A brain-inspired data service system and method thereof

Abstract

The application relates to a brain-inspired data service system and method, characterized by constructing a hierarchical architecture: a mixed deployment data node cluster, specifically composed of core nodes, collection nodes and gateway nodes, respectively processing core business data, real-time edge data and cross-domain connection; a small-world connection layer realizes efficient communication in the same domain through high-cluster local connection and breaks data islands through dynamic long-range cross-domain connection; an intelligent control layer integrates a DMN control hub, cooperates with a state monitoring engine, a prediction analysis module and a resource scheduler driven by reinforcement learning. The system initializes the topological structure through a small-world network manager, dynamically inserts a long-range connection when the cross-domain node communication frequency exceeds a threshold value, and ensures that the cross-domain path hop count is stable and does not exceed a preset value. The DMN control hub realizes double-mode switching based on the system load threshold value: in the idle period, data pre-cleaning, knowledge graph construction and predictive prefetching and resource pre-allocation are performed; in the task period, a routing optimization algorithm is combined with graph traversal.

Description

Technical Field

[0001] This application relates to the fields of distributed computing and artificial intelligence technology, and specifically to a brain-inspired data service system and method thereof. Background Technology

[0002] With the rapid development of big data, cloud computing, the Internet of Things, and artificial intelligence technologies, modern data service systems face the dual challenges of architectural complexity and service intelligence. In large enterprise or organizational environments, data is often scattered across multiple independent departmental systems, creating severe data silos. While traditional centralized architectures are simple to manage, as data scales up, the central node easily becomes a performance bottleneck and a single point of failure, and struggles to handle cross-departmental data collaborative analysis needs. Although existing distributed systems alleviate storage pressure through node expansion, they still have significant shortcomings in data discovery efficiency, service path optimization, and dynamic resource scheduling.

[0003] Current mainstream data service systems generally adopt a passive response model, only performing corresponding operations upon receiving user requests, lacking the ability to proactively integrate and preprocess data resources. This mechanism leads to three prominent problems: first, cross-domain data access requires multiple layers of relay, significantly increasing communication latency; second, system resource allocation lacks predictability and cannot be dynamically adjusted according to business fluctuations; and third, a large amount of computing resources remain idle during off-peak hours, resulting in serious resource waste. More importantly, existing systems lack the self-learning and adaptive capabilities similar to biological neural networks, making it difficult to establish intelligent data association networks and predictive service mechanisms. Summary of the Invention

[0004] The purpose of this application is to provide a brain-inspired data service system and method, which has the advantages of reducing cross-domain data access latency, realizing dynamic resource scheduling, actively integrating pre-processed data resources, and improving system response efficiency and resource utilization.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a brain-inspired data service system, comprising:

[0006] Data node cluster: Composed of several distributed data nodes, used to store, process, analyze data or provide data services. The data nodes are at least one of servers, containers, microservices, logical units or AIAgents, including core nodes deployed on cloud server clusters, edge nodes deployed on edge computing devices and gateway nodes deployed on containerized platforms, respectively corresponding to the core data processing, edge data collection and cross-domain data connection needs of the business domain.

[0007] Small-world connection layer: The connection topology between data nodes is constructed using a small-world network model, including a small-world network manager, local high-cluster connections, and long-range cross-domain connections. Local high-cluster connections enable dense interconnection of data nodes within the same business domain, while long-range cross-domain connections dynamically maintain cross-domain shortcuts through the small-world network manager.

[0008] Intelligent Control Layer: Includes the DMN control center and its subordinate status monitoring engine, predictive analysis module, and resource scheduler, serving as the system control center to simulate the default mode network functions of the brain; External Access Layer: Provides user application access interfaces and protocol adapters;

[0009] Infrastructure layer: A hybrid architecture deployed across cloud server clusters, containerized platforms, edge computing devices, and physical servers;

[0010] The intelligent control layer collects communication hop counts and latency metrics of the small-world connection layer in real time through the state monitoring engine, driving the predictive analysis module to generate topology optimization instructions; the DMN control center automatically triggers mode switching based on the real-time data from the state monitoring engine through a threshold comparison algorithm, and pushes hot zone weights to the small-world network manager.

[0011] By adopting the above technical solutions and constructing a data node cluster that includes a hybrid deployment of cloud core, edge nodes, and containerized gateways, and introducing a brain-inspired small-world connection layer and DMN intelligent control layer, the resource silos and efficiency bottlenecks of traditional systems are solved. The technical effect is that resources are accurately adapted according to scenarios: core processing, edge acquisition, and cross-domain connection. By utilizing the characteristics of small-world topology: efficient local aggregation and global short path reachability, data transmission efficiency is significantly improved. Furthermore, through the global perception and intelligent regulation of the DMN control center, such as hot zone weight push, the system's adaptability and cross-domain service capabilities are enhanced, laying the foundation for building an efficient, intelligent, and robust data service system.

[0012] The present invention is further configured such that: the small-world network manager of the small-world connection layer implements:

[0013] The network topology is initialized based on a variant of the Watts-Strogatz model algorithm, and long-range cross-domain connections are introduced by random reconnection edges.

[0014] The node communication logs are analyzed according to a preset period. When the frequency of communication between cross-domain nodes exceeds the threshold, long-range connections are dynamically inserted through the service mesh to reduce the number of communication path hops.

[0015] By adopting the above technical solution, the construction and dynamic adjustment mechanism of the small-world network manager is specifically defined. Based on Watts-Strogatz variant initialization, long-range connections are dynamically inserted according to communication frequency thresholds. The technical effect is to ensure that the network has high clustering from the beginning and can intelligently optimize the topology structure according to actual cross-domain communication needs. Through service mesh technology, direct "shortcuts" are dynamically established without being noticed, which effectively reduces the number of hops in the originally lengthy communication path, significantly reduces the latency of cross-business domain data access, improves the system's responsiveness to dynamic data access patterns, and solves the problem of low cross-domain efficiency caused by data silos.

[0016] The present invention is further configured such that the DMN control center of the intelligent control layer works collaboratively through the following modules:

[0017] Status monitoring engine: Real-time collection of CPU load, network throughput, and container resource utilization metrics of data nodes to generate system heatmaps;

[0018] Predictive analytics module: Equipped with a prediction engine, it uses an LSTM model to predict the peak value of real-time data collection at edge nodes and uses a Transformer model to analyze cross-domain query patterns.

[0019] Resource scheduler: Based on reinforcement learning algorithm, dynamically adjust the resource quotas of core nodes and edge nodes. When the data collection volume of edge nodes suddenly increases, automatically migrate the CPU cores / memory quota based on the containerized platform and dynamically expand the computing resources to the edge nodes according to 120% of the current load of the edge nodes.

