CN-122024494-A - Method and system for analyzing and processing all-link data of tunnel traffic abnormal event

CN122024494ACN 122024494 ACN122024494 ACN 122024494ACN-122024494-A

Abstract

The invention relates to the technical field of intelligent traffic systems and tunnel operation safety monitoring, in particular to a method and a system for analyzing and processing full-link data of abnormal events of tunnel traffic, comprising the steps of acquiring multi-source heterogeneous space-time sequence data in a target space; the method comprises the steps of constructing a space-time reference model, triggering a double-track differential analysis flow, including carrying out numerical calculation through a preset continuity equation to generate a standard state matrix, calling a preset abnormal dynamics operator set to synthesize a theoretical abnormal state matrix, calculating a first residual vector, calculating a second residual vector, calculating topological similarity among the first residual vector and the second residual vector, and generating an event diagnosis report corresponding to the selected abnormal dynamics operator.

Inventors

WANG SHOUNIAN
QIU LIN
WANG YAN
WANG WEI

Assignees

浙江佐通信息技术有限公司

Dates

Publication Date: 20260512
Application Date: 20260415

Claims (8)

1. The method for analyzing and processing the full-link data of the tunnel traffic abnormal event is characterized by comprising the following steps: acquiring multi-source heterogeneous space-time sequence data in a target space, wherein the multi-source heterogeneous data comprises dynamic entity motion data acquired by a first type sensor and environment state data acquired by a second type sensor; Constructing a space-time reference model based on the inherent parameters of the target space and the historical reference data; in response to the fluctuation of the dynamic entity motion data, triggering a double-track differential analysis flow, wherein the double-track differential analysis flow comprises: Step S1, based on the dynamic entity motion data and the environmental state data, combining the space-time reference model, carrying out numerical calculation through a preset continuity equation to generate a standard state matrix; s2, calling a preset abnormal dynamics operator set, and injecting the selected abnormal dynamics operator into the standard state matrix to synthesize a theoretical abnormal state matrix; Step S3, calculating a first residual vector based on the multi-source heterogeneous space-time sequence data and the standard state matrix, which are acquired in real time; Step S4, mapping the first residual vector and the second residual vector to the same high-dimensional feature space, and calculating the topological similarity between the first residual vector and the second residual vector; And S5, generating an event diagnosis report corresponding to the selected abnormal dynamics operator according to the calculation result of the topological similarity if the topological similarity is higher than a preset judgment threshold.
2. The method of claim 1, wherein the multi-source heterogeneous spatiotemporal sequence data comprises: Dynamic entity motion data which are collected by radar and a vehicle detector and contain entity ID, instantaneous speed, space coordinates and time intervals; environmental status data comprising light intensity values and medium transmittance collected by a luminance meter and a visibility meter.
3. The method of claim 1, wherein the generating a standard state matrix comprises: assuming that no abnormal event exists in the target space, and the environmental parameters accord with a preset ideal change curve; Using a fluid dynamics continuity equation to perform continuous medium simulation and numerical solution on the dynamic entity motion data; and outputting a standard state matrix which comprises the theoretical speed distribution and the theoretical density distribution along with the time.
4. The method of claim 1, wherein the invoking the set of preset anomaly dynamics operators comprises: invoking mathematical operators describing energy shock, motion stagnation or state propagation mechanisms from a pre-constructed operator knowledge base; parameterizing the mathematical operator to generate energy gradient abrupt change and entropy increase simulation features of a local area in the standard state matrix; And generating the abnormal dynamics operator based on the parameterized mathematical operator.
5. The method of claim 1, wherein the calculating the first residual vector and the second residual vector comprises: carrying out differential calculation on each dimension value of the multi-source heterogeneous space-time sequence data acquired in real time and the corresponding dimension value of the standard state matrix to obtain the first residual vector containing real fluctuation signals and noise; and carrying out differential calculation on each dimension value of the theoretical abnormal state matrix and the corresponding dimension value of the standard state matrix to obtain the second residual vector which purely represents the specific abnormal mode.
6. The method of claim 1, wherein said calculating the topological similarity therebetween comprises: Calculating similarity measurement values of the first residual vector and the second residual vector in the high-dimensional feature space by adopting a cosine similarity algorithm or a dynamic time warping algorithm; And if the fluctuation form of the first residual vector is highly consistent with the specific abnormal form represented by the second residual vector on the similarity measurement value, confirming that the topological similarity is higher than the judging threshold value.
7. The method according to claim 1, wherein the method further comprises: Based on the theoretical abnormal state matrix, deducing a propagation path and an influence range of an abnormal state in a future preset time window; Integrating the propagation path, the influence range and the corresponding abnormality type into the event diagnosis report and outputting the event diagnosis report.
8. A system for analyzing and processing all-link data of a tunnel traffic abnormal event, which is applied to the method of any one of claims 1 to 7, and is characterized by comprising the following steps: The data acquisition module is used for acquiring multi-source heterogeneous space-time sequence data in a target space; The reference modeling module is used for constructing a space-time reference model; an analytic reasoning module comprising: The reference calculation unit is used for generating a standard state matrix; the abnormal synthesis unit is used for synthesizing a theoretical abnormal state matrix; a residual analysis unit for calculating a first residual vector and a second residual vector; the matching judgment unit is used for calculating the topological similarity and generating an event diagnosis report; And the report generation module is used for outputting the event diagnosis report.

