and attribute integrated model that serves as the core spatiotemporal basis, and the other is various dynamic data streams covering the entire life cycle of the project.

[0025] Based on the consistent and analyzable dataset, a spatial geometry and attribute model of the engineering project is established, using the geographical environment and physical engineering entities as the foundation. On the basis of the spatial model, the engineering object is used as the element unit, and at least one time series data, including environmental monitoring and automated operation, is integrated. Using knowledge engineering technology, the engineering entities and their interrelationships are explicitly expressed, and a three-dimensional or four-dimensional integrated model of the engineering object is constructed and dynamically maintained.

[0026] The dynamic private domain data related to the engineering business includes: monitoring and surveillance data, equipment automation data, which reflect the dynamic full life cycle of the true picture of the project.

[0027] Optionally, the integrated 3D or 4D model of the engineering object responds to traditional query requests and, according to the needs of the upper-level module, directly returns the geometric data and attribute data contained in the 3D or 4D dimensions of the engineering object; or,

[0028] The integrated three-dimensional or four-dimensional model of the engineering object responds to traditional query requests and, according to the needs of the upper-level module, extracts the geometric data or attribute data from the integrated three-dimensional or four-dimensional model of the engineering object through the understanding and generation capabilities of the spatiotemporal large model base. The extraction includes at least one of the following: object attribute query, local graphic output, and three-dimensional model sectioning.

[0029] Based on the geometric data or attribute data, new generative content is output by comprehensive analysis of the spatiotemporal large model or by calling external tools. The generative content refers to outputs that have business value, which are generated after intelligent processing and go beyond the original data or simple query results. The types are at least one of text results, two-dimensional graphics, and three-dimensional models.

[0030] Optionally, the agent collaboration module provides management, communication, and collaboration for agents; and ensures the normal operation of each functional agent through registration, discovery, and management.

[0031] Multiple functional intelligent agents interact and execute data and status with each other through the communication channel provided by the intelligent agent collaboration module;

[0032] If any functional agent encounters an anomaly during execution, the agent collaboration module provides multiple rounds of task allocation and execution and anomaly resolution based on the analysis and understanding of the spatiotemporal large model base and the autonomous decision-making and control module.

[0033] Optionally, the autonomous analysis, decision-making, and control module is also used for the autonomous analysis and judgment of engineering problems, specifically including:

[0034] The autonomous analysis, decision-making, and control module, based on the engineering problem, utilizes the spatiotemporal large model base to output relevant engineering knowledge background and a list of steps for solving the problem. It obtains relevant spatiotemporal situational data from the three-dimensional or four-dimensional integrated model of the engineering object, forming complete knowledge, methods, and data for problem analysis and solution. The spatiotemporal large model base then autonomously analyzes and judges the data and calls the intelligent agent collaboration module to execute it, outputting the decision results or control instructions for the engineering problem.

[0035] Optionally, the spatiotemporal large model base is also used to process and parse the multimodal data, which includes at least one of text, sound, image, video, vector graphics and structured data.

[0036] Optionally, the specific implementation form of the three-dimensional or four-dimensional integrated model of the engineering object is: an engineering spatiotemporal knowledge graph or an engineering digital twin, which is a platform that integrates multi-source spatiotemporal data of engineering, and organizes it into a spatiotemporal data and knowledge platform with rich semantic relationships, which is used for reasoning and understanding of large spatiotemporal models;

[0037] The engineering spatiotemporal knowledge graph uses a graph structure to explicitly express engineering entities and the complex relationships between them, and contains and represents a large amount of business data.

[0038] The engineering digital twin constructs a high-fidelity virtual model of the physical engineering entity and uses fused dynamic data to achieve real-time or near-real-time synchronization between the virtual model and the physical entity's state, supporting simulation and interaction.

[0039] Optionally, the full-element, multi-dimensional geometric and attribute data of the engineering object includes, but is not limited to: full-stratum geological model, Building Information Modeling (BIM), City Information Modeling (CIM), or factory 3D model, infrastructure 3D model.

[0040] Optionally, the various dynamic data streams covering the entire life cycle of the project include at least one of the following: personnel activity data, safety monitoring data, equipment operation status data, environmental monitoring data, construction progress data, material consumption data, cost data, or production and operation management data.

[0041] Optionally, the functional agents in the agent collaboration module include at least one of the following: a data acquisition and preprocessing agent, a domain model computation agent, a risk identification and assessment agent, a resource scheduling and optimization agent, a report and visualization generation agent, or a human-computer interaction interface agent; or,

[0042] The functional intelligent agent includes at least one combination of spatiotemporal analysis intelligent agent, process execution intelligent agent, decision optimization intelligent agent and industrial control intelligent agent.

[0043] Optionally, the intelligent decision-making results or control instructions include at least one of the following: safety risk level assessment, abnormal event early warning, performance prediction analysis, resource allocation scheme, operation optimization suggestions, emergency response plan, or adaptive control parameter set.

[0044] Optionally, the intelligent control system is also used to receive feedback information at the execution level, integrate the feedback information into the three-dimensional or four-dimensional integrated model of the engineering object for updating, and trigger adaptive adjustments to the strategy of the spatiotemporal large model base or the scheduling of the autonomous analysis, decision-making and control module.

[0045] The adaptive adjustment is used to dynamically update the spatiotemporal data volume and automatically call and execute functional intelligent agents based on the spatiotemporal large model base to generate intelligent decision results or control instructions periodically or automatically.