[0020] By adopting the above technical solution, this paper describes in detail the collaborative working method and specific functions of the internal modules of the intelligent control layer, including the status monitoring engine, predictive analysis module, and resource scheduler. These include collecting multi-dimensional indicators, using LSTM / Transformer for prediction, and dynamically scheduling resources based on reinforcement learning. The technical effect is that it achieves refined and real-time monitoring and visualization of system status, including CPU, network, and container resources, and provides a heat map. It also uses advanced predictive models such as LSTM to capture short-term edge peaks and Transformer to analyze long-term cross-domain patterns, thus anticipating changes in demand. Furthermore, the reinforcement learning-driven resource scheduler enables intelligent and elastic migration and on-demand expansion of resources between core and edge nodes, effectively coping with sudden loads and significantly improving resource utilization and overall system stability.

[0021] The present invention is further configured such that: the DMN control center supports dynamic switching between idle mode and task mode.

[0022] When entering idle mode, if the system load, CPU utilization, and network I / O continuous preset time are all below the threshold, the following operations are triggered:

[0023] The prediction engine is invoked to analyze historical data access logs and node communication frequency to generate prediction results for future data demand.

[0024] Perform background tasks such as data cleaning, deduplication, and index optimization.

[0025] The predicted hotspot information is pushed to the small-world connection layer to pre-adjust the high-cluster connection weights of the local router;

[0026] When entering task mode, the routing optimization module is activated. Combining the real-time load status of the cross-domain scheduler in the small-world connection layer, the optimal task execution path between data service nodes is calculated using a graph traversal algorithm.

[0027] By adopting the above technical solution, the switching trigger conditions and core operations of the DMN control center in idle mode and task mode are defined. In idle mode, the prediction engine is called to analyze historical logs to generate future demand predictions, perform background tasks such as data cleaning and index optimization, and push hotspot information to the connection layer for pre-adjusting local router weights. In task mode, the routing optimization module is enabled to calculate the optimal path through graph traversal algorithm in combination with real-time load. The technical effect is to simulate the background thinking function of the network in the default mode of the brain, actively use idle resources to perform predictive analysis and optimization reserve value in idle time, and quickly switch to high-efficiency execution state when a task arrives, which greatly improves the intelligence of the system, resource utilization and task response speed.

[0028] The present invention is further configured such that the DMN control center triggers the following background task in idle mode:

[0029] Perform data cleaning, deduplication, and normalization.

[0030] Perform index and metadata optimization;

[0031] Based on knowledge graphs, cross-domain data relationships are integrated to form high-level knowledge representations to support complex queries;

[0032] Perform predictive data prefetching and cache frequently accessed data to hot zone nodes;

[0033] In idle mode, the DMN control center prefetches data that may be accessed in the future to neighboring data nodes and pre-allocates computing resources based on the output of the predictive analysis module.

[0034] By adopting the above technical solution, the specific background tasks performed by DMN in idle mode are further refined, including performing data cleaning, deduplication and normalization processing, indexing and metadata optimization, integrating cross-domain data relationships based on knowledge graphs to form high-level knowledge representations to support complex queries, performing predictive data prefetching to cache frequently accessed data to hot zone nodes, and prefetching data that may be accessed in the future to neighboring nodes and pre-allocating computing resources according to the output of the predictive analysis module. The technical effect is that it makes full use of the low load period of the system to perform key maintenance and optimization work. In particular, by building and updating the knowledge graph to integrate cross-domain relationships and predictive prefetching and pre-allocation, it significantly improves the response capability of subsequent complex queries, data access speed and resource readiness, effectively reduces user request latency and improves system throughput.

[0035] The present invention is further configured such that the prediction engine employs a dual-model fusion algorithm:

[0036] For short-term predictions, when the time frame is less than the preset time, an LSTM model is used to capture burst access patterns based on time series data.

[0037] For long-term forecasts exceeding the preset time, the Transformer model is used, combined with the topological relationships of the small-world connection layer to explore the need for cross-cluster data association.

[0038] By adopting the above technical solution, it is clarified that the prediction engine uses a dual-model fusion algorithm: when the short-term prediction is less than the preset time, the LSTM model is used to capture burst access patterns based on time-series data; when the long-term prediction is greater than the preset time, the Transformer model is used to mine cross-cluster data association needs by combining the topological relationships of the small-world connection layer. The technical effect is that the most suitable model is selected according to different prediction time scales. LSTM is good at capturing short-term burst traffic such as data peaks from edge devices, while Transformer can effectively analyze long-term dependencies and mine cross-business domain data associations. This targeted strategy integrates the advantages of the two models, which significantly improves the prediction accuracy and provides more reliable support for resource pre-allocation, data prefetching, and network optimization.

[0039] The present invention is further configured such that: the small-world connection layer includes a local router and a cross-domain scheduler, which respectively handle same-cluster and cross-domain requests;

[0040] The local router processes local data requests from data service nodes within the same cluster based on high-level clustering connections.

[0041] The cross-domain scheduler forwards cross-cluster data requests through long-range connections. When a topology update is triggered, the long-range connection topology of the cross-domain scheduler is adjusted first to ensure that the number of hops in the cross-domain request path is stable and does not exceed the preset number of hops.

[0042] The topology update is triggered by the following events:

[0043] The DMN control center outputs data on changes in access hot zones in predictive mode;

[0044] The number of consecutive long-range connection failures of the cross-domain scheduler is greater than or greater than the preset number;

[0045] The local router's high clustered connection load rate remains above a preset percentage for a preset period of time.

[0046] By adopting the above technical solution, the internal components of the small-world connection layer are defined: the local router handles same-cluster requests based on high-cluster connections, and the cross-domain scheduler handles cross-cluster requests through long-range connections. Its optimization mechanism prioritizes adjusting the cross-domain scheduler topology to ensure that the number of hops in cross-domain request paths remains stable and does not exceed a preset number. Specific events that trigger topology updates are specified, such as changes in hotspots in DMN prediction output, consecutive failures of long-range connections in the cross-domain scheduler exceeding a preset number, and the local router's high-cluster connection load rate continuously exceeding limits. The technical effect is to achieve a clear division of routing responsibilities and prioritize efficient and stable cross-domain paths through event-driven dynamic topology updates. This ensures rapid and accurate optimization of the network structure during hotspot migration, connection failures, or local congestion, maintaining low latency and high reliability in cross-domain communication, and improving the service quality and robustness of the system in dynamic environments.

[0047] A brain-inspired data service system approach includes the following steps:

[0048] Layered initialization steps: Deploy containerized data nodes in the infrastructure layer, generate the initial topology through the small-world network manager in the small-world connection layer, and start full-link monitoring through the status monitoring engine in the intelligent control layer;

[0049] Cross-layer collaborative processing steps: When the external access layer receives a cross-department query request, the DMN control center of the intelligent control layer calls the pre-calculated results of the predictive analysis module, and directly reaches the sales and finance gateway node through the long-distance cross-domain connection of the small world connection layer. The edge node synchronously provides real-time device data.

