CN-121984769-A - Security situation early warning system and method based on digital twinning and multi-mode sensing

CN121984769ACN 121984769 ACN121984769 ACN 121984769ACN-121984769-A

Abstract

The application relates to a security situation early warning system and method based on digital twinning and multi-modal sensing, wherein the system comprises a data fusion module, a twinning modeling module, a situation analysis module and an early warning generation module, wherein the data fusion module is used for respectively executing time synchronization, feature extraction and standardization processing actions on multi-modal raw data based on multi-modal raw data of a plurality of sensing sources to generate fused multi-modal feature data, the twinning modeling module is used for updating state information of a digital twinning model based on real-time sensing data to generate twinning state data, the situation analysis module is used for calculating security situation indexes based on twinning state data and historical situation data, and the early warning generation module is used for executing risk prediction actions according to situation data to generate early warning information comprising risk grades, influence areas and early warning time and executing security situation early warning according to control instructions. The application can improve the precision and consistency of multi-source data fusion and enhance the timeliness and reliability of risk assessment and early warning.

Inventors

QIAN CHENG
XIAO XIAO
LIN KAI
LU MAORONG
WEI DAI
HANG YIXUAN

Assignees

国电环境保护研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20260213

Claims (10)

1. A security posture early warning system based on digital twinning and multi-modal sensing, the system comprising: The data fusion module is used for respectively executing time synchronization, feature extraction and standardization processing actions on the multi-mode original data based on the multi-mode original data of the plurality of perception sources so as to generate fused multi-mode feature data; The twin modeling module is used for constructing a digital twin model corresponding to the target physical space according to the fused multi-modal characteristic data, updating the state information of the digital twin model based on real-time perception data and generating twin state data; the situation analysis module is used for calculating a safety situation index based on the twin state data and the historical situation data and generating situation data containing a risk assessment result according to the safety situation index; And the early warning generation module is used for executing risk prediction actions according to the situation data, generating early warning information comprising risk grades, influence areas and early warning time, and generating control instructions corresponding to the risk grades based on the early warning information so as to execute safety situation early warning according to the control instructions.
2. The security posture pre-warning system based on digital twinning and multi-modal awareness of claim 1, further comprising: And the feedback learning module is used for receiving the execution data of the control instruction, comparing the risk assessment results to generate feedback data, and updating the twin state data, the risk threshold value and the feature selection parameters based on the feedback data to obtain a final feedback learning result.
3. The security posture pre-warning system based on digital twinning and multi-modal awareness of claim 1, wherein the data fusion module comprises: the time synchronization unit is used for receiving the multi-mode original data of the plurality of perception sources, and performing time alignment processing on the multi-mode original data based on a unified clock source so as to obtain processed multi-mode original data; the feature extraction unit is used for extracting the processed multi-mode original data to generate a unified feature data vector; and the fusion unit is used for carrying out standardized processing comprising data normalization, denoising and outlier detection on the unified characteristic data vector so as to generate the fused multi-mode characteristic data.
4. The digital twinning and multi-modal awareness based security posture pre-warning system of claim 3, further comprising: The weighting fusion unit is used for calculating a weighting coefficient according to the historical reliability and the real-time confidence coefficient of each perception source after the fused multi-modal feature data is generated, and carrying out weighting fusion on the multi-modal feature vectors according to the weighting coefficient so as to obtain weighted fused data; and the upstream and downstream information injection unit is used for combining information of a target physical space in real-time perception data with static data and dynamic state data in the digital twin model to obtain target data, and injecting upstream and downstream dynamic information into the weighted and fused data based on the target data so as to optimize the digital twin model according to the upstream and downstream dynamic information.
5. The security posture pre-warning system based on digital twinning and multi-modal awareness of claim 1, wherein the twinning modeling module comprises: The data mapping unit is used for receiving the fused multi-modal feature data and mapping the fused multi-modal feature data according to the dynamic features of the target physical space so as to generate an initial state of the digital twin model; and the dynamic updating unit is used for updating the initial state of the digital twin model based on the real-time perception data and the historical state data and generating the twin state data.
6. The digital twinning and multi-modal awareness based security posture pre-warning system of claim 5, further comprising: And the self-adaptive adjusting unit is used for adjusting modeling parameters of the digital twin model according to real-time perception data change and scene change after the twin state data are generated so as to update the structure of the digital twin model.
7. The security situation early warning system based on digital twinning and multi-modal awareness according to claim 6, wherein the state evolution calculation formula of the digital twinning model is: Wherein, the For the moment of time Is a digital twin model state of (a), For the moment of time Is used to determine the model state of the model, For the moment of time Is used to determine the target state of the (c), For the moment of time Is a target physical space dynamic characteristic variation amount of the (c), The state evolution weight coefficient; the parameter increment updating value of the digital twin model is as follows: Wherein, the For the moment of time Is used for the model parameters of the model (a), For the moment of time Is used to determine the target state of the (c), For the moment of time Is provided with a set of historical state data, In order to lose the gradient of the function to the parameter, For the real-time response coefficient to be a real-time response coefficient, Is the learning rate; the adjusted modeling parameter set of the digital twin model is as follows: ; Wherein, the In order to adjust the set of modeling parameters before adjustment, In order to adaptively adjust the intensity coefficient, For the scene change sensitivity coefficient, For the moment of time Is used for the scene change amount of the (c) in the scene, For parameter adjustment vectors generated based on self-learning strategies, The function is activated for an S-shape.
8. The security posture early warning system based on digital twinning and multi-modal awareness according to claim 1, wherein the posture analysis module comprises: the situation calculation unit is used for calculating the security situation indexes according to the twin state data and the historical situation data and generating situation change vectors according to the security situation indexes; the risk assessment unit is used for identifying potential risk events according to the situation change vector and the real-time multi-mode characteristic data and generating risk data comprising risk grades and influence areas according to the potential risk events; and the situation data generation unit is used for constructing a risk association diagram according to the risk data and generating situation data containing the risk assessment result according to the risk association diagram.
9. The security posture early warning system based on digital twinning and multi-modal awareness according to claim 1, wherein the early warning generation module comprises: The risk judging unit is used for calculating risk influence weights according to the risk levels and the risk association strength, determining risk influence areas and duration according to a time attenuation function and a preset space diffusion model based on the risk influence weights, and generating risk judging results comprising the risk levels, the influence ranges and the risk evolution trend according to the risk influence areas and the duration; The early warning information generation unit is used for generating early warning information comprising a risk level, an influence area, early warning time and recommended response measures according to the risk level threshold and the system response priority based on the risk judgment result, dynamically adjusting early warning trigger conditions based on the early warning information, and generating a control instruction corresponding to the risk level, wherein the control instruction comprises a speed limit instruction, a signal priority instruction or a V2X cooperative control instruction.
10. A security posture pre-warning method based on digital twinning and multi-modal sensing, characterized in that a security posture pre-warning system based on digital twinning and multi-modal sensing as claimed in any one of the preceding claims 1-9 is adopted, the method comprising: Based on multi-mode original data of a plurality of perception sources, respectively executing time synchronization, feature extraction and standardization processing actions on the multi-mode original data to generate fused multi-mode feature data; Constructing a digital twin model corresponding to the target physical space according to the fused multi-modal feature data, and updating state information of the digital twin model based on real-time perception data to generate twin state data; calculating a security situation index based on the twin state data and the historical situation data, and generating situation data containing a risk assessment result according to the security situation index; And executing risk prediction actions according to the situation data, generating early warning information comprising risk grades, influence areas and early warning time, and generating control instructions corresponding to the risk grades based on the early warning information so as to execute security situation early warning according to the control instructions.