[0046] Optionally, the application method of the intelligent control system includes the following steps:

[0047] Step T1: Using the spatiotemporal large model base, deeply analyze the input engineering business requirements, and combine the state and spatiotemporal characteristics of the complete spatiotemporal data volume in the current three-dimensional or four-dimensional integrated model of the engineering object to automatically transform the engineering business requirements into a series of sub-tasks executed in the intelligent agent collaboration module.

[0048] Step T2: The autonomous analysis, decision-making and control module dynamically schedules the relevant functional agents in the agent collaboration module according to the sub-task and real-time status. The relevant functional agents use the communication and collaboration mechanism to exchange information and allocate tasks, and when needed, obtain the processed, generated or integrated spatiotemporal data, analysis results or generative content required to perform the task from the three-dimensional or four-dimensional integrated model of the engineering object to collaboratively execute the sub-task.

[0049] Step T3: The autonomous analysis, decision-making and control module obtains the global spatiotemporal situation information from the three-dimensional or four-dimensional integrated model of the engineering object, combines the execution results and status feedback of the aggregated functional intelligent agents, and uses the reasoning ability of the spatiotemporal large model base for the engineering field and the preset or dynamically generated rules and objectives to perform autonomous analysis and judgment, and generates intelligent decision results or control instructions based on this.

[0050] Step T4: Output the decision result or control command to the execution level, receive execution effect feedback or environmental change information, and use the execution effect feedback or environmental change information to update the three-dimensional or four-dimensional integrated model of the engineering object, and trigger adaptive adjustments to the strategy of the spatiotemporal large model base or the scheduling of the autonomous analysis decision and control module.

[0051] This invention aims to address the problems existing in the application of artificial intelligence in the engineering field, such as weak spatiotemporal correlation processing capabilities, difficulty in fusion of multi-source heterogeneous data, delayed business response, insufficient intelligent decision-making, and shallow integration of AI technology with engineering business. This invention introduces a spatiotemporal large-scale model optimized for the engineering field as the intelligent core, constructs a three-dimensional or four-dimensional integrated model of engineering objects with a multi-dimensional data model of all elements of the engineering object as its core, and achieves a closed loop of intelligent agent collaboration and autonomous decision-making. Compared with existing technologies, the intelligent management and control system provided by this invention has one or more of the following beneficial effects:

[0052] (1) Significantly enhances spatiotemporal cognition and processing capabilities: With the original multi-dimensional data model of engineering elements as the core, dynamic data is integrated and managed under a unified spatiotemporal benchmark. Combined with the understanding of spatiotemporal correlation and evolution patterns by the spatiotemporal big model, the system’s spatiotemporal perception and analysis capabilities for complex engineering environments are greatly enhanced, and it can effectively process multi-dimensional vector graphic information.

[0053] (2) Achieving efficient data fusion and intelligent services: The three-dimensional or four-dimensional integrated model of engineering objects not only solves the problem of data silos, but also responds to requests to actively process, generate and integrate new spatiotemporal data, analysis results or generative content (e.g., risk assessment results after spatiotemporal analysis, equipment health index generated based on multi-source monitoring data, daily summary automatically generated based on templates and real-time data, preliminary optimization layout suggestions or design elements, etc.), providing a powerful and intelligent new paradigm of data support for upper-level intelligent applications, which is different from the traditional data query mode.

[0054] (3) Improve business response speed and decision-making intelligence: The spatiotemporal big model can deeply understand business needs and quickly decompose and plan tasks. The autonomous decision-making module can make accurate judgments and optimize scheduling based on the global spatiotemporal situation and model reasoning, which significantly improves the business response speed and the scientific and forward-looking nature of decision-making.

[0055] (4) Promote the deep integration of AI and engineering business: The understanding and reasoning capabilities of the domain-adapted spatiotemporal model are closely combined with engineering spatiotemporal data and specific business processes (executed by intelligent agents), forming a closed loop from demand to execution to feedback, so that AI technology can be truly implemented and empower the core engineering business.

[0056] (5) Achieve adaptive and continuous optimization management: Through a closed-loop feedback mechanism, the system can continuously adjust strategies and optimize models based on actual execution results and environmental changes, thereby achieving adaptive and intelligent management of the entire life cycle of the project.

[0057] (6) Wide range of applications: The system and method can be flexibly applied to multiple engineering fields such as smart mines, smart cities, smart factories, smart transportation, smart water conservancy, smart energy, and smart environment, and have good versatility and scalability.

[0058] The intelligent control system proposed in this invention, based on a large-scale model and integrated multidimensional spatiotemporal data fusion, constructs a spatiotemporal large-scale model foundation optimized for the engineering field. It establishes a multidimensional spatiotemporal data platform with multidimensional geometric and attribute models at its core. This platform provides intelligent data services and information products, and through intelligent agent collaborative task execution and autonomous decision-making modules, it performs command, dispatch, and closed-loop control. This effectively solves many pain points in the application of AI large-scale models in the engineering field, significantly improving the level of intelligent control throughout the entire lifecycle of engineering projects. This invention can solve intelligent control problems in engineering fields such as mining, chemical plants, water conservancy projects, and building construction, significantly improving the practicality of artificial intelligence systems in the engineering field and possessing high applicability. Attached Figure Description

[0059] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0060] Figure 1 This is an architecture diagram of an intelligent control system based on a large model and integrated multidimensional spatiotemporal data fusion proposed in an embodiment of this application;

[0061] Figure 2This is an application method of an intelligent control system based on large model and integrated multidimensional spatiotemporal data fusion in the embodiments of this application. Detailed Implementation

[0062] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention, and are only some, not all, embodiments of the present invention, and are not intended to limit the present invention.