[0050] Layered adaptive optimization steps: The intelligent control layer triggers the small-world connection layer to adjust the distribution of long-range connections based on the state monitoring data, and at the same time guides the infrastructure layer to reallocate container resources.

[0051] By adopting the above technical solution, this application describes a layered method for applying the above system, which includes three core steps: a layered initialization step to deploy containerized nodes and generate an initial topology to start monitoring; a cross-layer collaborative processing step where, when the access layer receives a cross-departmental request, the intelligent control layer calls the pre-calculated results to directly reach the target node, such as the sales and finance gateway node, through a long-range connection and integrates real-time edge data; and a layered adaptive optimization step where the intelligent control layer triggers the connection layer to adjust the topology, such as the distribution of long-range connections, and guides the infrastructure layer to reallocate container resources based on monitoring data. The technical effect is that, through layer division and collaborative processes, the system can be quickly deployed and started, efficiently handle complex cross-domain requests, and dynamically trigger the linkage optimization of each layer to form an end-to-end closed-loop self-optimization capability, ensuring that the system continuously adapts to changes in business needs and data load fluctuations, and improving the overall agility, efficiency, and intelligence level.

[0052] The present invention is further configured such that the layered adaptive optimization includes: a small-world connection layer: when the communication between production data nodes and R&D data nodes continues for a preset number of days exceeding the system average, a new long-range connection is established between the two through the small-world network manager, and the number of hops in the communication path is shortened; an intelligent control layer: based on historical data, the predictive analysis module pre-allocates a preset proportion of database connection resources for the monthly financial settlement task; and an infrastructure layer: when the resource scheduler detects that the CPU utilization of the edge node exceeds a preset percentage, it automatically starts several new instances in the containerized platform for load balancing.

[0053] By adopting the above technical solutions, specific operational examples in the layered adaptive optimization steps are provided. For example, when the communication between production and R&D nodes exceeds the system average preset number of days in the small-world connection layer, the number of hops is significantly reduced by adding direct long-distance connections through the manager, such as reducing the number of hops from 5 to 1. The intelligent control layer predictive analysis module pre-allocates a preset ratio, such as 50%, of database connection resources for monthly financial settlement tasks based on historical data. When the infrastructure layer resource scheduler detects that the CPU utilization of edge nodes exceeds a preset percentage, it automatically starts new container instances for load balancing. The technical effect is to provide a specific practical solution for layered optimization. By automatically establishing critical connections, the path is greatly shortened. Resources are reserved in advance based on prediction to avoid contention. And elastic scaling is used to deal with bottlenecks. Together, they significantly reduce critical path latency, prevent performance degradation, and ensure the stability and scalability of services under load pressure.

[0054] The present invention is further configured such that: the intelligent control layer also includes a learning and adaptation engine, used for: analyzing historical access patterns and updating the access frequency model; evaluating task execution efficiency and optimizing resource scheduling strategies; adjusting LSTM / Transformer model parameters based on real-time feedback; and synchronously updating the knowledge base to reflect the latest data relationships.

[0055] The learning and adaptation engine updates the access pattern model based on task execution logs to optimize the long-range connection distribution of the small-world connection layer.

[0056] When the knowledge relationships in the knowledge base change, the index reconstruction and metadata update of the data nodes are triggered, and the pre-calculation query strategy is adjusted synchronously.

[0057] By adopting the above technical solution, the role of the learning and adaptation engine in the intelligent control layer is emphasized. It is used to analyze historical access patterns and update the frequency model, evaluate task execution efficiency and optimize resource scheduling strategies, adjust LSTM or Transformer model parameters based on real-time feedback, and synchronously update the knowledge base to reflect the latest data relationships, and correlate its impact on the connection layer and data nodes. That is, the learning engine updates the access pattern model based on task logs to optimize the small-world long-range connection distribution. When the knowledge base relationship changes, it triggers the index reconstruction and metadata update of data nodes and synchronously adjusts the pre-computation query strategy. Its technical effect is to give the system the ability to continuously evolve. By continuously learning historical patterns and feedback, it automatically optimizes core components such as updating model parameters, improving scheduling strategies, and adjusting network connections. When knowledge or relationships change, it intelligently triggers downstream synchronous updates, enabling the system to continuously adapt to new data patterns, business needs and performance goals, and maintain long-term operating efficiency and intelligent service levels.

[0058] The present invention has significant technical effects due to the adoption of the above technical solutions. Attached Figure Description

[0059] Figure 1 This is the overall system architecture diagram;

[0060] Figure 2 This is a small-world network topology diagram;

[0061] Figure 3 This is a flowchart of the internal workflow of DMN. Detailed Implementation

[0062] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0063] Example 1:

[0064] In existing technologies, with the widespread application of big data and cloud computing technologies, enterprise data service systems face problems such as difficulty in cross-departmental data integration, low resource utilization, and response delays. Traditional architectures rely on centralized nodes to process requests, resulting in excessive load on core nodes and inability to process edge data in a timely manner. Existing systems cannot effectively utilize resources for preprocessing during idle periods, and cross-domain queries require multiple jumps, increasing communication latency. For example, if a manufacturing company's production data and financial data belong to different server clusters, manual coordination of data interfaces is required when generating comprehensive reports, causing response times to exceed the business tolerance threshold.

[0065] The inventors observed that biological nervous systems possess adaptive connectivity and idle-period self-optimization characteristics. They proposed a working mechanism that simulates the default mode network of the brain. First, they analyzed the spatiotemporal distribution characteristics of cross-domain data access and found that local high-frequency communication and cross-domain low-frequency access coexist. Second, they studied the high clustering and short-path characteristics of small-world networks and attempted to apply them to the data node connection topology. They further designed a hierarchical control architecture to perform predictive resource pre-allocation during idle periods, reducing response latency when tasks are triggered. By forming a dynamic connection network between edge nodes and core nodes, they achieved direct routing for cross-domain requests.

[0066] This application proposes a brain-inspired data service system, comprising a data node cluster, a small-world connection layer, an intelligent control layer, an external access layer, and an infrastructure layer. The data node cluster consists of distributed servers, containers, microservices, etc., covering core nodes, edge nodes, and gateway nodes. The small-world connection layer adopts a topology combining high clustering and long-range connections. The intelligent control layer includes modules for status monitoring, predictive analysis, and resource scheduling. The external access layer provides standardized interfaces, and the infrastructure layer is deployed using a hybrid architecture.

