Information creation data operation and maintenance method and device

Abstract

The invention discloses a credential data operation and maintenance method and device, and belongs to the field of intelligent operation and maintenance and data management. Initiating a federal learning task among a plurality of data nodes, wherein each node trains a model locally; writing the operation and maintenance operation and the data access record into a block chain through an intelligent contract; establishing a digital twinborn model; simulation is operated on the digital twin, and data hotspots and performance bottlenecks are predicted; executing automatic optimization of data storage layout and computing resources; constructing and continuously enriching an operation and maintenance knowledge graph; a problem source is positioned through map reasoning, and a disposal scheme is recommended; continuously monitoring the system load; pre-judging resource demand change by the model; and executing the capacity expansion and contraction operation of the resource configuration to match the current service pressure. According to the method, the fundamental transformation of credential data operation and maintenance from passive response to active early warning, from experience driving to intelligent decision making, from data islanding to safety collaboration, and from resource fixation to elastic self-adaption is realized.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance and data management technology in the information technology application innovation (IT innovation) industry. In particular, it refers to a comprehensive IT innovation data operation and maintenance method and device that integrates multimodal perception, privacy computing, digital twin, knowledge graph and elastic scaling technologies. Background Technology

[0002] Against the backdrop of the large-scale deployment of the domestic IT innovation industry, ensuring the stable, secure, and efficient operation of massive amounts of data carrying critical business operations has become a core requirement. However, data operation and maintenance in the domestic IT innovation environment faces unprecedented challenges. For example, data operation and maintenance methods in this environment suffer from lagging and one-sided operational awareness: traditional operation and maintenance mainly relies on system logs and performance indicators, lacking real-time and multi-dimensional awareness of the physical environment (such as data center temperature and humidity, equipment noise), and cannot provide early warnings of hardware failures such as impending hard drive failure. Simultaneously, there is a conflict between data privacy and collaborative operation and maintenance: cross-departmental and cross-system collaborative operation and maintenance requires data support, but domestic IT innovation data often contains sensitive information, and directly sharing raw data poses a risk of leakage, restricting the efficiency and depth of collaborative operation and maintenance. Furthermore, data operation and maintenance methods in the context of domestic IT innovation suffer from serious shortcomings in operation and maintenance decision-making, fault location, and resource allocation: existing operation and maintenance are mostly "remedial after the fact," operation and maintenance decisions lack foresight, and there is a lack of tools for simulation and prediction on real systems, making it impossible to intervene and optimize before performance problems occur; moreover, the architecture of domestic IT innovation systems is complex, fault root cause location relies heavily on the experience of operation and maintenance personnel, knowledge transfer is difficult, decision support is insufficient when facing new types of faults, and fault location efficiency is low; in addition, in terms of resource allocation, system resources are usually static or semi-statically configured, which cannot quickly respond to sudden changes in business, resulting in performance degradation under high load and resource waste under low load. In summary, there is an urgent need for a next-generation intelligent data operation and maintenance solution for domestic IT innovation that can systematically address the above challenges. Summary of the Invention

[0003] To address the problems of outdated data operation and maintenance methods in the current information technology innovation environment, such as lagging dimension perception, contradiction between data privacy and collaborative operation and maintenance, lack of foresight in operation and maintenance decision-making, low efficiency in fault location, and rigid resource allocation, this invention provides an information technology innovation data operation and maintenance method and device, aiming to achieve a fundamental transformation in information technology innovation data operation and maintenance from passive response to proactive early warning, from experience-driven to intelligent decision-making, from data silos to secure collaboration, and from fixed resources to flexible self-adaptation.

