A data integration intelligent analysis system and method

Abstract

The application provides a data integrated intelligent analysis system and method, and belongs to the technical field of data intelligent analysis. The system comprises a data acquisition module which standardizes and acquires and accesses multi-source heterogeneous data of a target scene, carries out data cleaning and formatting processing, and generates standardized data events; an intelligent agent construction module which establishes a dynamically updated intelligent agent capability portrait for each analysis intelligent agent; a task publishing module which uses a preset distributed task parser to analyze a target macro task into a fine-grained sub task; an autonomous solving module which dynamically forms a dynamic temporary task chain network for an analysis task, and carries out analysis intelligent agent emergent solving based on group interaction; and a deep solving module which carries out emergent solving difficulty perception and external aid request generation, and constructs a mixed intelligent network integrating internal and external to solve the emergent solving difficulty. The application greatly improves the robustness of the system, improves the adaptive flexibility of the system to complex scenes, and significantly improves the solving depth and accuracy of complex problems.

Description

Technical Field

[0001] This invention relates to the field of data intelligent analysis technology, and specifically to an integrated data intelligent analysis system and method. Background Technology

[0002] In modern industry and people's lives, where big data and artificial intelligence are deeply integrated, integrated data intelligent analysis systems have become a core infrastructure for various industries to mine the core value of data and support scientific decision-making. The core function of such systems is to break down the format barriers of multi-source heterogeneous data, enabling unified access, integration, and intelligent analysis of structured, semi-structured, and unstructured data. This transforms dispersed data sources into insightful analytical results, and they are widely used in complex data analysis scenarios such as the Industrial Internet of Things and financial risk control, providing data support for efficient operation, risk prevention, and optimization across various fields.

[0003] In existing technologies, integrated data intelligent analysis systems still suffer from the following shortcomings in practical applications: poor system scalability and high adaptation costs. Development cycles are long and technically challenging, resulting in high adaptation and maintenance costs, and an inability to flexibly respond to diverse analytical needs across industries; insufficient system robustness and a high risk of single-point failures. If the central server experiences hardware failure, software crashes, or network outages, the entire analysis system will be paralyzed, unable to continuously provide analytical services; weak integration capabilities of external professional resources and limited solution depth. Existing systems mostly use a static binding model for connecting to heterogeneous external knowledge resources, failing to flexibly and on-demand invoke external professional resources according to the real-time needs of specific analytical tasks; lack of flexibility and self-evolution capabilities, insufficient adaptability, and a lack of self-organizing and self-cooperative collective intelligence characteristics. They cannot generate global intelligent analysis results through local interactions between individuals, resulting in insufficient flexibility when dealing with complex, variable, and highly uncertain practical analytical tasks. Summary of the Invention

[0004] The purpose of this invention is to provide an integrated intelligent data analysis system and method to solve the technical problems of poor scalability, high adaptation cost, insufficient robustness, high risk of single point of failure, limited solution depth, and lack of flexibility and self-evolution ability in existing intelligent data analysis systems.

[0005] To achieve the above objectives, this invention provides an integrated data intelligent analysis system, comprising: a data acquisition module, used to collect and access multi-source heterogeneous data from a target scenario in a standardized manner through a unified data access gateway, perform data cleaning and formatting processing, and generate standardized data events; an intelligent agent construction module, used to build a distributed cluster composed of lightweight analytical agents based on the standardized data events, establish a dynamically updated intelligent agent capability profile for each analytical agent, and inject a unified global collaboration protocol into all analytical agents; and a task publishing module, used to publish the target macroscopic data through various analytical task triggering modes and a preset distributed task parser. The task is parsed into fine-grained subtasks, and a decentralized publishing mechanism is adopted to publish the subtasks to the system's distributed task bulletin board. The autonomous solution module is used to bid for and accept analysis tasks for analysis agents based on the global collaboration protocol and the subtasks on the distributed task bulletin board, and dynamically form a dynamic temporary task chain network for analysis tasks to perform emergent solution based on group interaction. The deep solution module is used to use a preset dilemma judgment rule system to perceive emergent solution dilemmas and generate external assistance requests, match external knowledge brains through an external resource catalog, and build a hybrid intelligent network that integrates internal and external resources to solve emergent solution dilemmas and achieve deep solution of analysis tasks.

[0006] Optionally, the data cleaning and formatting process includes: performing a two-layer identification operation on multi-source heterogeneous data to generate data identification tags containing multi-dimensional information; matching the data identification tags with the corresponding data types using a preset dedicated data cleaning and formatting module; performing customized and refined processing operations on different types of data and generating processing logs to achieve full traceability of the data processing process.

[0007] Optionally, the step of establishing a dynamically updated agent capability profile for each analytical agent includes: maintaining a corresponding dynamic agent capability profile for each analytical agent through an agent registration center, wherein the dynamic agent capability profile is divided into a basic layer and a dynamic layer; the basic layer records the static core information of the analytical agent, and the dynamic layer records the dynamic performance indicators and resource consumption patterns exhibited by the analytical agent when processing historical analytical tasks; and setting a profile update triggering mechanism, wherein a small update is triggered after the analytical agent completes a single sub-task, and a full update is triggered after a preset number of sub-tasks are completed.

[0008] Optionally, the step of using a preset distributed task parser to parse the target macro task into fine-grained subtasks includes: deploying the preset distributed task parser across multiple nodes, performing parallel decomposition using a domain knowledge graph and task decomposition algorithms; the subtask decomposition follows the core principle of meta-capability matching, with each subtask requiring only one or a few specific meta-capabilities to complete independently, ensuring that the analysis agent matches and executes accordingly; clearly marking the logical relationships for each subtask and recording the correspondence between the subtask and the target macro task; generating a standardized subtask list after decomposition, assigning corresponding subtask core information to each subtask, binding it with the task number of the target macro task, and storing it in a task information database.

[0009] Optionally, the process of bidding and accepting analysis tasks for analytical agents, and dynamically forming a dynamic temporary task chain network for the analysis tasks, includes: after an analytical agent listens to a subtask on a distributed task bulletin board, it matches the subtask's requirements with its agent capability profile; within the response time threshold specified in the global collaboration protocol, the analytical agent submits bidding information to the system in a standardized encrypted format; a decentralized bidding negotiation mechanism is used to confirm the acceptance of the bidding information, without the participation of a central node, and all bidding analytical agents are comprehensively scored according to a preset scoring algorithm; the system sends an encrypted acceptance confirmation notification to the analytical agent ranked first in the score, and the accepted analytical agent locks the subtask; the accepted analytical agent generates standardized collaboration invitation information according to the provisions of the global collaboration protocol; after receiving the collaboration invitation information, other analytical agents decide whether to respond based on their own agent capability profile and current load status; the collaboration invitation initiator filters all responding analytical agents according to the rules of the global collaboration protocol, and through multiple rounds of collaboration invitations and responses, the system spontaneously forms a dynamic temporary task chain network for the analysis tasks.