[0067] Data node clusters refer to distributed computing units divided according to business needs. They can be rapidly deployed using containerization technology to handle data service needs from different business domains. The small-world connection layer refers to connection topology building modules based on complex network theory. It maintains communication efficiency within the business domain through local dense connections and achieves cross-domain direct access through random long-range connections. The intelligent control layer refers to the central control system that simulates the default mode of the brain. It drives system optimization through real-time monitoring and predictive analysis. The infrastructure layer refers to the combination of physical and virtualized resources that support system operation. For example, core nodes are deployed on cloud servers, and edge nodes are deployed on industrial gateway devices.

[0068] During system operation, the hybrid architecture of the infrastructure layer provides elastic resource support. When the external access layer receives user requests, the small-world connection layer selects local routing or cross-domain scheduling according to the request type. The intelligent control layer continuously monitors the node status and performs data preprocessing and resource pre-allocation during idle periods. For example, when frequent interaction between production nodes and quality nodes is detected, long-distance connection channels are automatically established. When the end-of-month financial settlement task is triggered, the pre-allocated database connection resources can be put into use immediately to avoid resource contention.

[0069] This solution achieves decoupling of control and execution through a layered architecture, avoiding bottlenecks from a single control node. The small-world connection topology reduces cross-domain hops while maintaining local efficiency, and reduces redundant links compared to the traditional fully connected mode. The predictive resource scheduling mechanism of the intelligent control layer transforms the traditional post-event response into pre-event preparation, effectively utilizing idle system resources.

[0070] This application achieves path optimization for cross-domain data services, shortens the average response time of cross-department queries to an acceptable range for business operations, improves the efficiency of handling sudden tasks through a dynamic resource pre-allocation mechanism, increases the utilization rate of edge node resources, and automatically executes data optimization tasks during idle periods to avoid wasting computing resources and form a continuous self-optimizing service capability.

[0071] The small-world network manager of the small-world connection layer initializes the network topology based on a variant of the Watts-Strogatz model algorithm, introduces long-range cross-domain connections through random reconnection edges, and analyzes node communication logs at preset intervals. When the frequency of communication between cross-domain nodes exceeds a threshold, long-range connections are dynamically inserted through the service mesh to reduce the number of hops in the communication path.

[0072] The Watts-Strogatz model variant algorithm refers to a topology generation algorithm based on small-world network theory. Specifically, it can introduce long-range connections into the initial highly clustered network by using a random reconnection edge mechanism. By adjusting the reconnection probability parameter, it balances the local connection density and global connectivity efficiency. Random reconnection edge refers to the probabilistic replacement of some connections in the initial ring topology. For example, local connections are replaced with cross-domain connections with a preset probability, thereby forming a network structure that has both high clustering and low path length. The service mesh dynamic insertion of long-range connections refers to the real-time adjustment of communication links between nodes through service mesh technology. For example, when the frequency of cross-domain communication exceeds a set threshold, a direct connection channel is automatically established between the corresponding nodes.

[0073] During the network initialization phase, an improved Watts-Strogatz algorithm is used to construct the basic topology. A random reconnection mechanism is used to introduce cross-domain connections while preserving the local high clustering characteristics. The communication logs of cross-domain nodes are analyzed periodically, such as counting the frequency of interactions between nodes on an hourly or daily basis. When the communication frequency of a specific pair of nodes exceeds a preset threshold, the dynamic link configuration function of the service mesh is triggered. At this time, the system automatically establishes long-range connections between the corresponding nodes, bypassing the original intermediate nodes to form a direct path, thereby reducing the number of hops in cross-domain communication from multiple hops in the original topology to a single hop.

[0074] Traditional systems typically employ fixed topologies or simple distributed architectures, with cross-domain communication paths relying on statically configured intermediate nodes. These paths cannot be dynamically optimized based on actual communication needs. In contrast, this solution utilizes periodic communication analysis and a dynamic service mesh adjustment mechanism to automatically identify high-frequency cross-domain interaction scenarios and optimize connection topologies, effectively addressing the issues of high latency and path redundancy in traditional architectures.

[0075] This application can automatically optimize cross-domain communication paths according to actual business needs, significantly reduce data transmission latency in high-frequency cross-domain interaction scenarios, reduce intermediate forwarding links by dynamically inserting long-range connections, avoid bandwidth resource waste caused by path redundancy, and improve the response efficiency of cross-domain data services.

[0076] The DMN control center of the intelligent control layer works collaboratively through the following modules: the status monitoring engine collects data node CPU load, network throughput, and container resource utilization indicators in real time and generates a system heat map; the predictive analysis module is equipped with a prediction engine that uses an LSTM model to predict the real-time data collection peak of edge nodes and uses a Transformer model to analyze cross-domain query patterns; the resource scheduler dynamically adjusts the resource quotas of core nodes and edge nodes based on reinforcement learning algorithms, and automatically migrates the number of CPU cores or memory quotas based on the containerized platform when the data collection volume of edge nodes suddenly increases, dynamically expanding the computing resources to the edge nodes according to 120% of the current load of the edge nodes.

[0077] The status monitoring engine is a module used to continuously collect system operation metrics. Specifically, it can be implemented using the Prometheus monitoring framework combined with a custom metric collector. It periodically acquires resource consumption data of each node through exposed API interfaces. The LSTM model in the predictive analysis module refers to a Long Short-Term Memory Neural Network, which can be implemented using the TensorFlow framework to build a three-layer network structure to handle periodic fluctuations in time series data. The Transformer model refers to a deep learning architecture based on a self-attention mechanism, which can be implemented using pre-trained models fine-tuned from the HuggingFace library to capture association patterns in cross-domain queries. The reinforcement learning algorithm in the resource scheduler refers to a dynamic decision-making method based on environmental feedback, which can be implemented using the Q-learning algorithm combined with the API interface of a container orchestration platform to optimize resource allocation strategies through a reward function.

[0078] The status monitoring engine periodically collects operational metrics from each node and generates a visual heatmap. The color depth of the heatmap reflects the resource consumption intensity of different areas. The predictive analysis module uses the historical load curves of edge nodes in the heatmap to call the LSTM model to predict the data collection peak within the next five minutes. At the same time, it uses the Transformer model to parse the semantic relationships in cross-department query requests. When a sudden surge in traffic occurs at an edge node, the resource scheduler calculates the optimal scaling solution based on reinforcement learning strategies. For example, in a containerized platform, the number of CPU cores of an edge node can be increased from 4 to 5, and the memory quota can be increased from 8GB to 9.6GB. The scaling range precisely matches the 120% increase in the current load.

[0079] Traditional systems typically employ static threshold alarm mechanisms for resource scheduling, which are inadequate for handling sudden surges in traffic. This solution, however, utilizes a dual-model architecture integrating time-series prediction and semantic analysis to anticipate changing resource demands. Combined with reinforcement learning algorithms, it enables precise elastic scaling up and down, avoiding response delays caused by manual intervention. In edge computing scenarios, this application can automatically identify sudden changes in data collection volume and adjust computing resource configurations promptly to prevent node overload. When edge nodes in a logistics and warehousing system experience a surge in order data during promotional periods, the system can automatically increase processing capacity above a preset safety threshold, ensuring the integrity and timeliness of real-time data collection while preventing excessive resource consumption of core nodes from impacting other business operations.

