CN-122024483-A - Traffic condition on-line monitoring system based on distributed optical fiber

CN122024483ACN 122024483 ACN122024483 ACN 122024483ACN-122024483-A

Abstract

The invention relates to the technical field of optical fiber intelligent traffic monitoring and discloses a traffic condition on-line monitoring system based on distributed optical fibers. The system comprises an optical signal sensing unit for capturing a sensing optical signal generated by traffic load of an optical fiber in a road, a disturbance track describing unit for carrying out time-frequency joint demodulation on the signal to generate a dynamic disturbance track describing the time-space motion characteristic of a vehicle, a scene slice generating unit for cutting the track into discretized traffic scene slices according to a preset time-space rule, a behavior pattern distilling unit for carrying out clustering and sequence analysis on historical slices and distilling a scene slice template library of a typical traffic behavior pattern, and a real-time state mapping unit for matching a new slice with the template library and calculating a state offset according to the degree of difference. The invention realizes the refined continuous depiction of traffic disturbance and intelligent state evaluation based on pattern learning.

Inventors

HE JIALIN
FAN ZHONGKAI
WANG YUNXIAO
WANG CUILI
KANG SHAOKUN

Assignees

河北冀翔通电子科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260316

Claims (10)

1. Traffic condition on-line monitoring system based on distributed optical fiber, characterized by comprising: The optical signal sensing unit is used for continuously capturing sensing optical signals generated by the distributed optical fibers laid in the road structure due to disturbance of traffic load; the disturbance track depiction unit is used for carrying out time-frequency joint demodulation on the sensing optical signals to generate dynamic disturbance tracks depicting the motion characteristics of vehicles or traffic flows in two dimensions of time and space, wherein each dynamic disturbance track comprises a starting position, a terminating position, duration and an intensity evolution profile of a vibration event on an optical fiber; The scene slice generation unit is used for cutting the dynamic disturbance track into discrete traffic scene slices with clear space-time boundaries according to preset road section division rules and time window rules, and attaching corresponding road identifiers and time stamps to each traffic scene slice; The behavior pattern distillation unit is used for carrying out clustering and serialization analysis on a plurality of traffic scene slices belonging to the same road identifier in a preset period, and distilling a scene slice template library representing a typical traffic behavior pattern of the road section; And the real-time state mapping unit is used for carrying out multi-granularity matching on the newly generated traffic scene slice and the scene slice template library under the corresponding road identifier, and calculating the state offset of the current traffic scene slice relative to the historical typical mode according to the matching difference degree.
2. The distributed optical fiber based traffic condition on-line monitoring system according to claim 1, wherein the disturbance trajectory characterization unit comprises: Converting the sensing optical signal captured by the optical signal sensing unit into a digital signal sequence containing phase and amplitude information by adopting a coherent detection technology; Performing space-time two-dimensional transformation on the digital signal sequence to generate a signal intensity distribution diagram taking the position of the optical fiber as a horizontal axis and taking time as a vertical axis; In the signal intensity distribution diagram, a self-adaptive threshold segmentation algorithm is applied to identify a significant vibration region exceeding the ambient noise substrate; carrying out contour tracing and feature point extraction on each remarkable vibration region, and connecting a plurality of feature points with space-time continuity to form an initial vibration event track; And carrying out smooth filtering and abnormal point removing processing on the initial vibration event track to form a final output dynamic disturbance track, wherein the dynamic disturbance track is formed by a series of track points which are arranged in time sequence, and each track point comprises accurate optical fiber position coordinates and instantaneous vibration intensity of the optical fiber position coordinates.
3. The distributed optical fiber based traffic condition online monitoring system according to claim 2, wherein the scene cut generation unit comprises: Presetting a road geographic information system database which defines a rule for dividing the whole road network into a plurality of independent monitoring sections, wherein each monitoring section has a unique road identifier and a physical length range; Presetting a sliding time window rule, wherein the sliding time window rule defines a mechanism for dividing a continuous time axis into time segments with equal or unequal lengths; Mapping the dynamic disturbance track generated by the disturbance track describing unit to a specific monitoring section in a road geographic information system database according to the starting position and the ending position of the vibration event contained in the dynamic disturbance track, and distributing a corresponding road identifier for the dynamic disturbance track; According to the sliding time window rule, the same time stamp is allocated for the dynamic disturbance track segments which fall in the same time window and have the same road identifier; And cutting and reorganizing the dynamic disturbance track by taking the road identifier and the time stamp as joint keys to generate structured traffic scene slices, wherein the data body of each traffic scene slice comprises track point sets, average intensity and disturbance density of all vibration events in space-time units corresponding to the traffic scene slice.
