CN-122020384-A - Data base model building method and system

CN122020384ACN 122020384 ACN122020384 ACN 122020384ACN-122020384-A

Abstract

The invention relates to the technical field of data modeling, and discloses a method and a system for establishing a data base model. The method comprises the steps of creating a data base model body, acquiring a basic operation parameter set containing physical environment parameters and logic behavior parameters in real time through a multi-mode heterogeneous data acquisition interface, performing feature decoupling on the parameter set, separating static feature vectors and dynamic feature sequences to construct a feature topological structure, building a virtual mapping model based on the structure, associating features to virtual space nodes through a dynamic topology mapping algorithm, loading a historical operation mode library, activating an operation rule engine according to the matching degree of real-time parameters and the historical modes, generating a dynamic strategy library containing data reconstruction strategies, abnormal monitoring thresholds and model correction instructions, and calling the correction instructions to update the virtual mapping model in an iterative mode when the parameters trigger the abnormal thresholds, and synchronously updating the data base model body, so that model adaptability and data processing accuracy are improved.

Inventors

ZHENG QUAN
WANG LAI
CHEN RONG
ZENG LI
GE ZHIBIN
WANG YUMING
ZHAO ZHIYUAN
WANG SHUAI
PAN JUN
LIU XINQIAO
ZHANG HAO

Assignees

中国民用航空飞行学院
四川捷途宇博航空科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. A method for building a data base model, comprising the steps of: Creating a data base model body, and acquiring a basic operation parameter set of a target object in real time through a multi-mode heterogeneous data acquisition interface, wherein the basic operation parameter set comprises physical environment parameters and logic behavior parameters; Performing feature decoupling processing on the basic operation parameter set, separating out static feature vectors and dynamic feature sequences, and constructing a feature topological structure of the data base model body; Establishing a virtual mapping model based on the characteristic topological structure, and associating the static characteristic vector and the dynamic characteristic sequence to a virtual space node through a dynamic topological mapping algorithm; Loading a historical operation mode library in the virtual mapping model, and activating a corresponding operation rule engine according to the matching degree of the basic operation parameter set acquired in real time and the historical operation mode library; Generating a dynamic strategy library through the operation rule engine, wherein the dynamic strategy library comprises a data reconstruction strategy, an anomaly monitoring threshold value and a model correction instruction; and when the basic operation parameter set triggers an abnormal monitoring threshold, calling a model correction instruction to update the parameters of the virtual mapping model in an iterative manner, and synchronizing the updated mapping relation to the data base model body.
2. The method for building a data base model according to claim 1, wherein the building the feature topology of the data base model body comprises: performing space dimension reduction processing on the static feature vector, extracting key dimension features and generating a dimension identifier; Performing time domain segmentation processing on the dynamic characteristic sequence, and calculating a fluctuation entropy value in each time period; And constructing the node connection relation of the characteristic topological structure according to the relevance of the dimension identifier and the fluctuation entropy value.
3. The method for building a data base model according to claim 2, wherein the building a virtual mapping model based on the feature topology comprises: Generating a space topology vector set according to the node connection relation; Setting a space weight distribution rule based on a pattern matching result in the history operation pattern library; and carrying out weighted fusion on the space topology vector set according to the space weight distribution rule to generate a coordinate mapping table of the virtual space node.
4. The method of claim 3, wherein generating the dynamic policy repository comprises: performing neighborhood scanning on the coordinate mapping table, and identifying high-density node clusters and sparse node areas; setting a data reconstruction strategy according to the distribution characteristics of the high-density node cluster; calculating an anomaly monitoring threshold based on the offset of the sparse node region; And binding the data reconstruction strategy with an anomaly monitoring threshold value to generate a model correction instruction set.
5. The method for building a data base model according to claim 4, wherein the calling model modification instructions to perform parameter iterative updating on the virtual mapping model comprises the following operations: extracting an offset vector of a sparse node area triggering an abnormal monitoring threshold value; Calculating the direction deviation value of the deviation vector and a reference vector in a historical operation mode library; adjusting a weighting coefficient in the space weight distribution rule according to the direction deviation value; and regenerating a coordinate mapping table by adopting the adjusted weighting coefficient.
6. The method of claim 1, wherein the configuration of the multi-modal heterogeneous data collection interface comprises the operations of: deploying an environmental sensor group on a physical layer of a target object, and collecting temperature gradient and vibration spectrum data in real time; Deploying a behavior capture agent on a logic layer, and continuously acquiring an operation instruction sequence and a state switching log; And aligning the temperature gradient, the vibration spectrum data, the operation instruction sequence and the state switching log according to time stamps, and then merging the aligned temperature gradient, the vibration spectrum data, the operation instruction sequence and the state switching log into the basic operation parameter set.
7. The data base model building method according to claim 6, wherein the feature decoupling process includes the operations of: Performing physical feature extraction on the temperature gradient and vibration spectrum data to generate a physical feature matrix; performing behavior mode analysis on the operation instruction sequence and the state switching log to generate a behavior state transition diagram; mapping the physical feature matrix into a static feature vector, and converting the behavior state transition diagram into a dynamic feature sequence.
8. The method for building a data base model according to claim 7, wherein the construction of the historical operating pattern library comprises the following operations: collecting a physical feature matrix and a behavior state transition diagram in a historical operation period of a target object; performing cluster analysis on the physical feature matrix to divide a plurality of physical feature clusters; Performing path mining on the behavior state transition diagram, and extracting a high-frequency state transition chain; and storing the corresponding relation between the physical feature cluster and the high-frequency state transition chain as a history operation mode library.
9. The method for building a data base model according to claim 8, wherein activating the corresponding operation rule engine according to the matching degree between the real-time collected basic operation parameter set and the historical operation mode library comprises: calculating similarity indexes of the real-time physical feature matrix and the historical physical feature clusters; Detecting the coincidence ratio of the real-time behavior state transition diagram and the high-frequency state transition chain; And when the similarity index and the coincidence degree simultaneously meet the preset conditions, activating an operation rule engine bound with the corresponding physical feature cluster.
10. A data base modeling system for implementing the method of any of claims 1-9, comprising: The multi-mode data acquisition module is deployed on a physical layer and a logic layer of the target object and is used for acquiring a basic operation parameter set; The characteristic decoupling engine is connected with the multi-mode data acquisition module and is used for generating a static characteristic vector and a dynamic characteristic sequence; the virtual modeling core is used for receiving the static feature vector and the dynamic feature sequence and constructing a feature topological structure and a virtual mapping model; The strategy library generator is used for loading the history operation mode library and connecting with the virtual modeling core to generate a dynamic strategy library; The model iteration controller is used for receiving an abnormal monitoring signal in the dynamic strategy library and triggering the parameter update of the virtual mapping model; and the data synchronization agent is connected with the model iteration controller and the data base model body and is used for synchronizing the updated mapping relation.