[0080] The DMN control center supports dynamic switching between idle and task modes: When entering idle mode, if the system load, CPU utilization, and network I / O continuous preset time are below the threshold, the following operations are triggered: call the prediction engine to analyze historical data access logs and node communication frequency to generate future data demand prediction results; perform background tasks such as data cleaning, deduplication, and index optimization; push the predicted hot zone information to the small-world connection layer for pre-adjusting the high-clustering connection weights of the local router; when entering task mode, enable the routing optimization module, and calculate the optimal task execution path between data service nodes through graph traversal algorithm, based on the real-time load status of the cross-domain scheduler in the small-world connection layer.

[0081] Idle mode refers to a low-power operating state that is automatically triggered when the system resource utilization rate is lower than a preset threshold. Specifically, it can be implemented using a load monitoring algorithm based on a sliding window, used to execute background optimization tasks during off-peak hours. Task mode refers to a high-performance operating state that is activated when the system responds to real-time requests. It is implemented through a dynamic path planning algorithm, such as an improved version of the Dijkstra algorithm to calculate the optimal task path. The prediction engine refers to a data access prediction module based on a machine learning model. It can be implemented using a fusion architecture of LSTM neural network and Transformer model, used to identify periodic access patterns and cross-domain association requirements. Data prefetching refers to a caching mechanism that preloads high-frequency access data to the target node based on the prediction results. It is implemented through a distributed cache management protocol, such as using an LRU cache replacement strategy in conjunction with a prefetch queue.

[0082] When the system detects that CPU utilization and network traffic metrics have been continuously below the set threshold for a preset duration, it automatically switches to idle mode. At this time, the prediction engine performs time series analysis on historical access logs to identify sets of data objects that may be accessed frequently in the future. The data cleaning task marks and merges redundant data in the storage nodes. The index optimization module reconstructs the B+ tree index structure to improve query efficiency. Predictive hot zone information is pushed to the local router through a message queue so that it can adjust the connection weight allocation strategy in advance. When an external request arrives, the system immediately switches to task mode. The routing optimization module uses a breadth-first search algorithm to traverse the topology graph based on the current load status of the cross-domain scheduler and selects the task execution path with the fewest hops and balanced load.

[0083] Traditional systems only maintain basic operation during idle periods, failing to effectively utilize idle resources to perform data optimization tasks, resulting in resource waste and response delays. Existing mode switching mechanisms mostly rely on manual configuration strategies and cannot dynamically adjust the operating status according to real-time load. This solution uses automated threshold monitoring and prediction-driven background task execution to proactively complete data preprocessing and resource pre-allocation during system idle periods, significantly reducing service response time in task mode.

[0084] This application achieves a dual improvement in system resource utilization and response efficiency. Index optimization executed in idle mode improves cross-domain query speed, data prefetching mechanism reduces the acquisition latency of frequently accessed data, dynamic path planning in task mode effectively avoids network congestion, ensures stable and controllable end-to-end processing time for cross-department data requests, and the system can autonomously switch operating modes according to real-time load status, reducing energy consumption and idle computing resources while ensuring service quality.

[0085] In idle mode, the following background tasks are triggered: perform data cleaning, deduplication and normalization; optimize indexes and metadata; integrate cross-domain data relationships based on knowledge graphs to form high-level knowledge representations to support complex queries; perform predictive data prefetching to cache frequently accessed data to hot zone nodes; and prefetch data that may be accessed in the future to neighboring data nodes and pre-allocate computing resources based on the output of the predictive analysis module.

[0086] Data cleaning refers to automatically identifying and correcting errors or redundant information in data through preset rules. This can be achieved by combining regular expression matching with machine learning models to improve data quality and consistency. Indexing and metadata optimization refers to reconstructing the database index structure. This can be achieved through B+ tree balancing algorithms or columnar storage optimization techniques to accelerate query response speed. Knowledge graph integration of cross-domain data relationships refers to building an entity relationship network through graph databases. This can be achieved using Neo4j or JanusGraph to discover potential data connections across business domains. Predictive data prefetching refers to loading data in advance based on access pattern prediction results. This can be achieved by combining LRU cache replacement algorithms with time series prediction models to reduce subsequent query latency.

[0087] When the system enters idle mode, the background task execution process is automatically started. The data cleaning module traverses the storage nodes, identifies abnormal data formats through the rule engine and performs standardization processing, the index optimization engine analyzes the query logs and reconstructs the index structure of frequently accessed fields, the knowledge graph engine synchronously updates cross-domain entity relationships and establishes new semantic association paths, the predictive analysis module generates prefetch instructions based on historical access patterns, distributes predictive data copies to the cache of the target nodes, and the pre-allocation of computing resources between neighboring nodes is achieved by dynamically adjusting resource quotas through the container orchestration platform to ensure that the prefetched data has the ability to be processed in real time.

[0088] Traditional systems only maintain basic operation during idle periods, failing to effectively utilize computing resources for data optimization. Existing data preprocessing technologies require manual triggering and lack cross-domain correlation analysis, making it impossible to form a global optimization strategy. This solution achieves collaborative optimization of system resources through automated background tasks, completing data quality improvement and access path pre-optimization during off-peak periods, overcoming the lag defects of traditional responsive processing models.

[0089] This application fully utilizes computing resources during system idle periods, effectively improving data query efficiency and accuracy. Continuous maintenance of cross-domain data relationships enhances the response capability of complex queries. The predictive data prefetching mechanism significantly reduces service latency in high-concurrency scenarios. The pre-allocation computing resource strategy ensures the availability of data prefetching, forming a complete offline optimization and online service collaboration mechanism.

[0090] The prediction engine employs a dual-model fusion algorithm. For short-term predictions (less than a preset timeframe), an LSTM model is used to capture sudden access patterns based on time-series data. For long-term predictions (more than a preset timeframe), a Transformer model is used, combining the topological relationships of small-world connection layers to mine cross-cluster data association needs. The LSTM model, or Long Short-Term Memory Neural Network, can be implemented using a recurrent neural network with forget gates, input gates, and output gates to capture short-term dependencies and sudden access patterns in time-series data. The Transformer model, a deep learning model based on self-attention mechanisms, can be implemented using an encoder structure composed of multi-head attention layers and feedforward neural network layers to handle long-distance dependencies and cross-cluster data association needs. The dual-model fusion algorithm is a technical solution that assigns prediction tasks at different time scales to different models. This can be achieved by setting preset time thresholds to classify prediction task types and establishing a model switching mechanism, thus balancing short-term fluctuations with long-term association patterns.