[0004] The technical solution adopted by this invention to solve the technical problem is as follows:

[0005] This invention provides a data operation and maintenance method for information technology innovation, comprising the following steps:

[0006] Step S1: Intelligent multimodal perception operation and maintenance;

[0007] Collect images, sounds, and various sensor data from the domestic IT innovation environment; perform data fusion and rapid analysis at the edge to identify abnormal device status and issue immediate alarms;

[0008] Step S2: Federated Learning and Blockchain Privacy Protection Operations and Maintenance;

[0009] A federated learning task is initiated among multiple domestic IT innovation data nodes, where each node trains the model locally and exchanges parameters; operation and maintenance operations and data access records are written to the blockchain through smart contracts.

[0010] Step S3: Intelligent optimization and maintenance using digital twins;

[0011] Establish and synchronize a digital twin model of the physical data system; run simulations on the digital twin to predict data hotspots and performance bottlenecks; and perform automated optimization of data storage layout and computing resources based on the prediction results.

[0012] Step S4: Knowledge Graph-Driven Decision Making;

[0013] Build and continuously enrich the operation and maintenance knowledge graph; receive user queries, use graph reasoning to locate the root cause of the problem and recommend solutions;

[0014] Step S5: Adaptive elastic resource scheduling;

[0015] Continuously monitor system load; use predictive models to anticipate changes in resource demand; automatically execute resource allocation scaling up and down operations to match current business pressure.

[0016] According to the present invention, in step S3, the effectiveness of different operation and maintenance strategies is evaluated in advance by virtual simulation of different operation and maintenance strategies in the digital twin model, so as to achieve zero-risk operation and maintenance strategy optimization of the physical system.

[0017] This invention provides a data operation and maintenance device for domestic IT innovation, used to implement the aforementioned data operation and maintenance method for domestic IT innovation. The device mainly includes:

[0018] The intelligent multimodal perception and operation and maintenance unit is used to collect operational data of domestically developed equipment and data center environment through deployed image sensors, acoustic sensors and environmental sensors, and to perform fusion analysis of multimodal data based on edge computing nodes in order to judge the health status of equipment in real time and trigger early warnings.

[0019] The federated learning and blockchain privacy protection operation and maintenance unit is connected to the intelligent multimodal perception operation and maintenance unit. It is used to coordinate multiple participants to jointly train models through federated learning without sharing the original data, and to use blockchain to store data access logs and operation and maintenance records, and to automatically execute access control policies through smart contracts.

[0020] The digital twin intelligent optimization and operation and maintenance unit is used to build and maintain a digital twin model synchronized with the physical information technology innovation data system. Through simulation analysis and machine learning on this digital twin model, it predicts system performance bottlenecks and generates resource optimization strategies.

[0021] The knowledge graph-driven decision support unit is used to build and update the knowledge graph of domestic IT innovation data operation and maintenance. This knowledge graph integrates system architecture, business logic, historical faults and solutions. It receives queries through a natural language processing interface, performs intelligent reasoning, and outputs operation and maintenance decision suggestions.

[0022] The adaptive elastic resource scheduling unit is used to monitor system business load indicators in real time, proactively allocate computing, storage and network resources through predictive algorithms, and realize on-demand scaling of operation and maintenance service functions based on microservice architecture.

[0023] According to the present invention, the intelligent multimodal sensing and maintenance unit performs multimodal data analysis to accurately diagnose potential faults in the equipment's heat dissipation system.

[0024] According to the present invention, the multimodal data analysis includes correlation analysis of temperature sensor data and device cooling fan speed.

[0025] According to the present invention, in the federated learning and blockchain privacy protection operation and maintenance unit, the model parameter update records of the federated learning process are stored as transactions on the blockchain to ensure the traceability and auditability of the training process.

[0026] According to the present invention, the digital twin intelligent optimization operation and maintenance unit dynamically adjusts the physical storage location of data based on the prediction of data access patterns in order to improve data read and write efficiency.

[0027] According to the present invention, the knowledge graph-driven decision support unit has a natural language processing interface, which can parse the fault phenomena described by the user in natural language and automatically match the relevant fault nodes and solution paths in the knowledge graph.

[0028] According to the present invention, the adaptive elastic resource scheduling unit performs resource allocation operations.