[0010] Optionally, the emergence solution based on group interaction of analytical agents includes: setting data and intermediate result flow rules for the dynamic temporary task chain network, and standardizing data events and intermediate analysis results flow in a directed manner along a preset encrypted collaborative link; each analytical agent, as a node, performs specialized processing only on the received standardized data events or intermediate analysis results within its own meta-capabilities, and generates new intermediate analysis results after specialized processing; setting the criteria for emergence convergence, and determining that the emergence of the dynamic temporary task chain network has reached a convergence state when the intermediate analysis results in the dynamic temporary task chain network reach the preset conditions or when all sub-tasks have been processed; integrating the analysis results generated by each analytical agent in the converged dynamic temporary task chain network to generate one or more preliminary candidate solution fragments for the target macro-task.

[0011] Optionally, the matching of external knowledge entities through the external resource catalog includes: maintaining a dynamically updated external resource catalog, which registers all interoperable external knowledge entity information; updating the external resource catalog by the corresponding catalog maintenance module, and performing availability checks on all external knowledge entities within a preset period; using a preset multi-dimensional weighted scoring algorithm, the system's external resource docking module matches external assistance requests with the external resource catalog; the system conducts secure service negotiations with candidate external knowledge entities, and after reaching an agreement, the system signs a standardized service collaboration agreement with the external knowledge entity.

[0012] Optionally, the construction of an internally and externally integrated hybrid intelligent network to solve the emergent problem and achieve in-depth solution of the analysis task includes: mapping successfully negotiated external knowledge entities as invited intelligent agents, and assigning them temporary cluster identifiers through an intelligent agent registration center; the invited intelligent agents using external knowledge entities as carriers dynamically linking to an internal dynamic temporary task chain network through encrypted channels to form an internally and externally integrated hybrid intelligent network; the internal analysis intelligent agent encrypting relevant standardized data events and preliminary candidate solution fragments, and transmitting them to the invited intelligent agents through a dedicated encrypted link; the invited intelligent agents analyzing the received information and encapsulating the processing results into a standardized data event format; and the internal analysis intelligent agent combining the processing results of the invited intelligent agents to refine the preliminary candidate solution fragments, thereby achieving in-depth solution of the analysis task.

[0013] Optionally, the integrated data intelligent analysis system further includes: a feedback module, used to integrate the emergent solution and the deep solution into an analysis report, cross-validate the analysis report in multiple dimensions, deliver the final analysis report and collect multi-dimensional feedback, assign multi-dimensional credit to the analysis agent and the external knowledge brain, and realize the adaptive evolution of the system's operating rules.

[0014] On the other hand, this invention provides a data-integrated intelligent analysis method, which includes: standardizing the collection and access of multi-source heterogeneous data from a target scenario through a unified data access gateway, performing data cleaning and formatting processing to generate standardized data events; based on the standardized data events, constructing a distributed cluster composed of lightweight analytical agents, establishing a dynamically updated agent capability profile for each analytical agent, and injecting a unified global collaboration protocol into all analytical agents; through multiple analytical task triggering modes, using a preset distributed task parser, parsing the target macro-task into fine-grained sub-tasks, and using a decentralized publishing mechanism to publish the sub-tasks to the system's distributed task bulletin board; and based on the global collaboration protocol... The system proposes a method for analyzing intelligent agents to bid for and accept analysis tasks from subtasks on a distributed task bulletin board. This process dynamically forms a temporary task chain network for the analysis tasks, enabling emergent problem-solving based on group interaction. A pre-defined dilemma judgment rule system is used to perceive emergent problem-solving dilemmas and generate external assistance requests. External knowledge entities are matched through an external resource catalog, and a hybrid intelligent network integrating internal and external resources is constructed to resolve emergent problem-solving dilemmas, achieving in-depth problem-solving for the analysis tasks. The emergent and in-depth problem-solving solutions are integrated into an analysis report, which undergoes multi-dimensional cross-validation. The final analysis report is delivered, and multi-dimensional feedback is collected. Multi-dimensional credit is assigned to the analysis agents and external knowledge entities, enabling adaptive evolution of system operating rules.

[0015] Through the above technical solutions, this invention employs a distributed analytical agent cluster to construct the core execution unit of the system, effectively avoiding system paralysis caused by single-point failures, significantly improving system robustness, enabling flexible expansion of analytical agents, reducing adaptation costs, and significantly improving system scalability, allowing for flexible responses to diverse analytical needs across various industries. By forming a dynamic temporary task chain network and interacting locally with agents to achieve emergent solutions to complex problems, it can flexibly handle complex and ever-changing analytical tasks across multiple domains and types, enhancing the system's adaptability to complex scenarios. By pre-setting dilemma judgment rules to perceive solution bottlenecks and match them with external knowledge entities, it breaks down the internal capability boundaries of the system, achieving the organic integration of internal and external resources, and significantly improving the depth and accuracy of solving complex problems.

[0016] Other features and advantages of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1This is a flowchart illustrating an integrated intelligent data analysis system according to the present invention. Figure 2 This is a flowchart illustrating an integrated intelligent data analysis method according to the present invention. Figure 1 ; Figure 3 This is a schematic diagram of the data cleaning and formatting process in this invention; Figure 4 This is a flowchart illustrating the process of establishing a dynamically updated capability profile for each analytical agent in this invention. Figure 5 This is a flowchart illustrating how the target macro-task is broken down into fine-grained sub-tasks in this invention. Figure 6 This is a flowchart illustrating the process of dynamically forming a dynamic temporary task chain network for analysis tasks in this invention. Figure 7 This is a schematic diagram of the process for solving the emergent problem of the analytical agent based on group interaction in this invention; Figure 8 This is a schematic diagram of the process of matching external knowledge brains through an external resource directory in this invention; Figure 9 This is a schematic diagram of the process for achieving deep solution of the analysis task in this invention; Figure 10 This is a flowchart illustrating an integrated intelligent data analysis method according to the present invention. Figure 2 . Detailed Implementation

[0018] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0019] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0020] Please refer to Figure 1This invention provides an integrated data intelligent analysis system, which may include: a data acquisition module, used to collect and access multi-source heterogeneous data of a target scenario in a standardized manner through a unified data access gateway, perform data cleaning and formatting processing, and generate standardized data events; an intelligent agent construction module, used to build a distributed cluster composed of lightweight analytical agents based on the standardized data events, and to establish a dynamically updated intelligent agent capability profile for each analytical agent, injecting a unified global collaboration protocol into all analytical agents; a task publishing module, used to parse the target macro-task into fine-grained sub-tasks through a preset distributed task parser using various analytical task triggering modes, and to publish the sub-tasks to the system's distributed task bulletin board using a decentralized publishing mechanism; and an autonomous solving module. The system is divided into three modules: a first, a deep solution module, and a third, a feedback module. The first module is used to bid for and accept analysis tasks for analytical agents based on the global collaboration protocol and sub-tasks on the distributed task bulletin board. It dynamically forms a temporary task chain network for analysis tasks and performs emergent solutions based on group interaction. The second module is used to perceive emergent solution dilemmas and generate external assistance requests using a preset dilemma judgment rule system. It matches external knowledge entities through an external resource catalog and constructs a hybrid intelligent network that integrates internal and external resources to solve emergent solution dilemmas and achieve deep solutions for analysis tasks. The third module integrates the emergent solution and the deep solution into an analysis report, performs multi-dimensional cross-validation of the analysis report, delivers the final analysis report and collects multi-dimensional feedback, assigns multi-dimensional credits to analytical agents and external knowledge entities, and achieves adaptive evolution of system operation rules.