[0091] When the prediction time range is below the preset time threshold, the system automatically calls the LSTM model to process the time series data. This model processes time series features layer by layer through a recurrent neural network structure and uses memory units to capture short-term fluctuation patterns of indicators such as the amount of data collected by device sensors and the frequency of user queries. When the prediction time range exceeds the preset time threshold, the system switches to the Transformer model, which analyzes the topological connection relationship across cluster nodes through a self-attention mechanism to identify potential data association needs between different business domains. The prediction results of the two models are used to generate the final output through a weighted fusion module, which is used to guide resource pre-allocation and data prefetching strategies.

[0092] Traditional forecasting methods typically use a single model to handle forecasting tasks at different time scales, resulting in short-term forecasts ignoring cross-domain correlation features and long-term forecasts failing to capture sudden fluctuations. This solution addresses the problem of insufficient adaptability of a single model by employing a dual-model division of labor mechanism. This maintains the sensitivity of short-term forecasts while effectively utilizing topological relationships to enhance the correlation analysis capabilities of long-term forecasts.

[0093] This application can automatically select the optimal model based on the time characteristics of the prediction task, improve the resource utilization of edge nodes and the efficiency of cross-domain data scheduling. The short-term prediction module accurately captures the peak of real-time data collection, avoiding resource shortages caused by sudden loads. The long-term prediction module optimizes the pre-allocation of cross-cluster resources through topological relationship analysis, reducing cross-domain query latency.

[0094] The small-world connection layer includes a local router and a cross-domain scheduler, which handle same-cluster and cross-domain requests respectively. The local router processes local data requests from data service nodes within the same cluster based on high-cluster connection relationships. The cross-domain scheduler forwards cross-cluster data requests through long-range connections. When a topology update is triggered, the long-range connection topology of the cross-domain scheduler is adjusted first to ensure that the number of hops in the cross-domain request path is stable and does not exceed the preset number of hops. The topology update is triggered by the following events: changes in the data access hotspot output by the DMN control center in predictive mode; the number of consecutive long-range connection failures of the cross-domain scheduler is greater than or equal to the preset number; and the high-cluster connection load rate of the local router is continuously greater than the preset proportion within a preset time.

[0095] A local router is a communication control unit deployed within a business domain. It can be implemented using routing devices based on the OSPF protocol. By maintaining high-cluster connectivity, it enables low-latency communication between nodes within the domain, reducing transmission latency for data interaction within the same business domain. A cross-domain scheduler is a relay device for cross-business domain communication. It can be implemented using a service mesh component extended with the BGP protocol. By dynamically maintaining long-range connections, it shortens cross-domain communication paths, addressing the problem of low efficiency in cross-domain access caused by data silos. Topology update triggering events are monitoring mechanisms for changes in network status. Specifically, they can use a sliding window algorithm to count connection failures and time-series analysis to detect abnormal load rates, enabling adaptive adjustments to the network structure to maintain service quality.

[0096] Local routers form an intra-domain communication backbone by maintaining high-cluster connections. When a request from the same cluster is received, it is transmitted directly through a pre-established physical link. The cross-domain scheduler continuously monitors the status of long-range connections. When it detects that the number of hops in the cross-domain communication path exceeds a threshold, it prioritizes triggering the topology adjustment operation of the cross-domain scheduler. For example, when the DMN control center predicts the migration of sales data hotspots, the cross-domain scheduler automatically establishes a new long-range connection between the sales domain and the financial domain, while removing inefficient and redundant connections. The local router and the cross-domain scheduler form a hierarchical control architecture, which optimizes cross-domain path selection while maintaining intra-domain communication efficiency.

[0097] Traditional data systems employ static routing strategies, which lead to a decrease in cross-domain communication efficiency as business fluctuates. In contrast, this solution achieves continuous optimization of cross-domain paths through a dynamic topology update mechanism. Existing technologies typically require manual intervention to adjust cross-domain connections. This solution achieves automated network reconstruction based on preset trigger conditions, effectively addressing sudden cross-domain access demands.

[0098] This application can automatically identify high-load communication paths and implement targeted optimizations, significantly reducing communication latency in cross-domain data access scenarios. While ensuring the efficiency of intra-domain communication through a layered processing mechanism, it can effectively avoid network congestion by dynamically adjusting the cross-domain connection topology, improve the response stability of cross-departmental data services, and the automated update mechanism based on preset trigger conditions can promptly eliminate the impact of single-point failures on the overall system performance, ensuring the continuous and efficient operation of the data service system in complex business scenarios.

[0099] The brain-inspired data service system approach includes a layered initialization step, a cross-layer collaborative processing step, and a layered adaptive optimization step. The layered initialization step deploys containerized data nodes at the infrastructure layer, generates an initial topology through the small-world network manager of the small-world connection layer, and initiates full-link monitoring through the status monitoring engine of the intelligent control layer. In the cross-layer collaborative processing step, when an external access layer receives a cross-department query request, the DMN control center of the intelligent control layer calls the pre-calculated results from the predictive analytics module, directly connecting to the sales and finance gateway node via a long-range cross-domain connection through the small-world connection layer, with edge nodes synchronously providing real-time device data. The layered adaptive optimization step involves the intelligent control layer triggering the small-world connection layer to adjust the distribution of long-range connections based on status monitoring data, while simultaneously guiding the infrastructure layer to reallocate container resources.

[0100] The layered initialization step refers to the process of establishing the system infrastructure. Specifically, it can be achieved by deploying data nodes using container orchestration tools, such as managing node instances in different locations through a Kubernetes cluster. This step lays the foundation for dynamic resource scheduling. The cross-layer collaborative processing step refers to the mechanism by which multi-level components collaborate to respond to complex requests. Specifically, it can be achieved through service mesh technology to realize cross-domain connection routing. This mechanism effectively solves the problem of latency in cross-departmental data access in traditional systems. The layered adaptive optimization step refers to the process of continuous self-adjustment of the system. Specifically, it can be achieved by using reinforcement learning algorithms to dynamically adjust resource quotas. This process achieves a dual improvement in resource utilization and service quality.

[0101] During system startup, the containerized platform can be configured to deploy core nodes in the cloud and lightweight container instances on edge devices. When a cross-domain query is received, the predictive analysis module can predict the target node based on historical access patterns. For example, it can directly route financial data requests to a preset gateway container instance. During operation, the status monitoring engine can periodically collect network latency data. When it detects that the number of communication hops between specific nodes exceeds the limit, it triggers the horizontal scaling operation of the container instance. This layered collaboration mechanism ensures that the data processing path is always in the optimal state, while avoiding system oscillations caused by adjustments at a single level.

[0102] Traditional centralized systems require forwarding requests layer by layer when querying across departments. This method, however, achieves direct cross-domain routing through pre-calculated paths, reducing intermediate steps. The static resource allocation mode in existing technologies cannot cope with sudden loads, while this method achieves elastic scaling of computing resources through layered adaptive optimization. Conventional distributed systems operate independently at each level, while this method achieves overall collaborative optimization through cross-level transmission of status monitoring data.