[0029] According to the present invention, the resource allocation action includes automatically and elastically expanding storage capacity and computing instances during sudden peaks in data traffic and automatically reclaiming idle resources during off-peak periods.

[0030] The beneficial effects of this invention are:

[0031] The domestic IT data operation and maintenance device of the present invention mainly includes five highly coordinated functional units:

[0032] The intelligent multimodal perception and operation and maintenance unit extends operation and maintenance perception from the digital space to the physical world, enabling precise early warning based on multi-source information fusion; the federated learning and blockchain privacy protection operation and maintenance unit achieves "data not leaving the domain and value circulating" through federated learning, and establishes the trust foundation of the operation and maintenance process through blockchain; the digital twin intelligent optimization operation and maintenance unit creates a digital mirror parallel to the physical system, previewing the future and optimizing the present in virtual space; the knowledge graph-driven decision support unit systematizes and structures the scattered operation and maintenance knowledge, building a tireless "expert brain"; the adaptive elastic resource scheduling unit enables the operation and maintenance system to have "breathing" elasticity, automatically adjusting resource forms according to business pulse.

[0033] This invention, through multimodal perception and digital twin technology, can detect potential faults and performance bottlenecks in advance, transforming the operation and maintenance mode from "passive firefighting" to "proactive prevention and control," significantly reducing unplanned downtime and the risk of data loss.

[0034] This invention combines federated learning with blockchain, enabling cross-domain collaborative analysis and intelligent operation and maintenance while strictly protecting data privacy. It also ensures that all operations are auditable and tamper-proof, greatly improving the compliance and credibility of operation and maintenance, and solving the challenges of privacy and collaboration.

[0035] This invention improves the efficiency and scientific rigor of decision-making. Knowledge graphs transform obscure operational data into reasonable knowledge, assisting operations personnel in quickly locating the root causes of complex problems. Furthermore, natural language interaction lowers the barrier to entry, enhancing the scientific rigor and efficiency of decision-making.

[0036] This invention optimizes resource utilization and ensures business flexibility. The adaptive elastic architecture ensures that system resources can be dynamically adjusted to meet business needs, guaranteeing business stability during peak periods while improving resource utilization during normal times, thereby systematically reducing operation and maintenance costs.

[0037] This invention forms a complete intelligent operation and maintenance closed loop from perception, analysis, decision-making to execution through the cooperation of five major units, which comprehensively improves the reliability, security and economy of the information technology innovation data system. Attached Figure Description

[0038] Figure 1 The present invention provides a detailed flowchart of a data operation and maintenance method for information technology innovation.

[0039] Figure 2 This invention provides a structural block diagram of a data operation and maintenance device for information technology innovation.

[0040] In the diagram, there are five units: Intelligent Multimodal Perception Operation and Maintenance Unit 1, Federated Learning and Blockchain Privacy Protection Operation and Maintenance Unit 2, Digital Twin Intelligent Optimization Operation and Maintenance Unit 3, Knowledge Graph-Driven Decision Support Unit 4, and Adaptive Elastic Resource Scheduling Unit 5. Detailed Implementation

[0041] Firstly, this invention provides a method for the operation and maintenance of information technology application innovation (ITAI) data.

[0042] See Figure 1 As shown, the present invention provides a data operation and maintenance method for information technology innovation, the specific implementation process of which is as follows:

[0043] Step S1: Intelligent multimodal perception operation and maintenance;

[0044] Collect images, sounds, and various sensor data from the domestic IT environment; perform data fusion and rapid analysis at the edge to identify abnormal device status and issue immediate alarms.

[0045] This invention uses intelligent analysis of the correlation between different modal data, such as combining data from computer room temperature sensors with the speed of equipment cooling fans, to accurately determine potential equipment failure risks.

[0046] This invention utilizes edge computing technology to quickly process some data locally, reducing data transmission latency, enabling immediate response to data anomalies, improving the timeliness and accuracy of data operation and maintenance in the field of information technology innovation, and reducing the risk of data loss.