Description

Security situation early warning system and method based on digital twinning and multi-mode sensing Technical Field The application relates to the technical field of digital twinning, in particular to a security situation early warning system and method based on digital twinning and multi-mode sensing. Background With the development of sensor network, internet of things, artificial intelligence and other technologies, the safety monitoring system gradually evolves towards the direction of intellectualization and data fusion. Various sensing devices are widely deployed in the scenes of traffic management, industrial production, energy scheduling, urban operation and the like, and sensing of the operation state and risk identification are realized through multi-source data acquisition. However, the existing security monitoring system generally depends on multiple types of data sources and a distributed analysis model, and different sources of information have differences in time, space and confidence, so that high-precision and low-delay global perception is difficult to realize. The whole system is still mainly based on static models or offline analysis, and has insufficient response to dynamic changes and risk evolution rules in a complex environment. In the related technology, the safety monitoring and early warning system has the following general technical problems that high consistency fusion of multisource perception data is difficult to realize, the information utilization rate is low, a model cannot be dynamically updated according to real-time environment changes, the system state lags behind the actual operation of a physical object, the risk analysis method is mostly based on fixed threshold values or single-dimensional characteristics, comprehensive judgment on risk propagation relations, time sequence change rules and multidimensional situation elements is lacking, an early warning mechanism mainly depends on preset rules, dynamic linkage of risk levels, influence ranges and response time cannot be realized, a feedback mechanism mostly comprises manual adjustment or fixed parameter iteration, and the system lacks self-learning capability and continuous optimization capability. The problems limit the accuracy and timeliness of the existing security situation prediction, and the requirements of high reliability and intelligent early warning in a complex environment are difficult to meet, so that improvement is needed. Disclosure of Invention The application provides a security situation early warning system and method based on digital twinning and multi-mode perception, which are used for solving the problems of insufficient precision of multi-source data fusion, delayed model updating, single risk assessment dimension, stiff early warning response, limitation of system lacking self-learning capacity and the like in the security monitoring and risk early warning technology in the related technology, realizing high-precision fusion of multi-source data, dynamic modeling of system state, comprehensive quantitative analysis of risk situation and self-adaptive optimization based on a feedback mechanism, thereby improving the security monitoring capacity, risk prediction precision and response intelligence level of the system in a complex environment. The embodiment of the first aspect of the application provides a digital twinning and multi-modal sensing-based security situation pre-warning system, which comprises a data fusion module, a twinning modeling module, a situation analysis module and a pre-warning generation module, wherein the data fusion module is used for respectively executing time synchronization, feature extraction and standardization processing actions on multi-modal raw data based on multi-modal raw data of a plurality of sensing sources to generate fused multi-modal feature data, the twinning modeling module is used for constructing a digital twinning model corresponding to a target physical space according to the fused multi-modal feature data and updating state information of the digital twinning model based on real-time sensing data to generate twinning state data, the situation analysis module is used for calculating security situation indexes based on the twinning state data and historical situation data and generating situation data containing risk assessment results according to the security situation indexes, and the pre-warning generation module is used for executing risk prediction actions according to the situation data to generate pre-warning information containing risk grades, influence areas and pre-warning time according to the pre-warning information corresponding to the grades to the control instruction to execute security pre-warning according to the control instruction. Optionally, in one embodiment of the present application, the system further includes a feedback learning module, configured to receive the execution data of the control instruction and compare t