[0021] In this embodiment of the invention, the data acquisition module can be used to collect and access multi-source heterogeneous data of the target scene in a standardized manner through a unified data access gateway, perform data cleaning and formatting processing, and generate standardized data events.

[0022] In a preferred embodiment of the present invention, a modular and loosely coupled architecture can be used to construct a unified data access gateway, which is set as the sole entry point for all raw data to enter the system. Based on various data protocol adapters of the unified data access gateway, full coverage access to all types of raw data sources can be achieved, including structured databases, distributed message queues, third-party open API interfaces, local and cloud file systems, IoT terminal sensors, etc. Data sources for batch processing, such as static ledgers and historical files, are accessed via batch file access, supporting both scheduled and manual triggering. A unique data source code is assigned to each data source, and all raw data is tagged with this data source code during the access process.

[0023] Please refer to Figure 3In a preferred embodiment of the present invention, data cleaning and formatting may include: Step S110: Perform a two-layer identification operation on the multi-source heterogeneous data (for example, the first layer identifies the data source encoding and matches the corresponding data source type, and the second layer identifies the format, structure and features of the data itself) to generate data identification labels containing multi-dimensional information.

[0024] Step S120: Based on the data identification label, use the preset dedicated data cleaning and formatting module to match it with the corresponding data type.

[0025] Step S130: Perform customized and refined processing operations on different types of data and generate processing logs to achieve full traceability of the data processing process.

[0026] In a preferred embodiment of the present invention, structured data completes intelligent filling of missing values ​​(e.g., interpolation based on domain rules or similar data), identification and filtering of outliers (e.g., based on the 3σ principle or domain thresholds), and normalization of field names and formats; unstructured text completes word segmentation, stop word removal, entity recognition, and semantic vectorization based on pre-trained models; image / video data completes keyframe extraction, feature point detection, and feature extraction and dimensionality reduction based on deep learning models; time-series data completes global timestamp alignment and time granularity unification; and geospatial data completes coordinate system unification and geolocation coding standardization.

[0027] In a preferred embodiment of the present invention, when generating standardized data events, the system can follow the preset unified encapsulation rules to encapsulate the standardized data after cleaning and formatting into standardized data events. Each standardized data event includes three core elements: data body, metadata, and initial simple label. A unique identifier code is assigned to each standardized data event, and the code adopts a fixed structure of "data source code - timestamp - random sequence".

[0028] In this embodiment of the invention, the agent building module can be used to build a distributed cluster composed of lightweight analytical agents based on standardized data events, and to establish a dynamically updated agent capability profile for each analytical agent, and to inject a unified global collaboration protocol into all analytical agents.

[0029] In a preferred embodiment of the present invention, the registration trigger condition for the analytical agent is the agent's first startup in the system or a restart after a function upgrade. Upon triggering, the analytical agent will proactively initiate a standardized registration request to the system's preset lightweight agent registration center. The registration request is transmitted in an encrypted format. The agent registration center adopts a stateless distributed architecture design, with multiple nodes deployed and data synchronized in real time between nodes, avoiding single points of failure and ensuring high availability and high concurrency processing capabilities for the analytical agent registration service. The registration information of the analytical agent is centered on meta-capability descriptions, which must strictly adhere to the system's preset standardized vocabulary set for definition. The agent clearly and accurately declares its specialized analytical capabilities; the agent registration center performs a legality verification on the registration information of the analytical agent, including the standardization of the meta-capability description, the completeness of the basic attributes, and the reachability of the network address. If the verification passes, the registration is completed; if the verification fails, the agent is informed of the specific reason for the failure and the registration is rejected; after registration, the agent registration center assigns a unique cluster identifier to the analytical agent. The cluster identifier adopts a fixed encoding structure and includes deployment node information and agent type information. At the same time, the registration information and cluster identifier are stored in a distributed registration information database to support fast query and retrieval.

[0030] Please refer to Figure 4 In a preferred embodiment of the present invention, establishing a dynamically updated agent capability profile for each analytical agent may include: Step S210: Through the agent registration center, maintain a corresponding agent dynamic capability profile for each analysis agent. The agent dynamic capability profile is divided into a basic layer and a dynamic layer.

[0031] Step S220: The base layer records the static core information of the analytical agent, and the dynamic layer records the dynamic performance indicators and resource consumption patterns exhibited by the analytical agent when processing historical analysis tasks.

[0032] In a preferred embodiment of the present invention, the static core information includes the registered meta-capability type, meta-capability boundary, applicable data type, and supported standardized data event format. This layer of information is only updated when the analysis agent's functions are upgraded and re-registered. The dynamic performance indicators include the accuracy, recall, F1 score, task completion rate, and response latency of the analysis results. The resource consumption mode includes computing power utilization, memory consumption, disk I / O, and network I / O. This layer of information is updated in real time as the analysis agent works.

[0033] Step S230: Set up a profile update trigger mechanism. After the analysis agent completes a single subtask, a small update is triggered. After a preset number of subtasks are completed, a full update is triggered.

[0034] In a preferred embodiment of the present invention, a minor update includes updating only the performance indicators and resource consumption data corresponding to the task, while a full update includes statistically analyzing, calculating and optimizing all indicators in the dynamic layer, and generating a profile update report. All data in the agent capability profile are bound to the cluster identifier of the analysis agent. The agent registration center provides a standardized profile query interface, and downstream modules can quickly query the agent capability profile of the analysis agent through the cluster identifier.

[0035] In a preferred embodiment of the present invention, the global collaboration protocol adopts a standardized and modular design, serving as a unified code of conduct for collaborative work among all analytical agents. No central controller intervenes throughout the process, and all collaborative actions are autonomously executed by the analytical agents according to the global collaboration protocol. Protocol injection is an automatic operation performed by the system. After an analytical agent completes registration and obtains a cluster identifier, the system immediately injects the global collaboration protocol into the local protocol storage module of the analytical agent via an encrypted channel. The protocol storage module employs a local persistent design to prevent protocol loss.