[0063] The present invention proposes an intelligent control system based on a large model and integrated multidimensional spatiotemporal data fusion, referring to... Figure 1 The structural diagram shown includes: a spatiotemporal large model base, a three-dimensional or four-dimensional integrated model of the engineering object, an intelligent agent collaboration module, and an autonomous decision-making and control module.

[0064] The spatiotemporal large model base is used to receive engineering-related input information and, based on its deep understanding, knowledge association, and logical reasoning of the data provided by the engineering domain professional knowledge, engineering-related input information, and the three-dimensional or four-dimensional integrated model of the engineering object, perform intelligent task planning and arrangement. The input information includes at least business requirements and multimodal data.

[0065] The "spatiotemporal large-scale model base" in this invention refers to a large-scale artificial intelligence model with powerful general capabilities, adapted and optimized for the engineering field. Here, the "spatiotemporal large-scale model base" does not refer to an unmodified general-purpose large model, but rather to its ability to deeply understand and handle problems specific to the engineering field. This capability is acquired through an adaptation process, which is one of the key features distinguishing this invention from simply calling general-purpose large-scale model APIs. This adaptation aims to enable it to deeply understand the professional knowledge, terminology, and standards of the engineering industry's public domain, as well as the business data, business processes, and the complex spatiotemporal relationships inherent in the private domain of engineering.

[0066] The adaptation method may include, but is not limited to, a combination of one or more of the following technologies:

[0067] Domain-specific fine-tuning: This is one of the key adaptation methods. It utilizes various public domain data, including large-scale text corpora from the engineering domain (such as industry reports, design specifications, construction plans, safety regulations, maintenance manuals, project management documents, etc.), code corpora (such as engineering calculation scripts, simulation model code), and multimodal data (detailed below), for supervised fine-tuning, instruction-based fine-tuning, or continuous pre-training. This enables the spatiotemporal large-scale model base to more accurately understand engineering language, follow engineering instructions, and generate text or code that conforms to engineering specifications. For example, a large number of mine safety accident reports can be used to fine-tune the model, enabling it to better understand mine risk-related terminology and causal relationships.

[0068] Knowledge Enhancement (DE) technology: another important adaptation method. It combines spatiotemporal data volumes (such as knowledge graphs) constructed from integrated 3D or 4D models of engineering objects, or external professional knowledge bases (such as materials databases or equipment knowledge bases). It employs Retrieval Enhancement Generation (RAG), knowledge graph embedding, or other knowledge injection techniques, and extracts knowledge relevant to engineering-specific problems, providing it as supplementary background knowledge to the spatiotemporal large-scale model base. During reasoning within the spatiotemporal large-scale model base, relevant structured engineering knowledge is dynamically retrieved and injected to improve the accuracy, professionalism, and factual consistency of the spatiotemporal large-scale model base's answers, reducing "illusions." For example, when the spatiotemporal large-scale model base answers a question about a certain material property, accurate data can be retrieved from a materials database and integrated into the answer.

[0069] Spatiotemporal capability enhancement technology: This technology targets the strong spatiotemporal characteristics of engineering fields through training. By designing tasks incorporating spatiotemporal concepts (such as understanding topological relationships, time series prediction, and spatial pattern recognition), and training with data bearing precise spatiotemporal labels, the technology enhances the spatiotemporal large-scale model's ability to understand spatiotemporal queries, reason about spatiotemporal relationships, and predict spatiotemporal evolution patterns. For example, training the spatiotemporal large-scale model to understand a spatiotemporal query such as "find all microseismic events that occurred within a 500-meter radius of point A in the past 24 hours."

[0070] Prompt Engineering: Designing optimized prompt templates for engineering tasks, containing clear instructions, contextual information, constraints, and expected output formats to better guide the adapted spatiotemporal large model base to complete the task.

[0071] For the aforementioned multimodal data: the spatiotemporal large model foundation is used to process multimodal data commonly found in the engineering field. Here, "multimodal data" refers to data types encompassing various information formats; its specific content varies depending on the engineering application scenario, but typically includes at least the following core types:

[0072] (1) Text: Various engineering documents, reports, records, standards and specifications, etc.

[0073] (2) Images and vector graphics: Visual representations of engineering drawings (GIS, CAD, scanned copies), BIM / CIM models, site photos, satellite / UAV remote sensing images, and screenshots from surveillance videos. The model needs to be able to understand graphic symbols, recognize image content, and associate graphic and textual information.

[0074] (3) Structured data: time series data collected by sensors, monitoring reports, equipment operation parameter tables, bill of materials, cost data, project schedules, etc.

[0075] In some preferred embodiments, the spatiotemporal large model base can also be configured to process other unstructured data, such as:

[0076] (1) Sound: such as the operating sound of key equipment, used for status monitoring or abnormality diagnosis (such as bearing noise, rock fracture sound) through sound pattern recognition.

[0077] (2) Video: such as continuous monitoring video of the construction site, used for personnel behavior recognition, safety violation detection, automatic progress tracking, equipment action analysis, etc.

[0078] The spatiotemporal large model base needs to have the ability to fuse and process these different modal information. For example, it should be able to understand a text describing a device malfunction and associate it with the corresponding device operation sound clips or monitoring video footage.

[0079] Core Function Summary: Receives engineering business requirements; analyzes requirements using its adapted engineering understanding and spatiotemporal cognition capabilities; queries 3D or 4D integrated models of engineering objects to obtain relevant information; performs task planning; generates instructions and issues them to the intelligent agent collaboration module; understands intelligent agent feedback; and provides analytical and judgmental support for the autonomous decision-making and control module based on engineering knowledge and spatiotemporal reasoning.