[0047] Step S2: Federated Learning and Blockchain Privacy Protection Operations and Maintenance;

[0048] A federated learning task is initiated among multiple domestic IT innovation data nodes, where each node trains the model locally and exchanges parameters; operation and maintenance operations and data access records are written to the blockchain through smart contracts.

[0049] This invention leverages federated learning technology to achieve joint analysis and model training of multi-source domestic IT innovation data without disclosing the original data, meeting the collaborative operation and maintenance needs of different departments under data privacy protection. Utilizing the immutability and traceability of blockchain, data access and operation records are encrypted and stored, ensuring the security and auditability of data operations. Smart contracts regulate data usage permissions and operation and maintenance processes, preventing data misuse, protecting the privacy and security of domestic IT innovation data throughout its entire lifecycle, and improving the compliance and credibility of data operation and maintenance.

[0050] Step S3: Intelligent optimization and maintenance using digital twins;

[0051] Establish and synchronize a digital twin model of the physical data system; run simulations on the digital twin to predict data hotspots and performance bottlenecks; and perform automated optimization of data storage layout and computing resources based on the prediction results.

[0052] This invention pre-evaluates the effectiveness of different operation and maintenance strategies by virtually simulating them in a digital twin model, thereby achieving zero-risk operation and maintenance strategy optimization for the physical system.

[0053] This invention constructs a digital twin model of a domestically developed data system to simulate the flow, storage, and processing of data in real time. Through analysis of the digital twin model, it can predict issues such as data hotspots and performance bottlenecks in advance.

[0054] This invention utilizes machine learning algorithms to automatically optimize data storage layout and resource allocation strategies, such as dynamically adjusting storage locations based on data access frequency to improve data read and write performance.

[0055] This invention combines virtual simulation technology to pre-simulate operation and maintenance strategies, assess their impact on data systems, realize intelligent and precise operation and maintenance of domestically developed data, and improve the overall performance of data systems.

[0056] Step S4: Knowledge Graph-Driven Decision Making;

[0057] Build and continuously enrich the operation and maintenance knowledge graph; receive user queries, use the graph to reason and locate the root cause of the problem and recommend solutions.

[0058] This invention integrates knowledge from various aspects such as data structure, business logic, and operation and maintenance experience by constructing a knowledge graph of information technology innovation data.

[0059] This invention uses knowledge graph association analysis to quickly locate the root cause of data problems, such as finding key links that affect data accuracy based on data dependencies.

[0060] This invention utilizes natural language processing technology to enable intelligent interaction between users and knowledge graphs, allowing users to obtain data operation and maintenance solutions through natural language queries.

[0061] This invention uses knowledge graphs for intelligent reasoning to predict potential data risks, providing decision support for operations and maintenance personnel and improving operational efficiency and the scientific nature of decision-making.

[0062] Step S5: Adaptive elastic resource scheduling;

[0063] Continuously monitor system load; use predictive models to anticipate changes in resource demand; automatically execute resource allocation scaling up and down operations to match current business pressure.

[0064] This invention designs an adaptive and elastic architecture that automatically adjusts the configuration of operation and maintenance resources based on changes in the business load of domestically developed data. Simultaneously, it utilizes real-time monitoring data, such as data traffic and concurrent access volume, to predict load trends through intelligent algorithms, and pre-allocates or releases computing and storage resources.

[0065] When data volume experiences a sudden surge, this invention automatically expands storage capacity and computing power. When the load decreases, idle resources are reclaimed, improving resource utilization. A microservice architecture decouples operational functions, facilitating flexible expansion and upgrades, ensuring stable and efficient operation of the domestically developed data system under varying loads, and reducing operational costs.