[0103] This application enables end-to-end direct processing of cross-domain data requests, effectively reducing query response latency. The dynamic resource reallocation mechanism ensures the service stability of edge nodes under sudden loads, avoiding data processing interruptions due to insufficient resources. Multi-level collaborative optimization enables the system to automatically adapt to changes in business needs and improve the processing efficiency of complex query tasks.

[0104] The specific implementation of the hierarchical adaptive optimization steps includes a collaborative optimization mechanism of the small-world connection layer, intelligent control layer, and infrastructure layer. When the communication between production data nodes and R&D data nodes continues for a preset number of days exceeding the system average, a new long-range connection is established between the two through the small-world network manager, and the number of communication path hops is shortened. The predictive analysis module pre-allocates a preset proportion of database connection resources for monthly financial settlement tasks based on historical data. When the resource scheduler detects that the CPU utilization of edge nodes exceeds a preset percentage, it automatically starts several new instances in the containerized platform for load balancing.

[0105] Layered adaptive optimization refers to the system solving cross-domain communication and resource allocation problems through coordinated adjustments at different levels. Specifically, it can be implemented by dynamically adjusting connection topology and resource allocation algorithms. The small-world connection layer optimizes communication paths by adding new cross-domain long-range connections. For example, when the communication frequency between nodes exceeds a threshold, a topology update is triggered. The intelligent control layer responds to periodic task requirements through predictive resource pre-allocation. For example, database connection resources are pre-allocated based on historical access patterns. The infrastructure layer achieves load balancing through the dynamic scaling of containerized instances. For example, when the resource utilization of edge nodes reaches a preset threshold, a new instance is automatically launched.

[0106] When the communication frequency between production and R&D data nodes is consistently higher than the system average, the Small World Network Manager automatically establishes long-distance cross-domain connections, shortening the communication path that originally required forwarding through multiple intermediate nodes into a direct link. The intelligent control layer analyzes historical task execution logs and pre-allocates database connection pool resources before the monthly financial settlement task starts, avoiding resource contention during peak task periods. The infrastructure layer's resource scheduler monitors the load of edge nodes in real time, and when it detects that the CPU utilization exceeds a preset threshold, it automatically starts a new instance in the containerized platform, migrating some computing tasks to the newly added instance for execution.

[0107] In some specific implementations, the preset number of days can be, for example, 30 days, and the system average can be dynamically calculated using a sliding window algorithm; the preset proportion of database connection resources can be, for example, 40% of the total connection pool; the preset percentage can be, for example, an 80% CPU utilization threshold; and the number of new instances can be dynamically calculated based on the current load and resource availability.

[0108] Traditional systems typically employ fixed routing strategies when optimizing cross-domain communication, making it impossible to dynamically adjust the topology based on actual communication patterns. In contrast, this solution achieves adaptive optimization of cross-domain connections through a hierarchical collaborative mechanism. Existing resource allocation methods often employ static quotas or passive responsive scaling. This solution achieves proactive resource provisioning through predictive pre-allocation and dynamic instance creation. Existing load balancing technologies rely on fixed thresholds to trigger scaling. This solution combines real-time monitoring and containerization technologies to achieve elastic resource scheduling.

[0109] This application effectively reduces the number of path hops and transmission latency in cross-domain communication, improves the resource supply efficiency of periodic tasks, enhances the dynamic load processing capability in edge computing scenarios, optimizes communication paths to reduce the latency of cross-departmental data interaction, prevents resource bottlenecks during task execution through predictive resource pre-allocation, and ensures the stability of the system under high load scenarios through the dynamic scaling of containerized instances.

[0110] The intelligent control layer also includes a learning and adaptation engine, which analyzes historical access patterns and updates the access frequency model, evaluates task execution efficiency and optimizes resource scheduling strategies, adjusts LSTM or Transformer model parameters based on real-time feedback, and synchronously updates the knowledge base to reflect the latest data relationships. The learning and adaptation engine updates the access pattern model based on task execution logs, optimizes the long-range connection distribution of the small-world connection layer, and triggers index reconstruction and metadata updates of data nodes when knowledge relationships in the knowledge base change, while synchronously adjusting the pre-computation query strategy.

[0111] The learning and adaptation engine refers to a computing module with autonomous learning and dynamic adjustment capabilities. Specifically, it can be implemented using a machine learning framework combined with a rule engine. It optimizes strategies by continuously analyzing system operation data. The access frequency model refers to a probability distribution model describing the data access patterns. Specifically, it can be constructed using a hidden Markov model or time series analysis algorithm to predict future data access hotspots. The resource scheduling strategy refers to the decision rules for dynamically allocating computing resources. Specifically, it can be implemented using a reinforcement learning algorithm combined with a queuing theory model. It adjusts resource quotas based on real-time load status. The knowledge base refers to a structured storage system that stores data relationships. Specifically, it can use a graph database or triplet storage architecture and maintain cross-domain data relationships through knowledge graph technology.

[0112] The learning and adaptation engine continuously collects task execution logs and system operation metrics, and updates the access pattern model through a combination of offline training and online inference. When a significant change in the access frequency of a specific data node is detected, long-range connection optimization suggestions are automatically generated and pushed to the small-world connection layer. When a knowledge base update event is triggered, the system automatically starts the index reconstruction process, recalculates the association weights between data nodes, and generates an optimized query execution plan based on the new associations. The model parameter adjustment process adopts an online learning mechanism, combining real-time feedback data with historical training data, and iteratively updates the weight parameters of the prediction model through the gradient descent algorithm.

[0113] Traditional systems lack a continuous learning mechanism and cannot dynamically optimize resource allocation strategies based on operational data. This solution, however, introduces a learning and adaptation engine to achieve automatic tuning of system parameters and real-time synchronization of the knowledge base. Existing technologies often employ static resource scheduling strategies, which are difficult to adapt to dynamic load changes. This solution, on the other hand, achieves dynamic optimization of resource allocation through reinforcement learning algorithms.

[0114] This application can effectively solve the problems of low resource utilization and poor cross-domain query efficiency in traditional data service systems. Through a self-learning mechanism, it continuously optimizes system configuration and reduces manual maintenance costs. The real-time synchronization mechanism of the knowledge base can eliminate data silos and improve the response speed of cross-domain data association queries. The dynamic model parameter adjustment function enhances the system's adaptability to sudden access patterns and ensures that prediction accuracy continues to improve over time.