[0036] In this embodiment of the invention, the task publishing module can be used to analyze various task triggering modes, use a preset distributed task parser to parse the target macro task into fine-grained sub-tasks, and use a decentralized publishing mechanism to publish the sub-tasks to the system's distributed task bulletin board.

[0037] In a preferred embodiment of the present invention, the system supports two analysis task triggering modes: user-initiated triggering and system-automatic triggering. The two modes work together to meet the full-scenario analysis requirements of both proactive needs and passive monitoring. User-initiated triggering is achieved through the system's human-computer interaction interface. Users can submit abstract analysis requirements through a visual interface, and customize basic information such as analysis objectives, result output format, task priority, and completion time requirements. After receiving the information, the system automatically completes the initial information entry of the task and generates a task draft. System-automatic triggering relies on a preset multi-scenario monitor. The multi-scenario monitor continuously monitors the flow status, data characteristic changes, data volume fluctuations, and abnormal data ratio of standardized data events in the distributed data buffer in real time. When any data indicator is detected to meet the system's preset triggering conditions, an analysis task is automatically generated, and initial information such as task background, analysis scope, and data source is supplemented according to the monitoring scenario.

[0038] Please refer to Figure 5 In a preferred embodiment of the present invention, a preset distributed task parser is used to parse the target macro-task into fine-grained subtasks, which may include: Step S310: Deploy the preset distributed task parser across multiple nodes and perform parallel task decomposition using domain knowledge graphs and task decomposition algorithms.

[0039] In a preferred embodiment of the present invention, the core decomposition is based on a domain knowledge graph and a task decomposition algorithm. The domain knowledge graph provides a professional logical decomposition framework for analysis tasks in different domains, and the task decomposition algorithm ensures the uniformity of granularity and executability of subtasks.

[0040] Step S320: Subtask decomposition follows the core principle of meta-capability matching. Each subtask can be completed independently with only one or a few specific meta-capabilities, ensuring that the analysis agent matches and executes accordingly.

[0041] Step S330: Clearly label the logical relationships between each subtask and record the correspondence between the subtask and the target macro task.

[0042] In a preferred embodiment of the present invention, during the decomposition process, a clear logical relationship is marked for each subtask, clarifying the parallel, serial, and dependency relationships between subtasks. At the same time, the correspondence between subtasks and the target macro task is recorded to ensure that the final answer to the target macro task can be obtained by integrating the results of the subtasks.

[0043] Step S340: After decomposition, a standardized subtask list is generated, each subtask is assigned corresponding subtask core information, and bound to the task number of the target macro task, and stored in the task information database.

[0044] In a preferred embodiment of the present invention, the core information of the subtask may include the subtask objective, the required meta-capability type, the characteristics of the input standardized data events, the logical relationship with other subtasks, and the result output format, etc.

[0045] In a preferred embodiment of the present invention, the distributed task parser encapsulates each subtask into a standardized task announcement according to the system's preset subtask announcement encapsulation specification; a decentralized publishing mechanism is adopted to synchronously publish all standardized task announcements to the system's distributed task announcement board; the standardized task announcements are continuously retained on the distributed task announcement board until the subtask is completed or times out, the timeout time is set according to the priority of the subtask, the timeout time for higher priority subtasks is shorter, and the system sends an alert to the operation and maintenance end for subtasks that have not been accepted after the timeout, reminding manual intervention.

[0046] For example, taking an industrial IoT production line scenario as an example, the system completes the triggering, parsing, and distributed deployment of analysis tasks: The product yield of this production line experiences unexplained fluctuations. Factory maintenance personnel proactively submit a macro-analysis task—"diagnosing the root cause of recent product yield fluctuations on a certain production line"—through the system's visual human-machine interface. The task is set as high priority with a completion time of 2 hours. The system assigns a unique task number and completes initial information entry. The system's distributed task parser, based on an industrial production knowledge graph and task decomposition algorithm, breaks down this macro-task into three logically related fine-grained sub-tasks: Sub-task 1, "Anomaly detection of production line equipment timing data"; Sub-task 2, "Production line material batch data and yield data"; and Sub-task 3, "Anomaly detection of production line equipment timing data"; Sub-task 4, "Production line material batch data and yield data"; and Sub-task 5, "Production line material batch data and yield data"; and Sub-task 6, "Production line material batch data and yield data"; and Sub-task 7, "Production line material batch data and yield data"; and Sub-task 8, "Production line material batch data and yield data"; and Sub-task 9, "Production line material batch data and yield data"; and Sub-task 1, "Production line equipment timing data anomaly detection"; and Sub-task 1, "Production line material batch data and yield data"; and Sub-task 1, "Production line material batch data and yield data"; and Sub-task 1, "Production line material batch data and yield data"; and Sub-task 2 ... The system performs three tasks: "Relational Analysis of Data," "Keyword Extraction and Event Sorting from Workshop Maintenance Logs," and "Keyword Extraction and Event Sorting from Workshop Maintenance Logs." These three tasks are marked as parallel and a standardized sub-task list is generated. The distributed task parser encapsulates the three sub-tasks into a standardized task announcement. It specifies that Subtask 1 requires the "Time-Series Data Anomaly Detection" meta-capability and inputs standardized data events related to equipment operation; Subtask 2 requires the "Data Relational Analysis" meta-capability and inputs standardized data events related to material batches and yield; and Subtask 3 requires the "Text Keyword Extraction" meta-capability and inputs standardized data events related to maintenance logs. The system also specifies task quality and time requirements. Subsequently, the three standardized task announcements are simultaneously published to the distributed task bulletin board and a lightweight notification is sent to all analytical agents.

[0047] In this embodiment of the invention, the autonomous solution module can be used to bid for and accept analysis tasks for analysis agents based on the global collaboration protocol and the sub-tasks on the distributed task bulletin board, and dynamically form a dynamic temporary task chain network for analysis tasks to perform emergent solution of analysis agents based on group interaction.

[0048] Please refer to Figure 6 In a preferred embodiment of the present invention, the process of bidding for and accepting analysis tasks by the analytical agent, and dynamically forming a dynamic temporary task chain network for the analysis tasks, may include: Step S410: After the agent listens to the subtasks on the distributed task bulletin board, it matches the agent's capability profile with the subtask's requirements (for example, determining that the meta-capability type, performance indicators, and resource status all match the subtask's requirements).

[0049] Step S420: Analyze the intelligent agent to submit bidding information to the system in a standardized encrypted format within the response time threshold specified by the global cooperation protocol (for example, the bidding information includes its own cluster identifier, meta-capability matching degree, historical performance indicators of similar tasks, current load status, expected task completion time, resource usage commitment and other core contents).

[0050] Step S430: A decentralized bidding negotiation mechanism is adopted to confirm the acceptance of bidding information. There is no central node involved. All bidding analysis agents are comprehensively scored according to the preset scoring algorithm.