[0066] As a specific implementation method, the present invention provides a data operation and maintenance method for information technology innovation, the specific implementation process of which is as follows:

[0067] (1) System initialization: Deploy sensors such as image sensors, acoustic sensors and environmental sensors, and construct an initial digital twin model and knowledge graph;

[0068] (2) The intelligent multimodal perception operation and maintenance process runs continuously, providing real-time monitoring of the physical environment and equipment health status;

[0069] (3) When an anomaly is detected or an analysis task is triggered as planned, other units are invoked according to the task type;

[0070] If cross-security domain collaborative analysis is required, initiate the federated learning and blockchain privacy protection operation and maintenance process;

[0071] If system optimization or strategy verification is required, simulation should be performed in the digital twin model.

[0072] If a complex fault occurs, intelligent diagnosis will be performed using a knowledge graph.

[0073] (4) Adaptive elastic resource scheduling continuously monitors system load and dynamically adjusts resources according to system load;

[0074] (5) The logs, decision-making basis and execution results generated by each unit are recorded and used to optimize the model and knowledge graph, forming a closed loop of continuous learning.

[0075] Secondly, the present invention provides a data operation and maintenance device for domestic IT innovation, which is used to implement the data operation and maintenance method for domestic IT innovation provided in the first aspect.

[0076] This invention provides a data operation and maintenance device for domestically developed IT applications, integrating five core subsystems: Intelligent multimodal perception operation and maintenance unit 1, which achieves early warning of equipment status through the fusion analysis of image, sound, and sensor data; Federated learning and blockchain privacy protection operation and maintenance unit 2, which achieves collaborative analysis and security auditing while protecting data privacy; Digital twin intelligent optimization operation and maintenance unit 3, which predicts performance bottlenecks and optimizes resource allocation through virtual models; Knowledge graph-driven decision support unit 4, which integrates diverse knowledge to achieve rapid root cause location and intelligent decision-making for faults; and Adaptive elastic resource scheduling unit 5, which dynamically adjusts resources according to business load to ensure system stability. This invention achieves real-time, intelligent, secure, and efficient data operation and maintenance for domestically developed IT applications.

[0077] See Figure 2 The present invention provides a data operation and maintenance device for information technology innovation, comprising the following modules:

[0078] (1) Intelligent multimodal perception and maintenance unit 1 is used to collect the operation data of the information technology innovation equipment and the computer room environment through deployed image sensors, acoustic sensors and environmental sensors, and perform multimodal data fusion analysis based on edge computing nodes to judge the health status of the equipment in real time and trigger early warning.

[0079] According to the present invention, the intelligent multimodal sensing and maintenance unit 1 is used to perform multimodal data analysis, which includes correlation analysis of temperature sensor data and equipment cooling fan speed, so as to accurately diagnose potential faults in the equipment cooling system.

[0080] As one embodiment, the intelligent multimodal perception and maintenance unit 1 can consist of cameras, microphones, temperature and humidity sensors, vibration sensors, etc., deployed in the domestic IT data center, and the data is processed in real time by an edge server deployed locally in the data center. For example, when the analysis model detects abnormal noise from a server's cooling fan (sound mode) and an abnormal increase in its air outlet temperature (temperature mode), it immediately sends an early warning of the server fan failure to the maintenance center.

[0081] (2) Federated learning and blockchain privacy protection operation and maintenance unit 2 is connected to the intelligent multimodal perception operation and maintenance unit 1. It is used to coordinate multiple participants to jointly train the model through federated learning without sharing the original data, and to use the blockchain to store data access logs and operation and maintenance records, and to automatically execute access control policies through smart contracts.

[0082] According to the present invention, in the federated learning and blockchain privacy protection operation and maintenance unit 2, the model parameter update records of the federated learning process are stored as transactions on the blockchain to ensure the traceability and auditability of the training process.

[0083] As one example, the Federated Learning and Blockchain Privacy Protection Operations Unit 2 is initiated when cross-departmental analysis of abnormal data access patterns is required. For instance, data center A and data center B want to jointly train a more accurate intrusion detection model, but they cannot directly exchange user access logs. The Federated Learning and Blockchain Privacy Protection Operations Unit 2 coordinates the two parties to train the model locally using their own data, exchanging only model parameters (such as gradients) for aggregation. Simultaneously, the task ID, participants, and aggregation result hash of this federated learning process are all recorded on the blockchain for auditing purposes.