Claims

1. A brain-inspired data service system, characterized in that, include: Data node cluster: Composed of several distributed data nodes, used to store, process, analyze data or provide data services. The data nodes are at least one of servers, containers, microservices, logical units or AIAgents, including core nodes deployed on cloud server clusters, edge nodes deployed on edge computing devices and gateway nodes deployed on containerized platforms, respectively corresponding to the core data processing, edge data collection and cross-domain data connection needs of the business domain. Small-world connection layer: The connection topology between data nodes is constructed using a small-world network model, including a small-world network manager, local high-cluster connections, and long-range cross-domain connections. Local high-cluster connections enable dense interconnection of data nodes within the same business domain, while long-range cross-domain connections dynamically maintain cross-domain shortcuts through the small-world network manager. Intelligent Control Layer: Includes the DMN control center and its subordinate status monitoring engine, predictive analysis module, and resource scheduler, serving as the system control center to simulate the default mode network functions of the brain; External Access Layer: Provides user application access interfaces and protocol adapters; Infrastructure layer: A hybrid architecture deployed across cloud server clusters, containerized platforms, edge computing devices, and physical servers; The intelligent control layer collects communication hop count and latency metrics of the small-world connection layer in real time through the state monitoring engine, driving the predictive analysis module to generate topology optimization instructions; the DMN control center automatically triggers mode switching based on the real-time data from the state monitoring engine through a threshold comparison algorithm, and pushes hot zone weights to the small-world network manager. The DMN control center supports dynamic switching between idle mode and task mode: When entering idle mode, if the system load, CPU utilization, and network I / O continuous preset time are all below the threshold, the following operations are triggered: The prediction engine is invoked to analyze historical data access logs and node communication frequency to generate prediction results for future data demand. Perform background tasks such as data cleaning, deduplication, and index optimization. The predicted hotspot information is pushed to the small-world connection layer to pre-adjust the high-cluster connection weights of the local router; When entering task mode, the routing optimization module is activated. Combining the real-time load status of the cross-domain scheduler in the small-world connection layer, the optimal task execution path between data service nodes is calculated using a graph traversal algorithm.

2. The brain-inspired data service system according to claim 1, characterized in that, The small-world network manager of the small-world connection layer is implemented as follows: The network topology is initialized based on a variant of the Watts-Strogatz model algorithm, and long-range cross-domain connections are introduced by random reconnection edges. The node communication logs are analyzed according to a preset period. When the frequency of communication between cross-domain nodes exceeds the threshold, long-range connections are dynamically inserted through the service mesh to reduce the number of communication path hops.

3. The brain-inspired data service system according to claim 2, characterized in that, The DMN control center of the intelligent control layer works collaboratively through the following modules: Status monitoring engine: Real-time collection of CPU load, network throughput, and container resource utilization metrics of data nodes to generate system heatmaps; Predictive analytics module: Equipped with a prediction engine, it uses an LSTM model to predict the peak value of real-time data collection at edge nodes and uses a Transformer model to analyze cross-domain query patterns. Resource scheduler: Based on reinforcement learning algorithm, dynamically adjust the resource quotas of core nodes and edge nodes. When the data collection volume of edge nodes suddenly increases, automatically migrate the CPU cores / memory quota based on the containerized platform and dynamically expand the computing resources to the edge nodes according to 120% of the current load of the edge nodes.

4. A brain-inspired data service system according to claim 3, characterized in that, The DMN control center triggers the following background task in idle mode: Perform data cleaning, deduplication, and normalization. Perform index and metadata optimization; Based on knowledge graphs, cross-domain data relationships are integrated to form high-level knowledge representations to support complex queries; Perform predictive data prefetching and cache frequently accessed data to hot zone nodes; In idle mode, the DMN control center prefetches data that may be accessed in the future to neighboring data nodes and pre-allocates computing resources based on the output of the predictive analysis module.

5. A brain-inspired data service system according to claim 4, characterized in that, The prediction engine employs a dual-model fusion algorithm: For short-term predictions, when the time frame is less than the preset time, an LSTM model is used to capture burst access patterns based on time series data. For long-term forecasts exceeding the preset time, the Transformer model is used, combined with the topological relationships of the small-world connection layer to mine cross-cluster data association needs.

6. A brain-inspired data service system according to claim 3, characterized in that, The small-world connection layer includes a local router and a cross-domain scheduler, which handle same-cluster and cross-domain requests, respectively. The local router processes local data requests from data service nodes within the same cluster based on high-level clustering connections. The cross-domain scheduler forwards cross-cluster data requests through long-range connections. When a topology update is triggered, the long-range connection topology of the cross-domain scheduler is adjusted first to ensure that the number of hops in the cross-domain request path is stable and does not exceed the preset number of hops. The topology update is triggered by the following events: The DMN control center outputs data on changes in access hot zones in predictive mode; The number of consecutive long-range connection failures of the cross-domain scheduler is greater than or greater than the preset number; The local router's high clustered connection load rate remains above a preset percentage for a preset period of time.

7. A brain-inspired data service system method, applied to the brain-inspired data service system according to any one of claims 1-6, characterized in that, Includes the following steps: Layered initialization steps: Deploy containerized data nodes in the infrastructure layer, generate the initial topology through the small-world network manager in the small-world connection layer, and start full-link monitoring through the status monitoring engine in the intelligent control layer; Cross-layer collaborative processing steps: When the external access layer receives a cross-department query request, the DMN control center of the intelligent control layer calls the pre-calculated results of the predictive analysis module, and directly reaches the sales and finance gateway node through the long-distance cross-domain connection of the small world connection layer. The edge node synchronously provides real-time device data. Layered adaptive optimization steps: The intelligent control layer triggers the small-world connection layer to adjust the distribution of long-range connections based on the state monitoring data, and at the same time guides the infrastructure layer to reallocate container resources.

8. The method according to claim 7, characterized in that, The hierarchical adaptive optimization includes: Small World Connection Layer: When the communication between production data nodes and R&D data nodes continues for a preset number of days exceeding the system average, a new long-range connection is established between the two through the Small World Network Manager, and the number of hops in the communication path is shortened. Intelligent control layer: The predictive analysis module pre-allocates a preset proportion of database connection resources for monthly financial settlement tasks based on historical data; Infrastructure layer: When the resource scheduler detects that the CPU utilization of the edge node exceeds a preset percentage, it automatically starts several new instances in the containerized platform for load balancing.

9. The method according to claim 8, characterized in that, The intelligent control layer also includes a learning and adaptation engine, used for: analyzing historical access patterns and updating the access frequency model; evaluating task execution efficiency and optimizing resource scheduling strategies; adjusting LSTM / Transformer model parameters based on real-time feedback; and synchronously updating the knowledge base to reflect the latest data relationships. The learning and adaptation engine updates the access pattern model based on task execution logs to optimize the long-range connection distribution of the small-world connection layer. When the knowledge relationships in the knowledge base change, the index reconstruction and metadata update of the data nodes are triggered, and the pre-computation query strategy is adjusted synchronously.

Images