[0051] In a preferred embodiment of the present invention, the scoring dimensions may include meta-capability matching degree (e.g., weight 40%), historical performance indicators of similar tasks (e.g., weight 30%), current load status (e.g., weight 20%), and reputation score (e.g., weight 10%), and the scores are sorted.

[0052] Step S440: The system sends an encrypted order acceptance confirmation notification to the top-ranked analytical agent. After accepting the order, the analytical agent locks the subtask.

[0053] In a preferred embodiment of the present invention, the top-ranked analytical agent or group of analytical agents becomes the subtask execution subject. The system sends an encrypted order acceptance confirmation notification to it, while simultaneously sending a bid failure notification to other bidding analytical agents. After accepting the order, the analytical agent immediately locks the subtask to avoid accepting duplicate orders, and locks its committed hardware resources during task execution, and does not participate in bidding for other subtasks.

[0054] Step S450: After receiving the order, the analytical agent generates standardized collaboration invitation information according to the provisions of the global collaboration protocol (for example, the collaboration invitation information includes its own cluster identifier, the number of the currently executed subtask, the required meta-capability type, the characteristics of the intermediate analysis results to be processed, the description of collaboration requirements, the collaboration time requirements, etc.).

[0055] Step S460: After receiving the collaboration invitation, other analytical agents decide whether to respond based on their own agent capability profile and current load status.

[0056] In a preferred embodiment of the present invention, the responding analytical agent is required to submit encrypted response information to the collaboration invitation initiator within a specified time, including its own cluster identifier, meta-capability matching degree, historical collaboration efficiency, reputation score, and other content.

[0057] Step S470: The collaboration invitation initiator filters all responding analytical agents according to the rules of the global collaboration protocol. Through multiple rounds of collaboration invitations and responses, the system spontaneously forms a dynamic temporary task chain network for the analysis task.

[0058] In a preferred embodiment of the present invention, the selection criteria are meta-capability matching degree, historical collaboration efficiency, and reputation score. After selecting the optimal collaborative analysis agent, a collaboration confirmation notification is sent to it, and the two establish a point-to-point encrypted collaboration link. The dynamic temporary task chain network uses sub-tasks as nodes and collaboration relationships as links, supporting the dynamic joining and leaving of analysis agents. When an analysis agent fails, a collaboration invitation can be quickly re-initiated, ensuring the flexibility and robustness of the network.

[0059] Please refer to Figure 7 In a preferred embodiment of the present invention, the emergence solution of analytical agents based on group interaction may include: Step S401: Set data and intermediate result flow rules for the dynamic temporary task chain network, and standardize the flow of data events and intermediate analysis results along the preset encrypted collaboration link.

[0060] Step S402: Each analytical agent, as a node, performs specialized processing on the received standardized data events or intermediate analysis results only within its own meta-capabilities, and generates new intermediate analysis results after specialized processing.

[0061] Step S403: Set the criteria for emergence convergence. When the intermediate analysis results in the dynamic temporary task chain network reach the preset conditions (e.g., the intermediate analysis results no longer generate new effective information, the analysis conclusions tend to be stable, or the preset number of iterations is reached), or all sub-tasks are completed, the emergence of the dynamic temporary task chain network is determined to have reached the convergence state.

[0062] Step S404: Integrate the analysis results generated by each analysis agent in the converged dynamic temporary task chain network to generate one or more preliminary candidate solution fragments for the target macro task (e.g., the fragments include information such as analysis process, core conclusions, data support, result confidence, and processing logs).

[0063] For example, continuing the scenario of production line yield fluctuation analysis, the self-organization and emergence of analytical agents are solved: After analytical agent D (e.g., its meta-capability is time-series data anomaly detection) listens to the standardized task announcement of subtask 1, it is determined to meet the bidding conditions through precise matching and submits encrypted bidding information within 5 seconds. After decentralized bidding negotiation, the system determines that it is the executor of subtask 1. Analytical agents E and B become the executors of subtasks 2 and 3, respectively. The system updates the subtask status in the task information database to "in execution". When analyzing agent D is processing standardized data events related to equipment time-series operation, it finds that data anomalies are strongly correlated with specific time periods. It needs to sort out the operation and maintenance events in that time period. Therefore, it broadcasts standardized collaboration invitation information to the cluster according to the global collaboration protocol. Analytical agent B responds to the invitation and passes the screening. The two establish a point-to-point encrypted collaboration link. Agent B extracts the core event of "preventive maintenance of equipment" from the standardized data events related to the operation and maintenance work order logs during this period. It finds that this maintenance operation involves material replacement and requires further analysis of the correlation between materials and yield. Therefore, it encapsulates this clue into a standardized data event and passes it to the analysis agent E. The two establish an encrypted collaborative link. The three analysis agents spontaneously form a dynamic temporary task chain network. The standardized data event and intermediate analysis results flow along the link and are processed layer by layer. When the analysis conclusion tends to stabilize, the network reaches a convergence state and jointly emerges a preliminary candidate solution fragment: "The abnormal fluctuation of the production line equipment is related to the material change after preventive maintenance, and the yield fluctuation may be caused by this material change." The fragment contains abnormal data characteristics, operation and maintenance event records, correlation analysis process, data support, and conclusion confidence. At the same time, the fragment is bound to the macro task number and stored in the solution library.

[0064] In this embodiment of the invention, the deep solution module can be used to use a preset dilemma judgment rule system to perceive emergent dilemmas and generate external assistance requests, match external knowledge brains through an external resource catalog, and construct a hybrid intelligent network that integrates internal and external resources to solve emergent dilemmas and achieve deep solution of analysis tasks.

[0065] In a preferred embodiment of the present invention, the judgment indicators of the preset dilemma judgment rule system include the number of solution iterations, the confidence level of the analysis results, the effective output rate of intermediate results, and the stall time of subtask processing. Each indicator has a clearly defined threshold. When any indicator reaches the preset threshold, the dilemma perception is triggered. The group of analytical agents in the dynamic temporary task chain network adopts a distributed negotiation mechanism to cross-confirm the dilemma perception results at multiple nodes, ensuring the accuracy of the dilemma judgment and avoiding the false triggering of external assistance requests due to the failure of a single analytical agent. After the dilemma is confirmed through negotiation, the group of analytical agents in the dynamic temporary task chain network jointly generates a standardized external assistance request. The request strictly follows the system's preset encapsulation specifications, and its core includes the feature description of the required external capabilities, the details of the current dilemma, the obtained preliminary candidate solution fragments, the standardized data event features required for input, the expected requirements of the analysis results, and the problem context information.

[0066] Please refer to Figure 8 In a preferred embodiment of the present invention, matching external knowledge entities through an external resource directory may include: Step S510: The system maintains and updates the external resource catalog, registering all interoperable external knowledge brain information in the external resource catalog.