[0084] (3) Digital twin intelligent optimization operation and maintenance unit 3 is used to build and maintain a digital twin model synchronized with the physical information technology innovation data system. Through simulation analysis and machine learning on the digital twin model, the system performance bottleneck is predicted and resource optimization strategies are generated.

[0085] According to the present invention, the digital twin intelligent optimization operation and maintenance unit 3 dynamically adjusts the physical storage location of data based on the prediction of data access patterns in order to improve data read and write efficiency.

[0086] As one example, the digital twin intelligent optimization and maintenance unit 3 creates a high-fidelity virtual model for the physical domestic IT data cluster (such as a storage system based on Kunpeng servers). Maintenance personnel can simulate peak business activity like during "Double Eleven" on this model, proactively identify potential bottlenecks in storage I / O, and execute optimization strategies accordingly, such as migrating hot data to high-performance SSDs.

[0087] (4) Knowledge graph-driven decision support unit 4 is used to build and update the knowledge graph of information technology innovation data operation and maintenance. The knowledge graph of information technology innovation data operation and maintenance integrates system architecture, business logic, historical faults and solutions. It receives queries through natural language processing interface and performs intelligent reasoning to output operation and maintenance decision suggestions.

[0088] According to the present invention, the knowledge graph-driven decision support unit 4 has a natural language processing interface, which can parse the fault phenomena described by the user in natural language and automatically match the relevant fault nodes and solution paths in the knowledge graph.

[0089] As one example, the knowledge graph of the knowledge graph-driven decision support unit 4 contains entities and relationships such as "Application A depends on database B" and "Table C of database B is generated by ETL task D". When a user queries "the reason for the delay in displaying data in application A", the system parses the query through natural language processing, then reasons in the knowledge graph, and finally outputs a possible cause chain: "ETL task D failed -> Table C data not updated -> Application A reads old data", and recommends the solution "check and restart task D".

[0090] (5) Adaptive elastic resource scheduling unit 5 is used to monitor the system business load indicators in real time, and to proactively allocate computing, storage and network resources through prediction algorithms, and to realize on-demand scaling of operation and maintenance service functions based on microservice architecture.

[0091] According to the present invention, the adaptive elastic resource scheduling unit 5 performs resource allocation actions. These resource allocation actions mainly include automatically and elastically expanding storage capacity and computing instances during sudden data traffic spikes and automatically reclaiming idle resources during periods of low load.

[0092] As one implementation, the adaptive elastic resource scheduling unit 5 monitors the number of concurrent connections to the data query service in real time. When it is predicted that the number of connections will exceed the threshold within the next 5 minutes, it automatically calls the cloud platform API to elastically scale up a new query service instance to share the load.

[0093] This invention utilizes multimodal sensing technologies such as image recognition, sound monitoring, and sensor data acquisition to conduct comprehensive real-time monitoring of the storage environment and equipment operating status of information technology innovation data.

[0094] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A data operation and maintenance method for information technology innovation, characterized in that, Includes the following steps: Step S1: Intelligent multimodal perception operation and maintenance; Collect images, sounds, and various sensor data from the information technology innovation environment; Perform data fusion and rapid analysis at the edge to identify abnormal device status and issue immediate alerts; Step S2: Federated Learning and Blockchain Privacy Protection Operations and Maintenance; A federated learning task is initiated among multiple domestic IT innovation data nodes, where each node trains the model locally and exchanges parameters; operation and maintenance operations and data access records are written to the blockchain through smart contracts. Step S3: Intelligent optimization and maintenance using digital twins; Establish and synchronize a digital twin model of the physical data system; run simulations on the digital twin to predict data hotspots and performance bottlenecks; and perform automated optimization of data storage layout and computing resources based on the prediction results. Step S4: Knowledge Graph-Driven Decision Making; Build and continuously enrich the operation and maintenance knowledge graph; Receive user queries, use graph reasoning to locate the root cause of the problem and recommend solutions; Step S5: Adaptive elastic resource scheduling; Continuously monitor system load; use predictive models to anticipate changes in resource demand; automatically execute resource allocation scaling up and down operations to match current business pressure.