Description

Data base model building method and system Technical Field The invention relates to the technical field of data modeling, in particular to a method and a system for establishing a data base model. Background In the current digital transformation process, data generated by various system operations show obvious characteristics of multi-source, isomerization and dynamic, a data base is used as a core framework for supporting efficient operation and decision analysis of the system, and the construction quality of the data base directly influences the mining and utilization of the data value. However, existing data base model building methods gradually expose a number of problems when faced with complex scenarios. The traditional data acquisition mode depends on a single or limited type of interface, and is difficult to cover multidimensional parameters such as physical environment, logic behavior and the like of a target object, so that the acquired basic data has one-sidedness, and the real running state of the object cannot be comprehensively reflected. Meanwhile, the real-time performance of data acquisition is insufficient, and a certain delay often exists, so that the subsequent analysis and decision based on the data are delayed from the actual operation requirement. In the data feature processing link, the existing method lacks an effective decoupling mechanism for the acquired basic parameter set, static features and dynamic features are mixed together, and a clear feature structure is difficult to form. This results in masked correlations between features, failing to build feature topologies that are stable and have a clear hierarchical relationship, affecting the accuracy and reliability of the subsequent model. The virtual mapping is used as a key means for connecting a physical entity and a digital space, and in the prior art, the association between a virtual model and an actual feature mostly adopts a static mapping mode, so that the dynamic adjustment capability is lacked. When the running state of the target object changes, the virtual space node cannot respond to the dynamic change of the feature in time, so that the deviation between the virtual mapping model and the running state of the physical entity occurs, and the reference value of the model is reduced. In addition, inefficiency in the use of historical operating data is also a significant problem. Most methods fail to combine historical operating modes with real-time data effectively, and lack a rule engine activation mechanism based on matching degree makes it difficult for historical experience to provide effective guidance for real-time operating decisions. Meanwhile, when the abnormal situation is faced, the correction mechanism of the existing model is passive, manual intervention adjustment is needed after the abnormal situation occurs, and automatic dynamic strategy library support is lacked, so that the model is slow in iterative update and difficult to adapt to the continuously-changing operation environment. The problems restrict the adaptability, accuracy and real-time performance of the data base model together, and cannot meet the high requirements of a complex system on data support. Disclosure of Invention The invention aims to provide a data base model building method and system for solving the problems in the background technology. In order to achieve the above object, the present invention provides a method for establishing a data base model, the method comprising: Creating a data base model body, and acquiring a basic operation parameter set of a target object in real time through a multi-mode heterogeneous data acquisition interface, wherein the basic operation parameter set comprises physical environment parameters and logic behavior parameters; Performing feature decoupling processing on the basic operation parameter set, separating out static feature vectors and dynamic feature sequences, and constructing a feature topological structure of the data base model body; Establishing a virtual mapping model based on the characteristic topological structure, and associating the static characteristic vector and the dynamic characteristic sequence to a virtual space node through a dynamic topological mapping algorithm; Loading a historical operation mode library in the virtual mapping model, and activating a corresponding operation rule engine according to the matching degree of the basic operation parameter set acquired in real time and the historical operation mode library; Generating a dynamic strategy library through the operation rule engine, wherein the dynamic strategy library comprises a data reconstruction strategy, an anomaly monitoring threshold value and a model correction instruction; and when the basic operation parameter set triggers an abnormal monitoring threshold, calling a model correction instruction to update the parameters of the virtual mapping model in an iterative manner, and synchronizing the upd