[0067] In a preferred embodiment of the present invention, the external knowledge brain information may include professional capabilities, service invocation methods, data interaction formats, reputation ratings, invocation costs, service response timeliness, data security measures, etc. The types of external knowledge brains include third-party professional simulation APIs, domain-specific knowledge models, real-time expert databases, industry large models, professional computing clusters, etc.

[0068] Step S520: The external resource catalog is updated by the corresponding catalog maintenance module. Within a preset period, all external knowledge entities are tested for availability (e.g., external knowledge entities that are unavailable, have timed out, or have inaccurate results are removed, while external knowledge entities that have passed compliance review are added).

[0069] Step S530: Using a preset multi-dimensional weighted scoring algorithm, the system's external resource docking module matches external assistance requests with external resource directories.

[0070] In a preferred embodiment of the present invention, the scoring dimensions may include professional competence fit (e.g., weight 50%), reputation rating (e.g., weight 20%), service response timeliness (e.g., weight 15%), and call cost (e.g., weight 15%), and the optimal 1-3 candidate external knowledge bodies are selected based on the scoring results.

[0071] Step S540: The system conducts secure service negotiations with the candidate external knowledge entity. After reaching an agreement, the system and the external knowledge entity sign a standardized service collaboration agreement.

[0072] In a preferred embodiment of the present invention, the negotiation process adopts end-to-end asymmetric encrypted communication throughout. The negotiation content includes the final confirmation of the call cost, the encryption algorithm and key exchange method for data transmission, the data format of input and output, the time commitment for result return, the definition of data security responsibility, and the service termination conditions. If the negotiation fails, the next external knowledge entity is selected from the candidate list and the negotiation is carried out again until the negotiation is successful or the candidate list is empty.

[0073] Please refer to Figure 9 In a preferred embodiment of the present invention, constructing a hybrid intelligent network that integrates internal and external technologies to solve the emergent problem and achieve deep solving of analysis tasks may include: Step S501: Map the successfully negotiated external knowledge brain to an invited intelligent agent, and assign it a temporary cluster identifier through the intelligent agent registration center.

[0074] In a preferred embodiment of the present invention, the temporary cluster identifier includes an "invited" identifier, enabling it to adapt to the system's global collaboration protocol and achieve seamless integration with the dynamic temporary task chain network.

[0075] Step S502: The invited intelligent agent uses the external knowledge brain as a carrier and dynamically links to the internal dynamic temporary task chain network through an encrypted channel to form a hybrid intelligent network that integrates internal and external systems.

[0076] In a preferred embodiment of the present invention, the original collaborative links of internal analysis agents are retained in the hybrid intelligent network, and a new point-to-point encrypted collaborative link is added between invited agents and relevant internal analysis agents.

[0077] Step S503: The internal analysis agent encrypts the relevant standardized data events and fragments of preliminary candidate solutions, and transmits them to the invited agent through a dedicated encrypted link.

[0078] In a preferred embodiment of the present invention, data transmission strictly follows the provisions of the service collaboration protocol. The internal analysis agent encrypts the relevant standardized data events and preliminary candidate solution fragments and then transmits them to the invited agent through a dedicated encrypted link. The invited agent's calculation process is completed independently on its own platform and does not occupy the system's internal computing resources.

[0079] Step S504: The invited intelligent agent analyzes the received information and encapsulates the processing results into a standardized data event format.

[0080] In a preferred embodiment of the present invention, the invited intelligent agent uses its own professional capabilities to analyze, verify or quantify the received information, and after completion, encapsulates the processing result into a standardized data event format, along with information such as result confidence and calculation log, and feeds it back to the internal hybrid intelligent network via an encrypted link.

[0081] Step S505: The internal analysis agent, in conjunction with the processing results of the invited agent, refines the preliminary candidate solution fragments to achieve in-depth solution of the analysis task.

[0082] In a preferred embodiment of the present invention, the release trigger condition for the external link is that the invited intelligent agent completes the solution of the corresponding sub-problem, helps the internal hybrid intelligent network overcome the solution bottleneck, and the internal analysis intelligent agent completes the integration and improvement of the external processing results, enhancing the end of the solution process. The system automatically executes the link release process according to the provisions of the service collaboration agreement. First, it removes the temporary cluster identifier of the invited intelligent agent, causing it to leave the system's analysis intelligent agent cluster. Then, it disconnects the point-to-point encrypted collaboration link with the internal hybrid intelligent network. Finally, it releases the relevant docking resources of the system's external resource docking module. The system records the participation process of the external knowledge brain (e.g., the invited intelligent agent) in all dimensions. Based on the recorded all-dimensional information, the system's directory maintenance module updates the reputation rating, applicable scenario tags, service evaluation, and other content of the external knowledge brain in the external resource directory.

[0083] For example, in the scenario of analyzing production line yield fluctuations, the system dynamically links and enhances the solution for external knowledge entities: The dynamic temporary task chain network formed by the internal analysis agents proposes a preliminary candidate solution fragment for "material switching causing yield fluctuations." However, none of the internal analysis agents are capable of quantifying and verifying this hypothesis, nor can they analyze the specific impact of material switching on the production process. The confidence level of the analysis results is lower than a preset threshold, triggering the perception of a solution dilemma. After distributed negotiation by the group, the solution dilemma is confirmed. The group generates a unified standardized external assistance request, clearly specifying the need for causal quantitative analysis capabilities for "materials-process-product quality," and submits it to the system's external resource docking module by the master node. The system's external resource docking module accurately matches the external assistance request with the external resource catalog, selecting a professional industrial materials simulation platform as a candidate external knowledge entity. Subsequently, end-to-end encrypted negotiation is conducted with this platform, determining the asymmetric encryption transmission method, the 2-hour result return time, and the per-call cost, and signing a standardized service collaboration agreement. The materials simulation platform is mapped as an invited intelligent agent. After being assigned a temporary cluster identifier by the agent registration center, it is dynamically linked to the internal dynamic temporary task chain network through an encrypted link, forming a hybrid intelligent network. The internal analysis agent encrypts and transmits standardized data events related to the production line, such as material parameters, process parameters, and yield data, to the invited intelligent agent. The platform uses its own professional simulation model to simulate the product quality fluctuation curve under different material parameters, quantitatively verifying the hypothesis that "some parameters do not match the production process after material switching, resulting in significant fluctuations in product yield," and provides optimization suggestions for material parameters. The results are then encrypted and fed back to the internal hybrid intelligent network. The internal analysis agent combines these results to improve the preliminary candidate solution fragments and complete the enhanced solution. After the solution is completed, the system automatically releases the external link with the industrial materials simulation platform according to the service collaboration agreement, removes its temporary cluster identifier, and records the accuracy, response speed, and collaboration efficiency of its simulation results. The directory maintenance module updates its reputation rating in the external resource directory based on the recorded information.