[0080] An integrated 3D or 4D model of an engineering object is built and dynamically updated under a unified spatiotemporal benchmark. Based on the full-element, 3D or 4D geometric and attribute data of the engineering object, it integrates dynamic private domain data related to engineering business. Through integrated spatiotemporal expression and processing, it constructs and dynamically updates a spatiotemporal data volume that reflects the true overall picture of the project and dynamically reflects its entire lifecycle. This spatiotemporal data volume serves as the integrated 3D or 4D model of the engineering object. It responds to data requests (from the spatiotemporal large model base, intelligent agent collaboration module, or autonomous analysis, decision-making, and control module), processes, generates, or synthesizes global spatiotemporal situational information, and provides the spatiotemporal data, analysis results, or generative content required for the training of the spatiotemporal large model base, intelligent agent collaboration execution, analysis, decision-making, and control. The full-element, multi-dimensional geometric and attribute data of the engineering object includes, but is not limited to: full-stratum geological models, Building Information Modeling (BIM), City Information Modeling (CIM), or 3D factory models and 3D infrastructure models.

[0081] The "integrated three-dimensional or four-dimensional model of the engineering object" in this invention is a core data hub. It is not only a data repository but also a provider of data fusion, knowledge construction, and intelligent services. The core of the intelligent control system of this invention is based on a unified benchmark, which integrates data into a spatiotemporal data volume and provides intelligent analysis in conjunction with a large spatiotemporal model base.

[0082] Unified spatiotemporal reference: Define and implement a globally consistent spatiotemporal reference framework to provide a unified spatiotemporal "anchor" for all data, which is the foundation for data fusion and spatiotemporal analysis.

[0083] Data Access and Fusion: Responsible for accessing two main categories of core data: first, integrated 3D geometric and attribute models (such as BIM and geological models) serving as the core spatiotemporal foundation; and second, various dynamic data streams covering the entire project lifecycle (from planning and design to operation and maintenance stages) (such as monitoring, status, cost, and activity data). "Fusion" here refers to the process of associating and integrating these data from different sources and in different formats under a unified spatiotemporal benchmark, forming a consistent and analyzable data set. This includes the integrated multidimensional model of the entire mine strata, roadways, pipelines, and equipment. The various dynamic data streams covering the entire project lifecycle include at least one of the following: personnel activity data, safety monitoring data, equipment operating status data, environmental monitoring data, construction progress data, material consumption data, cost data, or production and operation management data.

[0084] Construction and Management of Integrated 3D or 4D Models of Engineering Objects: This is a crucial step in transforming fused data into knowledge. Based on the unified dataset provided by the data access and fusion layer, this layer utilizes knowledge engineering techniques (such as ontology modeling, entity recognition, relation extraction, semantic linking, and rule-based reasoning) to construct and dynamically maintain an integrated 3D or 4D model of an engineering object. The term "integrated model" here emphasizes the structured, semantic, and reasonable characteristics of its content, far exceeding those of traditional databases. A preferred specific implementation could be:

[0085] Engineering spatiotemporal knowledge graph: It uses a graph structure to explicitly express engineering entities (such as equipment, components, measurement points, events) and the complex relationships between them (such as spatial inclusion, topological connection, temporal sequence, causal influence, attribute association, etc.), and contains and represents a large amount of business data (for example, the ID, attributes, and historical status changes of a specific device as nodes and attribute values ​​in the graph).

[0086] Engineering digital twin: It constructs a high-fidelity virtual model of a physical engineering entity and uses fused dynamic data to achieve real-time or near-real-time synchronization between the virtual model and the physical entity's state, supporting simulation and interaction.

[0087] Therefore, in this invention, a "three-dimensional or four-dimensional integrated model of an engineering object" refers to a high-level data and knowledge set that exists in the form of a knowledge graph or digital twin, is constructed based on fused engineering spatiotemporal data, contains rich semantic relationships and business information, and supports intelligent reasoning and analysis. In other words, it is a platform that integrates multi-source spatiotemporal engineering data, organizes it into a spatiotemporal data and knowledge set with rich semantic relationships, and provides a basis for reasoning and understanding within a large spatiotemporal model.

[0088] Data Processing and Services (A New Paradigm of Input and Output): Providing intelligent data services. It not only responds to traditional query requests but also proactively utilizes integrated 3D or 4D models of engineering objects for data processing (such as spatiotemporal aggregation and feature extraction), information synthesis (such as weighted assessment of multi-source risk factors), or content generation (such as automated report summaries and preliminary design elements) based on the needs of upper-level modules, providing these new "generative contents." Here, "generative contents" refers to outputs with business value generated after intelligent processing, going beyond raw data or simple query results. For example, it could be: a comprehensive risk assessment result including risk level, impact range, and confidence level; a spatiotemporal heatmap predicting traffic congestion for the next hour; a list of optimized material distribution suggestions generated based on current inventory and production plans; or a preliminary draft of a compliant equipment installation flowchart containing key parameters. This model makes the data platform a crucial support for intelligent decision-making, rather than merely a passive data source.

[0089] The integrated 3D or 4D model of the engineering object responds to traditional query requests and, based on the needs of the upper-level modules, directly returns the geometric data and attribute data contained in the 3D or 4D model of the engineering object; or, the integrated 3D or 4D model of the engineering object responds to traditional query requests and, based on the needs of the upper-level modules, extracts geometric data or attribute data from the integrated 3D or 4D model of the engineering object through the understanding and generation capabilities of the spatiotemporal large model base. This extraction includes at least one of the following: object attribute query, local graphic output, and 3D model sectioning.