Description

Method and system for analyzing and processing all-link data of tunnel traffic abnormal event Technical Field The invention relates to the technical field of intelligent traffic systems and tunnel operation safety monitoring, in particular to a method and a system for analyzing and processing full-link data of tunnel traffic abnormal events. Background Along with the improvement of the digitization level of the traffic infrastructure, the multi-source heterogeneous data scale in the tunnel traffic scene is exponentially increased, and the data complexity provides higher requirements for the real-time monitoring and intelligent analysis of the traffic running state; At present, millimeter wave radar, a video vehicle detector, an environment monitoring station and other devices arranged on the side wall of a tunnel are used for collecting space-time sequence data, a preset fixed threshold value or a simple statistical model is utilized for monitoring and judging indexes such as speed, flow and occupancy, however, the traditional monitoring data processing method depends on numerical comparison of single dimension and is limited by special environment factors of a tunnel enclosed space, in practical application, data fluctuation caused by severe changes of illumination intensity, radar multipath effect or equipment vibration and other environmental noise is extremely easy to be misjudged as traffic abnormality by a system, meanwhile, when the flow is saturated or defensive deceleration of a driver due to the environment factors, a data analysis model lacking physical mechanism constraint is difficult to effectively distinguish physical congestion and visual interference, so that the false alarm rate is higher, and the requirement of accurate sensing of a full link cannot be met. Disclosure of Invention The invention aims to provide a method and a system for analyzing and processing tunnel traffic abnormal event full-link data, which can effectively eliminate false alarm caused by environmental noise under multi-source heterogeneous data interference, and can accurately distinguish physical congestion from defensive deceleration caused by visual environment mutation from a physical layer, and the technical scheme of the invention is as follows: a method for analyzing and processing all-link data of a tunnel traffic abnormal event comprises the following steps: Acquiring multi-source heterogeneous space-time sequence data in a target space, wherein the multi-source heterogeneous data comprises dynamic entity motion data acquired by a first type sensor and environment state data acquired by a second type sensor; constructing a space-time reference model based on inherent parameters of the target space and historical reference data; In response to fluctuation of dynamic entity motion data, triggering a double-track differential analysis flow, wherein the double-track differential analysis flow comprises: step S1, based on dynamic entity motion data and environment state data, combining a space-time reference model, carrying out numerical calculation through a preset continuity equation to generate a standard state matrix; s2, calling a preset abnormal dynamics operator set, and injecting the selected abnormal dynamics operator into a standard state matrix to synthesize a theoretical abnormal state matrix; Step S3, calculating a first residual vector based on the multi-source heterogeneous space-time sequence data acquired in real time and a standard state matrix, and calculating a second residual vector based on the theoretical abnormal state matrix and the standard state matrix; Step S4, mapping the first residual vector and the second residual vector to the same high-dimensional feature space, and calculating the topological similarity between the first residual vector and the second residual vector; And S5, generating an event diagnosis report corresponding to the selected abnormal dynamics operator according to the calculation result of the topological similarity if the topological similarity is higher than a preset judgment threshold. Preferably, the multi-source heterogeneous spatiotemporal sequence data comprises: Dynamic entity motion data which are collected by radar and a vehicle detector and contain entity ID, instantaneous speed, space coordinates and time intervals; environmental status data comprising light intensity values and medium transmittance collected by a luminance meter and a visibility meter. Preferably, generating the standard state matrix includes: Assuming that no abnormal event exists in the target space, and the environmental parameters accord with a preset ideal change curve; carrying out continuous medium simulation and numerical solution on dynamic entity motion data by utilizing a fluid dynamics continuity equation; and outputting a standard state matrix which comprises the theoretical speed distribution and the theoretical density distribution along with the time. Preferably, the