[0084] In this embodiment of the invention, the feedback module can be used to integrate the emergent solution and the deep solution into an analysis report, cross-validate the analysis report in multiple dimensions, deliver the final analysis report and collect multi-dimensional feedback, assign multi-dimensional credit to the analysis agent and the external knowledge brain, and realize the adaptive evolution of the system operation rules.

[0085] In a preferred embodiment of the present invention, after the preliminary analysis report draft is generated, it is immediately sent to the system's verification module for multi-dimensional cross-verification. The verification methods include: comparison with the results of similar historical analysis cases, verification of the logical consistency of the report through the domain rule base, backtracking verification of the core conclusions in the report using original standardized data events, and cross-verification of the analysis results of external knowledge bodies. If the verification fails, the verification module reports the specific verification issues to the distributed solution synthesizer, which makes targeted corrections and improvements based on the issues. If the verification passes, the verification module adds a verification pass mark to the preliminary analysis report draft and proceeds to the next step.

[0086] In a preferred embodiment of the present invention, the verification module, based on the results of multi-dimensional cross-validation, makes targeted corrections and optimizations to unreasonable parts of the preliminary analysis report draft, supplements missing analysis information, and corrects erroneous conclusions or data; adopts a multi-terminal adapted delivery method to simultaneously deliver the final analysis report to the user's human-computer interaction interface, mobile terminal, email, and other user-specified channels; conducts explicit feedback collection, setting standardized feedback entry points on the system's human-computer interaction interface and report delivery page to collect explicit feedback information such as user quality ratings of the report, suggestions for correcting core conclusions, feasibility evaluations of solutions, and completeness evaluations of the report content; conducts implicit feedback collection, collecting implicit feedback data such as the user's subsequent operational behavior based on the analysis report, operation time, the degree of consistency between operation results and report suggestions, and the actual problem-solving effect through the system's user behavior monitoring module; sets a feedback collection cycle, with explicit feedback collected in real time, and implicit feedback having different collection cycles set according to the type of analysis task, ensuring that the collected feedback information can truly reflect the actual use effect of the report.

[0087] In a preferred embodiment of the present invention, the system establishes a multi-dimensional credit allocation index system, setting different indicators and weights for the analytical agent and the external knowledge brain respectively. The system adopts a decentralized credit allocation algorithm, calculating the credit score of each analytical agent and the external knowledge brain based on the completion quality of the final analysis report, user feedback information, and the actual performance of each participating entity in the task processing process, combined with the credit allocation index system and weight settings. The system adjusts the reputation score of the analytical agent based on the credit score, updates the relevant indicators of the dynamic layer of the agent's capability profile, and the analytical agent with a high credit score receives a higher score weight in subsequent task bidding. The system also completes the rating adjustment of the external knowledge brain based on the credit score. The system's directory maintenance module updates the reputation rating of the external knowledge brain in the external resource directory based on the credit score, and refines the applicable scenario tags based on its performance. The external knowledge brain with a high credit score receives a higher priority in subsequent external assistance matching.

[0088] In a preferred embodiment of the present invention, the system's rule evolution engine performs in-depth analysis and mining of the entire process data. The analysis logic is based on machine learning algorithms to extract success patterns, failure problems, optimization space, and user needs and preferences in the current task collaboration process, and accurately identify the deficiencies in the system's operating rules. Based on the results of the in-depth analysis, the core operating rules of the system are finely tuned and optimized. The optimization types include: analyzing the bidding strategies and scoring weights of intelligent agents in specific scenarios, analyzing the reputation scoring formulas and penalty measures between intelligent agents, the matching algorithm and scoring dimensions of external knowledge brains, the task decomposition granularity rules of the distributed task parser, and the collaborative behavior norms and response thresholds in the global collaboration protocol.

[0089] Please refer to Figure 2 and Figure 10 This invention also provides an integrated intelligent data analysis method, which may include: Step S100: Through a unified data access gateway, collect and access multi-source heterogeneous data of the target scenario in a standardized manner, perform data cleaning and formatting processing, and generate standardized data events.

[0090] Step S200: Based on standardized data events, construct a distributed cluster composed of lightweight analytical agents, establish a dynamically updated agent capability profile for each analytical agent, and inject a unified global collaboration protocol into all analytical agents.

[0091] Step S300: Through various analysis task triggering modes, a preset distributed task parser is used to parse the target macro task into fine-grained sub-tasks, and a decentralized publishing mechanism is adopted to publish the sub-tasks to the system's distributed task bulletin board.

[0092] Step S400: Based on the global collaboration protocol and the subtasks on the distributed task bulletin board, the analysis agents bid for and accept analysis tasks, and dynamically form a temporary task chain network for the analysis tasks to perform emergent solutions based on group interaction.

[0093] Step S500: Using a pre-defined dilemma judgment rule system, perform emergent problem perception and external assistance request generation, match external knowledge brains through an external resource catalog, and construct a hybrid intelligent network that integrates internal and external resources to solve emergent problem dilemmas and achieve in-depth solution of analysis tasks.

[0094] Step S600: Integrate the emergent solution and the deep solution into an analysis report, cross-validate the analysis report in multiple dimensions, deliver the final analysis report and collect multi-dimensional feedback, assign multi-dimensional credits to the analysis agent and the external knowledge brain, and realize the adaptive evolution of the system operation rules.

[0095] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0100] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0101] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0102] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0103] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A data-integrated intelligent analysis system, characterized in that, The integrated data intelligent analysis system includes: The data acquisition module is used to collect and access multi-source heterogeneous data from the target scenario in a standardized manner through a unified data access gateway, perform data cleaning and formatting, and generate standardized data events. The agent building module is used to build a distributed cluster of lightweight analytical agents based on standardized data events, and to create a dynamically updated agent capability profile for each analytical agent, and to inject a unified global collaboration protocol into all analytical agents. The task publishing module is used to analyze various task triggering modes, use a preset distributed task parser to parse the target macro task into fine-grained sub-tasks, and use a decentralized publishing mechanism to publish the sub-tasks to the system's distributed task bulletin board. The autonomous solution module is used to bid for and accept analysis tasks for analysis agents based on the global collaboration protocol and sub-tasks on the distributed task bulletin board, and dynamically form a dynamic temporary task chain network for analysis tasks to perform emergent solution of analysis agents based on group interaction. The deep solution module is used to perceive emergent solution dilemmas and generate external assistance requests by utilizing a preset dilemma judgment rule system. It matches external knowledge brains through an external resource catalog and constructs a hybrid intelligent network that integrates internal and external resources to solve emergent solution dilemmas, thereby achieving deep solution of analysis tasks.

2. The integrated data intelligent analysis system according to claim 1, characterized in that, The data cleaning and formatting process includes: Perform a two-layer identification operation on multi-source heterogeneous data to generate data identification labels containing multi-dimensional information; Based on data identification tags, a pre-set dedicated data cleaning and formatting module is used to match the corresponding data types; Perform customized and refined processing operations on different types of data and generate processing logs to achieve full traceability of the data processing process.