[0090] Based on geometric or attribute data, the spatiotemporal large model is used for comprehensive analysis or external tools are called to process and output new generative content. Generative content refers to outputs that go beyond the original data or simple query results, are generated after intelligent processing, and have business value. The type is at least one of text results, two-dimensional graphics, and three-dimensional models.

[0091] The agent collaboration module contains multiple (two or more) functional agents and provides communication and collaboration mechanisms between them. The agent collaboration module receives task instructions from the spatiotemporal large model base corresponding to intelligent task planning and orchestration. Under the scheduling and monitoring of the autonomous analysis, decision and control module, it drives the functional agents to use the communication and collaboration mechanisms to conduct information interaction and task allocation. It also obtains the spatiotemporal data, analysis results or generative content required to execute tasks from the three-dimensional or four-dimensional integrated model of the engineering object. The agents then collaboratively execute task instructions to process data, perform calculations or interact with external systems, and feed back the status information or final results during the execution process to the autonomous analysis, decision and control module or the spatiotemporal large model base.

[0092] The agent collaboration module provides agent management, communication infrastructure, and a collaborative mechanism and protocol framework. Agent management is responsible for agent registration, discovery, and lifecycle management. Communication infrastructure ensures reliable and efficient information exchange between agents. Collaboration mechanisms and protocols implement collaborative logic such as task allocation, information sharing, and conflict resolution.

[0093] The agent collaboration module provides management, communication, and coordination for agents; it ensures the normal operation of each functional agent through registration, discovery, and management; multiple functional agents interact and execute data and states with each other through the communication channel provided by the agent collaboration module; if any functional agent encounters an anomaly during execution, the agent collaboration module provides multi-round task allocation and execution and anomaly resolution based on the analysis, understanding, autonomous decision-making, and control module of the spatiotemporal large model.

[0094] Functional agent instances: These consist of at least two functional agents. They can be of general functional types (data acquisition, model computation, risk assessment, optimization, reporting, interaction, etc.) or more focused on engineering processes (spatiotemporal analysis, process execution, decision optimization, industrial control, etc.). Specifically, the functional agents in the agent collaboration module include at least one of the following: data acquisition and preprocessing agent, domain model computation agent, risk identification and assessment agent, resource scheduling and optimization agent, report and visualization generation agent, or human-computer interaction interface agent; or, functional agents include at least one combination of: spatiotemporal analysis agent, process execution agent, decision optimization agent, and industrial control agent. These functional agents are the basic units for performing specific tasks.

[0095] The autonomous analysis, decision-making, and control module is used to acquire global spatiotemporal situational information from the three-dimensional or four-dimensional integrated model of the engineering object, receive execution feedback results from the functional intelligent agent collaboration module, and utilize the reasoning capabilities of the spatiotemporal large model base for the engineering field. Combining global spatiotemporal situational information, execution feedback results, and preset or dynamically generated rules and objectives, it performs autonomous analysis and judgment. Based on the results of autonomous analysis and judgment, it autonomously schedules the intelligent agent collaboration module to execute subsequent tasks and generates intelligent decision results or control instructions.

[0096] The autonomous analysis, decision-making, and control module is responsible for high-level situational awareness, analysis, judgment, and command and dispatch. It includes: information acquisition, core decision-making logic, autonomous dispatch, and decision or control output.

[0097] Information acquisition includes: obtaining processed, generated, or integrated global spatiotemporal situational information from the three-dimensional or four-dimensional integrated model of the engineering object, as well as the execution feedback from the intelligent agent collaboration module.

[0098] Core decision-making logic (embodying autonomy): Its "autonomy" lies in leveraging the reasoning capabilities of a large spatiotemporal model for the engineering field. It inputs comprehensive information, rules, and objectives into the large model, which then autonomously analyzes and judges them, outputting decision suggestions or judgment results.

[0099] Autonomous scheduling: Based on the analysis and judgment results of the large model, this module autonomously issues scheduling instructions to the intelligent agent collaboration module 130.

[0100] Decision or control output: Generates the final intelligent decision result or control command. The intelligent decision result or control command includes: safety risk level assessment, abnormal event early warning, performance prediction analysis, resource allocation plan, operation optimization suggestion, emergency response plan, or at least one of the following adaptive control parameter sets.

[0101] The autonomous analysis, decision-making, and control module is also used for the autonomous analysis and judgment of engineering problems. Specifically, the autonomous analysis, decision-making, and control module outputs relevant engineering knowledge background and a list of steps to solve the problem based on the engineering problem using the spatiotemporal large model base. It also obtains relevant spatiotemporal situational data from the three-dimensional or four-dimensional integrated model of the engineering object, forming complete knowledge, methods, and data for problem analysis and solution. The spatiotemporal large model base then performs autonomous analysis and judgment, and calls the intelligent agent collaboration module to execute the results, outputting the decision results or control instructions for the engineering problem.

[0102] Furthermore, the intelligent control system also receives feedback information from the execution level and integrates this feedback information into a three-dimensional or four-dimensional integrated model of the engineering object. This triggers adaptive adjustments to the scheduling of the strategy or autonomous analysis, decision-making, and control modules of the spatiotemporal large model base. The adaptive adjustment is used to dynamically update the spatiotemporal data volume and automatically invoke and execute functional intelligent agents based on the spatiotemporal large model base to periodically or automatically generate intelligent decision results or control commands.

[0103] Based on the large-scale model and integrated multi-dimensional spatiotemporal data fusion intelligent control system proposed in this invention, referring to... Figure 2 The method of applying the system shown includes:

[0104] Step 201: Utilize the spatiotemporal large model base to deeply analyze the input engineering business requirements, and combine the state and spatiotemporal characteristics of the complete spatiotemporal data volume in the current three-dimensional or four-dimensional integrated model of the engineering object to automatically transform the engineering business requirements into a series of sub-tasks executed in the intelligent agent collaboration module.