2. The data operation and maintenance method for information technology innovation according to claim 1, characterized in that, In step S3, the effectiveness of different operation and maintenance strategies is evaluated in advance by virtual simulation of different operation and maintenance strategies in the digital twin model, so as to achieve zero-risk operation and maintenance strategy optimization of the physical system.

3. A data operation and maintenance device for domestically developed information technology (IT) innovation, used to implement the data operation and maintenance method for IT innovation as described in claim 1 or 2, characterized in that, The device includes: The intelligent multimodal perception and operation and maintenance unit (1) is used to collect the operation data of the information technology innovation equipment and the computer room environment through the deployed image sensor, acoustic sensor and environmental sensor, and perform multimodal data fusion analysis based on edge computing nodes to judge the health status of the equipment in real time and trigger early warning; The federated learning and blockchain privacy protection operation and maintenance unit (2) is connected to the intelligent multimodal perception operation and maintenance unit (1) to coordinate multiple participants to jointly train the model through federated learning without sharing the original data, and to use the blockchain to store data access logs and operation and maintenance records, and to automatically execute access control policies through smart contracts. The digital twin intelligent optimization operation and maintenance unit (3) is used to build and maintain a digital twin model that is synchronized with the physical information technology innovation data system. Through simulation analysis and machine learning on the digital twin model, the system performance bottleneck is predicted and resource optimization strategies are generated. The knowledge graph-driven decision support unit (4) is used to build and update the knowledge graph of the operation and maintenance of information technology innovation data. The knowledge graph of the operation and maintenance of information technology innovation data integrates system architecture, business logic, historical faults and solutions. It receives queries through the natural language processing interface and performs intelligent reasoning to output operation and maintenance decision suggestions. The adaptive elastic resource scheduling unit (5) is used to monitor the system's business load indicators in real time, and to proactively allocate computing, storage and network resources through prediction algorithms, and to realize on-demand scaling of operation and maintenance service functions based on microservice architecture.

4. The information technology innovation data operation and maintenance device according to claim 3, characterized in that, The intelligent multimodal perception and maintenance unit (1) performs multimodal data analysis to accurately diagnose potential faults in the equipment's heat dissipation system.

5. The information technology innovation data operation and maintenance device according to claim 4, characterized in that, The multimodal data analysis includes correlation analysis of temperature sensor data and equipment cooling fan speed.

6. The information technology application innovation data operation and maintenance device according to claim 3, characterized in that, In the federated learning and blockchain privacy protection operation and maintenance unit (2), the model parameter update records of the federated learning process are stored as transactions on the blockchain to ensure the traceability and auditability of the training process.

7. The data operation and maintenance device for information technology innovation according to claim 3, characterized in that, The digital twin intelligent optimization operation and maintenance unit (3) dynamically adjusts the physical storage location of data based on the prediction of data access patterns in order to improve data read and write efficiency.

8. The information technology innovation data operation and maintenance device according to claim 3, characterized in that, The knowledge graph-driven decision support unit (4) has a natural language processing interface, which can parse the fault phenomena described by the user in natural language and automatically match the relevant fault nodes and solution paths in the knowledge graph.

9. The data operation and maintenance device for information technology innovation according to claim 3, characterized in that, The adaptive elastic resource scheduling unit (5) performs resource allocation actions.

10. The information technology innovation data operation and maintenance device according to claim 9, characterized in that, The resource allocation actions include automatically and elastically expanding storage capacity and computing instances during sudden peaks in data traffic and automatically reclaiming idle resources during off-peak periods.

Images