3. The integrated data intelligent analysis system according to claim 1, characterized in that, The process of establishing a dynamically updated agent capability profile for each analytical agent includes: Through the agent registration center, a corresponding dynamic capability profile of the agent is maintained for each analytical agent. The dynamic capability profile of the agent is divided into a basic layer and a dynamic layer. The base layer records and analyzes the static core information of the intelligent agent, while the dynamic layer records and analyzes the dynamic performance indicators and resource consumption patterns exhibited by the intelligent agent when processing historical analysis tasks. Set up a profile update trigger mechanism: a small update is triggered after the intelligent agent completes a single subtask, and a full update is triggered after a preset number of subtasks are completed.

4. The integrated data intelligent analysis system according to claim 1, characterized in that, The step of using a pre-defined distributed task parser to parse the target macro-task into fine-grained subtasks includes: The pre-defined distributed task parser is deployed across multiple nodes, and parallel task decomposition is performed using domain knowledge graphs and task decomposition algorithms. Subtask decomposition follows the core principle of meta-capability matching. Each subtask can be completed independently with only one or a few specific meta-capabilities, ensuring that the analysis agent matches and executes accordingly. Clearly label the logical relationships between each subtask and record the correspondence between the subtask and the target macro task; After decomposition, a standardized list of subtasks is generated. Each subtask is assigned corresponding core information and bound to the task number of the target macro task, and then stored in the task information database.

5. The integrated data intelligent analysis system according to claim 1, characterized in that, The process of bidding for and accepting analysis tasks by the analytical agent, and dynamically forming a dynamic temporary task chain network for the analysis tasks, includes: After the intelligent agent listens to the subtasks on the distributed task bulletin board, it matches the requirements of the subtasks with the intelligent agent's capability profile. The analysis shows that the intelligent agent submits bidding information to the system in a standardized encrypted format within the response time threshold specified by the global cooperation protocol; A decentralized bidding negotiation mechanism is adopted to confirm the acceptance of bidding information. There is no central node involved. All bidding analysis agents are comprehensively scored according to a preset scoring algorithm. The system sends an encrypted order acceptance confirmation notification to the top-ranked analytical agent, which then locks the subtask after accepting the order. After receiving the order, the analytical agent generates standardized collaboration invitation information according to the global collaboration protocol. After receiving the collaboration invitation, other analytical agents will decide whether to respond based on their own agent capability profile and current load status. The party initiating the collaboration invitation filters all responding analytical agents according to the rules of the global collaboration protocol. Through multiple rounds of collaboration invitations and responses, the system spontaneously forms a dynamic temporary task chain network for the analysis task.

6. The integrated data intelligent analysis system according to claim 5, characterized in that, The emergence solution of the analytical agent based on group interaction includes: Set data and intermediate result flow rules for dynamic temporary task chain networks, and standardize the flow of data events and intermediate analysis results along the preset encrypted collaboration links; Each analytical agent, acting as a node, performs specialized processing on the received standardized data events or intermediate analysis results within its own meta-capabilities, generating new intermediate analysis results after specialized processing. Set the criteria for emergence convergence. When the intermediate analysis results in the dynamic temporary task chain network reach the preset conditions, or when all subtasks are completed, the emergence of the dynamic temporary task chain network is determined to have reached convergence. The analysis results generated by each analytical agent in the converged dynamic temporary task chain network are integrated to generate one or more preliminary candidate solution fragments for the target macroscopic task.

7. The integrated data intelligent analysis system according to claim 1, characterized in that, The matching of external knowledge entities through external resource directories includes: The system maintains and updates a dynamic external resource catalog, which registers all interoperable external knowledge information. The external resource catalog is updated by the corresponding catalog maintenance module, and the availability of all external knowledge entities is tested within a preset period. Using a pre-defined multi-dimensional weighted scoring algorithm, the system's external resource docking module matches external assistance requests with the external resource directory; The system conducts secure service negotiations with candidate external knowledge entities. Once an agreement is reached, the system and the external knowledge entity sign a standardized service collaboration agreement.

8. The integrated data intelligent analysis system according to claim 7, characterized in that, The construction of a hybrid intelligent network integrating internal and external technologies addresses the emergent problem and enables in-depth solutions to analytical tasks, including: The successfully negotiated external knowledge entities are mapped as invited intelligent agents, and temporary cluster identifiers are assigned to them through the intelligent agent registration center. The specially invited intelligent agents use external knowledge brains as carriers and dynamically link to the internal dynamic temporary task chain network through encrypted channels, forming a hybrid intelligent network that integrates internal and external systems. The internal analysis agent encrypts relevant standardized data events and fragments of preliminary candidate solutions, and transmits them to the invited agent through a dedicated encrypted link. The invited intelligent agent analyzes the received information and encapsulates the processing results into a standardized data event format. The internal analysis agent, in conjunction with the processing results of the invited agent, refines the preliminary candidate solution fragments, thereby achieving in-depth solution of the analysis task.

9. The integrated data intelligent analysis system according to claim 1, characterized in that, The integrated data intelligent analysis system also includes: The feedback module integrates the emergent solution and the deep solution into an analysis report, performs multi-dimensional cross-validation of the analysis report, delivers the final analysis report and collects multi-dimensional feedback, assigns multi-dimensional credits to the analysis agent and the external knowledge brain, and realizes the adaptive evolution of the system's operating rules.

10. A data-integrated intelligent analysis method, characterized in that, The integrated intelligent data analysis method includes: Through a unified data access gateway, multi-source heterogeneous data from the target scenario is collected and accessed in a standardized manner, and data is cleaned and formatted to generate standardized data events. Based on standardized data events, a distributed cluster composed of lightweight analytical agents is formed, and a dynamically updated agent capability profile is established for each analytical agent, injecting a unified global collaboration protocol into all analytical agents. By analyzing various task triggering modes and using a preset distributed task parser, the target macro task is parsed into fine-grained sub-tasks, and a decentralized publishing mechanism is used to publish the sub-tasks to the system's distributed task bulletin board. Based on the global collaboration protocol and the subtasks on the distributed task bulletin board, the analysis agents bid for and accept analysis tasks, and dynamically form a temporary task chain network for the analysis tasks, and perform emergent solutions for analysis agents based on group interaction. By utilizing a pre-defined system of rules for identifying dilemmas, the system can perceive emerging dilemmas and generate requests for external assistance. It can also match external knowledge entities through an external resource catalog and construct a hybrid intelligent network that integrates internal and external resources to solve emerging dilemmas and achieve in-depth solutions to analytical tasks. The emergent solution and the deep solution are integrated into an analysis report. The analysis report is cross-validated in multiple dimensions. The final analysis report is delivered and multi-dimensional feedback is collected. Multi-dimensional credit is assigned to the analysis agent and the external knowledge brain, so as to realize the adaptive evolution of the system operation rules.

Images