[0105] Step 202: The autonomous analysis, decision-making and control module dynamically schedules relevant functional agents in the agent collaboration module according to the sub-tasks and real-time status. The relevant functional agents use communication and collaboration mechanisms to exchange information and allocate tasks, and when needed, obtain the processed, generated or integrated spatiotemporal data, analysis results or generative content required to perform the tasks from the three-dimensional or four-dimensional integrated model of the engineering object to collaboratively execute the sub-tasks.

[0106] Step 203: The autonomous analysis, decision-making and control module obtains global spatiotemporal situation information from the three-dimensional or four-dimensional integrated model of the engineering object. Combining the execution results and status feedback of the aggregated functional intelligent agents, it uses the spatiotemporal large model base to perform autonomous analysis and judgment based on the reasoning ability of the engineering field and the preset or dynamically generated rules and objectives, and generates intelligent decision results or control instructions based on this.

[0107] Step 204: Output the decision results or control instructions to the execution level, receive execution effect feedback or environmental change information, and use the execution effect feedback or environmental change information to update the three-dimensional or four-dimensional integrated model of the engineering object, and trigger adaptive adjustments to the scheduling of the strategy or autonomous analysis decision and control module of the spatiotemporal large model base.

[0108] In summary, the intelligent management and control system based on a large model and integrated multidimensional spatiotemporal data fusion of the present invention aims to solve the problems existing in the application of artificial intelligence in the engineering field, such as weak spatiotemporal correlation processing capabilities, difficulty in fusion of multi-source heterogeneous data, delayed business response, insufficient intelligent decision-making, and shallow integration of AI technology with engineering business. The present invention introduces a spatiotemporal large model optimized for the engineering field as the intelligent core, constructs a three-dimensional or four-dimensional integrated model of the engineering object with a multidimensional data model of all elements of the engineering object as its core, and achieves a closed loop of intelligent agent collaboration and autonomous decision-making. Compared with the prior art, the intelligent management and control system provided by the present invention has one or more of the following beneficial effects:

[0109] (1) Significantly enhances spatiotemporal cognition and processing capabilities: With the original multi-dimensional data model of engineering elements as the core, dynamic data is integrated and managed under a unified spatiotemporal benchmark. Combined with the understanding of spatiotemporal correlation and evolution patterns by the spatiotemporal big model, the system’s spatiotemporal perception and analysis capabilities for complex engineering environments are greatly enhanced, and it can effectively process multi-dimensional vector graphic information.

[0110] (2) Achieving efficient data fusion and intelligent services: The three-dimensional or four-dimensional integrated model of engineering objects not only solves the problem of data silos, but also responds to requests to actively process, generate and integrate new spatiotemporal data, analysis results or generative content (e.g., risk assessment results after spatiotemporal analysis, equipment health index generated based on multi-source monitoring data, daily summary automatically generated based on templates and real-time data, preliminary optimization layout suggestions or design elements, etc.), providing a powerful and intelligent new paradigm of data support for upper-level intelligent applications, which is different from the traditional data query mode.

[0111] (3) Improve business response speed and decision-making intelligence: The spatiotemporal big model can deeply understand business needs and quickly decompose and plan tasks. The autonomous decision-making module can make accurate judgments and optimize scheduling based on the global spatiotemporal situation and model reasoning, which significantly improves the business response speed and the scientific and forward-looking nature of decision-making.

[0112] (4) Promote the deep integration of AI and engineering business: The understanding and reasoning capabilities of the domain-adapted spatiotemporal model are closely combined with engineering spatiotemporal data and specific business processes (executed by intelligent agents), forming a closed loop from demand to execution to feedback, so that AI technology can be truly implemented and empower the core engineering business.

[0113] (5) Achieve adaptive and continuous optimization management: Through a closed-loop feedback mechanism, the system can continuously adjust strategies and optimize models based on actual execution results and environmental changes, thereby achieving adaptive and intelligent management of the entire life cycle of the project.

[0114] (6) Wide range of applications: The system and method can be flexibly applied to multiple engineering fields such as smart mines, smart cities, smart factories, smart transportation, smart water conservancy, smart energy, and smart environment, and have good versatility and scalability.

[0115] The intelligent control system proposed in this invention, based on a large-scale model and integrated multidimensional spatiotemporal data fusion, constructs a spatiotemporal large-scale model foundation optimized for the engineering field. It establishes a multidimensional spatiotemporal data platform with multidimensional geometric and attribute models at its core. This platform provides intelligent data services and information products, and through intelligent agent collaborative task execution and autonomous decision-making modules, it performs command, dispatch, and closed-loop control. This effectively solves many pain points in the application of AI large-scale models in the engineering field, significantly improving the level of intelligent control throughout the entire lifecycle of engineering projects. This invention can solve intelligent control problems in engineering fields such as mining, chemical plants, water conservancy projects, and building construction, significantly improving the practicality of artificial intelligence systems in the engineering field and possessing high applicability.

[0116] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0117] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0118] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. An intelligent control system based on large-scale models and integrated multi-dimensional spatiotemporal data fusion, characterized in that, include: The spatiotemporal large model base is used to receive engineering-related input information and, based on its deep understanding, knowledge association and logical reasoning of the data provided by the engineering domain professional knowledge, the engineering-related input information and the three-dimensional or four-dimensional integrated model of the engineering object, to perform intelligent task planning and arrangement. The input information includes at least business requirements and multimodal data. A three-dimensional or four-dimensional integrated model of an engineering object, under a unified spatiotemporal reference, is based on the geometric and attribute data of all elements, three-dimensional or four-dimensional of the engineering object, and integrates dynamic private domain data related to engineering business. Through integrated spatiotemporal expression and processing, a spatiotemporal data body reflecting the true overall picture and full life cycle of the project is constructed and dynamically updated. This spatiotemporal data body serves as the three-dimensional or four-dimensional integrated model of the engineering object, which is used to respond to data requests and is processed, generated or integrated to form global spatiotemporal situation information. The intelligent agent collaboration module contains multiple functional intelligent agents and provides a communication and collaboration mechanism between the functional intelligent agents. The intelligent agent collaboration module is used to receive task instructions corresponding to intelligent task planning and orchestration issued by the spatiotemporal large model base, and under the scheduling and monitoring of the autonomous analysis decision and control module, it obtains the data, analysis results or generative content required to execute the task from the three-dimensional or four-dimensional integrated model of the engineering object, and drives the functional intelligent agents to execute. The autonomous analysis, decision-making, and control module is used to obtain global spatiotemporal situation information from the three-dimensional or four-dimensional integrated model of the engineering object, receive execution feedback results from the functional intelligent agent collaboration module, and use the reasoning ability of the spatiotemporal large model base for the engineering field, combined with the global spatiotemporal situation information, the execution feedback results, and preset or dynamically generated rules and objectives, to perform autonomous analysis and judgment. Based on the results of the autonomous analysis and judgment, it autonomously schedules the intelligent agent collaboration module to execute subsequent tasks and generates intelligent decision results or control instructions. The integrated three-dimensional or four-dimensional model of the engineering object is a high-level data and knowledge set that exists in the form of a knowledge graph or digital twin, is constructed based on fused engineering spatiotemporal data, contains rich semantic relationships and business information, and supports intelligent reasoning and analysis.

2. The intelligent control system according to claim 1, characterized in that, The aforementioned spatiotemporal large model base is an intelligent model with the ability to deeply understand and process problems specific to the engineering field. This ability to deeply understand and process problems specific to the engineering field is obtained through an adaptation method. The adaptation method includes a combination of one or more of the following technologies: Domain fine-tuning technology refers to the use of large-scale text corpora, code corpora, and various public domain data of multimodal data in the engineering field for supervised fine-tuning, instruction fine-tuning, or continuous pre-training of intelligent models. Knowledge enhancement technology refers to the following: combining a spatiotemporal data volume constructed from a three-dimensional or four-dimensional integrated model of the engineering object or an external professional knowledge base, using techniques such as retrieval enhancement generation, knowledge graph embedding or arbitrary knowledge injection, and extracting knowledge related to problems specific to the engineering field and providing it as additional background knowledge to the spatiotemporal large model base; Spatiotemporal capability enhancement technology refers to training based on the spatiotemporal characteristics of the engineering field. Specifically, it includes designing tasks that incorporate spatiotemporal concepts, constructing datasets with precise spatiotemporal labels, training on a large-scale base model, and enhancing the spatiotemporal base model's ability to understand spatiotemporal queries, reason about spatiotemporal relationships, and predict spatiotemporal evolution patterns. Prompt engineering technology refers to designing optimized prompt templates for engineering tasks, which include clear instructions, contextual information, constraints, and expected output formats.

3. The intelligent control system according to claim 1, characterized in that, The integrated three-dimensional or four-dimensional model of the engineering object is a high-level data and knowledge set that exists in the form of a knowledge graph or digital twin, is constructed based on fused engineering spatiotemporal data, contains rich semantic relationships and business information, and supports intelligent reasoning and analysis. It is obtained through the following methods: Based on the engineering scenario, select either relative time or absolute time as the time reference, and select either local coordinate system or geographic coordinate system as the spatial reference to define a unified spatiotemporal benchmark. Under a unified spatiotemporal benchmark, data from different sources and in different formats are accessed, correlated, and integrated to form a consistent and analyzable data set. The data from different sources and in different formats are divided into two main categories of core data: one is the data corresponding to the three-dimensional geometry and attribute integrated model that serves as the core spatiotemporal basis, and the other is various dynamic data streams covering the entire life cycle of the project. Based on the consistent and analyzable dataset, a spatial geometry and attribute model of the engineering project is established, using the geographical environment and physical engineering entities as the foundation. On the basis of the spatial model, the engineering object is used as the element unit, and at least one time series data, including environmental monitoring and automated operation, is integrated. Using knowledge engineering technology, the engineering entities and their interrelationships are explicitly expressed, and a three-dimensional or four-dimensional integrated model of the engineering object is constructed and dynamically maintained. The dynamic private domain data related to the engineering business includes: monitoring and surveillance data, equipment automation data, which reflect the dynamic full life cycle of the true picture of the project.

4. The intelligent control system according to claim 3, characterized in that, The integrated 3D or 4D model of the engineering object responds to traditional query requests and, based on the needs of upper-level modules, directly returns the geometric and attribute data contained in the 3D or 4D dimensions of the engineering object; or, The integrated three-dimensional or four-dimensional model of the engineering object responds to traditional query requests and, according to the needs of the upper-level module, extracts the geometric data or attribute data from the integrated three-dimensional or four-dimensional model of the engineering object through the understanding and generation capabilities of the spatiotemporal large model base. The extraction includes at least one of the following: object attribute query, local graphic output, and three-dimensional model sectioning. Based on the geometric data or attribute data, new generative content is output by comprehensive analysis of the spatiotemporal large model or by calling external tools. The generative content refers to outputs that have business value, which are generated after intelligent processing and go beyond the original data or simple query results. The types are at least one of text results, two-dimensional graphics, and three-dimensional models.

5. The intelligent control system according to claim 1, characterized in that, The agent collaboration module provides management, communication, and collaboration for agents; it ensures the normal operation of each functional agent through registration, discovery, and management. Multiple functional intelligent agents interact and execute data and status with each other through the communication channel provided by the intelligent agent collaboration module; If any functional agent encounters an anomaly during execution, the agent collaboration module provides multiple rounds of task allocation and execution and anomaly resolution based on the analysis and understanding of the spatiotemporal large model base and the autonomous analysis, decision-making and control module.

6. The intelligent control system according to claim 1, characterized in that, The autonomous analysis, decision-making, and control module is also used for the autonomous analysis and judgment of engineering problems, specifically including: The autonomous analysis, decision-making, and control module, based on the engineering problem, utilizes the spatiotemporal large model base to output relevant engineering knowledge background and a list of steps for solving the problem. It obtains relevant spatiotemporal situational data from the three-dimensional or four-dimensional integrated model of the engineering object, forming complete knowledge, methods, and data for problem analysis and solution. The spatiotemporal large model base then autonomously analyzes and judges the data and calls the intelligent agent collaboration module to execute it, outputting the decision results or control instructions for the engineering problem.

7. The intelligent control system according to claim 1, characterized in that, The spatiotemporal large model base is also used to process and parse the multimodal data, which includes at least one of text, sound, image, video, vector graphics and structured data.

8. The intelligent control system according to claim 3, characterized in that, The specific implementation form of the three-dimensional or four-dimensional integrated model of the engineering object is: an engineering spatiotemporal knowledge graph or an engineering digital twin, which is a platform that integrates multi-source spatiotemporal data of engineering, and organizes it into a spatiotemporal data and knowledge platform with rich semantic relationships, which is used for reasoning and understanding of large spatiotemporal models; The engineering spatiotemporal knowledge graph uses a graph structure to explicitly express engineering entities and the complex relationships between them, and contains and represents a large amount of business data. The engineering digital twin constructs a high-fidelity virtual model of the physical engineering entity and uses fused dynamic data to achieve real-time or near-real-time synchronization between the virtual model and the physical entity's state, supporting simulation and interaction.

9. The intelligent control system according to claim 8, characterized in that, The engineering object's full-element, multi-dimensional geometric and attribute data includes, but is not limited to: full-stratum geological model, Building Information Modeling (BIM), City Information Modeling (CIM), or factory 3D model, infrastructure 3D model.

10. The intelligent control system according to claim 3, characterized in that, The various dynamic data streams covering the entire lifecycle of the project include at least one of the following: personnel activity data, safety monitoring data, equipment operation status data, environmental monitoring data, construction progress data, material consumption data, cost data, or production and operation management data.

11. The intelligent control system according to claim 1, characterized in that, The functional agents in the agent collaboration module include at least one of the following: a data acquisition and preprocessing agent, a domain model calculation agent, a risk identification and assessment agent, a resource scheduling and optimization agent, a report and visualization generation agent, or a human-computer interaction interface agent; or... The functional intelligent agent includes at least one combination of spatiotemporal analysis intelligent agent, process execution intelligent agent, decision optimization intelligent agent and industrial control intelligent agent.

12. The intelligent control system according to claim 11, characterized in that, The intelligent decision-making results or control instructions include at least one of the following: safety risk level assessment, abnormal event early warning, performance prediction analysis, resource allocation scheme, operation optimization suggestions, emergency response plan, or adaptive control parameter set.

13. The intelligent control system according to claim 12, characterized in that, The intelligent control system is also used to receive feedback information from the execution level, integrate the feedback information into the three-dimensional or four-dimensional integrated model of the engineering object for updating, and trigger adaptive adjustments to the strategy of the spatiotemporal large model base or the scheduling of the autonomous analysis, decision and control module. The adaptive adjustment is used to dynamically update the spatiotemporal data volume and automatically call and execute functional intelligent agents based on the spatiotemporal large model base to generate intelligent decision results or control instructions periodically or automatically.

14. The intelligent control system according to claim 13, characterized in that, The application method of the intelligent control system includes the following steps: Step T1: Using the spatiotemporal large model base, deeply analyze the input engineering business requirements, and combine the state and spatiotemporal characteristics of the complete spatiotemporal data volume in the current three-dimensional or four-dimensional integrated model of the engineering object to automatically transform the engineering business requirements into a series of sub-tasks executed in the intelligent agent collaboration module. Step T2: The autonomous analysis, decision-making and control module dynamically schedules the relevant functional agents in the agent collaboration module according to the sub-task and real-time status. The relevant functional agents use the communication and collaboration mechanism to exchange information and allocate tasks, and when needed, obtain the processed, generated or integrated spatiotemporal data, analysis results or generative content required to perform the task from the three-dimensional or four-dimensional integrated model of the engineering object to collaboratively execute the sub-task. Step T3: The autonomous analysis, decision-making and control module obtains the global spatiotemporal situation information from the three-dimensional or four-dimensional integrated model of the engineering object, combines the execution results and status feedback of the aggregated functional intelligent agents, and uses the reasoning ability of the spatiotemporal large model base for the engineering field and the preset or dynamically generated rules and objectives to perform autonomous analysis and judgment, and generates intelligent decision results or control instructions based on this. Step T4: Output the decision result or control command to the execution level, receive execution effect feedback or environmental change information, and use the execution effect feedback or environmental change information to update the three-dimensional or four-dimensional integrated model of the engineering object, and trigger adaptive adjustments to the strategy of the spatiotemporal large model base or the scheduling of the autonomous analysis decision and control module.

Images