4. The distributed fiber based traffic condition on-line monitoring system of claim 3, wherein the behavior pattern distillation unit comprises: Setting a fixed statistical period, and screening all traffic scene slices with the same road identifier from the historical traffic scene slices output by the scene slice generating unit; Grouping the screened traffic scene slices according to the spatial distribution characteristics and disturbance density time sequence characteristics of the track point set contained in the traffic scene slices by adopting an unsupervised clustering algorithm, and classifying the traffic scene slices with similar characteristics into the same cluster; Performing feature alignment and averaging calculation on traffic scene slices in the same cluster to generate an aggregate scene slice representing central features of the same cluster as an initial scene slice template; Analyzing the appearance sequence and the transition probability of scene slice templates of different clusters on a time axis, and constructing a scene slice template chain based on a time sequence for representing the evolution rule of traffic flow; And storing all the initial scene slice templates and the relation thereof in a template chain together as a scene slice template library corresponding to the road identifier.
5. The distributed optical fiber based traffic condition online monitoring system according to claim 4, wherein the real-time status mapping unit comprises: receiving a traffic scene slice newly generated by the scene slice generation unit and reading a road identifier attached to the traffic scene slice; According to the road identifier, a corresponding scene slice template and a template chain are retrieved from a scene slice template library maintained by the behavior pattern distillation unit; Carrying out similarity calculation on the newly generated traffic scene slice and each retrieved scene slice template in a plurality of dimensions such as track point space distribution, intensity profile, disturbance density curve and the like; Selecting a scene slice template with the highest similarity calculation value as an optimal matching template, and taking the similarity calculation value as a component of the state offset of the current traffic scene slice after supplementing the similarity calculation value; Further checking whether the front-to-back relation of the newly generated traffic scene slice in the scene slice template chain accords with the historical evolution rule, and if not, generating a relation violation degree as another component of the state offset; And synthesizing the similarity complement value and the relation violation degree, and generating a final state offset value through weighted fusion to quantify the difference degree between the current traffic state and the historical typical mode.
6. The distributed fiber based traffic condition online monitoring system of claim 5, further comprising a state offset analysis unit comprising: Receiving state offset values generated by the real-time state mapping unit for a series of continuous time windows to form a state offset time sequence; Trend analysis and mutation point detection are carried out on the state offset time sequence, and an abnormal period in which the state offset is continuously higher or is rapidly increased is identified; The identified abnormal time period is associated and packaged with the corresponding traffic scene slice and the road identifier, and a structured traffic state abnormal event record is generated; the traffic state anomaly event record is transmitted to a subsequent decision unit for further analysis.
7. The distributed fiber based traffic condition online monitoring system of claim 6, further comprising a provenance locator unit comprising: when a traffic state abnormal event record from the state offset analysis unit is received, extracting an associated traffic scene slice; Performing reverse backtracking analysis on a dynamic disturbance track contained in the traffic scene slice, and positioning the accurate position and the starting moment of an optical fiber which cause the initial disturbance of state deviation; combining with a road geographic information system database, converting the accurate position of the optical fiber into specific lane or road stake number information; And generating an abnormal event detail report containing the traceability positioning result, wherein the abnormal event detail report is combined with the original traffic state abnormal event record to form an enhanced abnormal event description.
8. The distributed fiber based traffic condition online monitoring system of claim 7, further comprising a template library adaptive update unit for: Continuously monitoring the state offset value output by the real-time state mapping unit; when the state offset is in a low level for a long time and the newly generated traffic scene slice shows a new stable mode under the specific road identifier, triggering a template library updating flow; Generating a new scene slice template after cluster analysis of the new traffic scene slices meeting the conditions, integrating the new scene slice template into an original scene slice template library, and simultaneously adjusting the transfer relation of a template chain; The updated scene slice template library will replace the old version for the subsequent real-time state mapping process.
9. The distributed fiber based traffic condition online monitoring system of claim 8, further comprising a visual presentation unit for: Establishing a mapping relation with a road geographic information system map to form a visual base map; superposing the state offset calculated by the real-time state mapping unit on the corresponding road section of the visual base map in a thermodynamic diagram or color coding mode; Warning and displaying the abnormal event detail report generated by the tracing positioning unit on a visual base map in a pop-up mark or animation mode; And providing an interaction control, and allowing a user to inquire a historical scene slice template library, a real-time traffic scene slice and a state offset historical curve of any road section.
10. The distributed fiber based traffic condition online monitoring system of claim 9, further comprising a control instruction generation unit for: Receiving a traffic state abnormal event record from the state offset analysis unit and an enhanced abnormal event description from the tracing positioning unit; Matching corresponding response strategies for different types of abnormal events according to a preset traffic management strategy rule base; And automatically generating a specific equipment control instruction sequence according to the matched response strategy, wherein the equipment control instruction sequence can be sent to a corresponding traffic signal controller, a variable information board or a lane indicator.

Description

Traffic condition on-line monitoring system based on distributed optical fiber Technical Field The invention relates to the technical field of optical fiber intelligent traffic monitoring, in particular to a traffic condition on-line monitoring system based on distributed optical fibers. Background In the existing traffic condition monitoring technology, distributed optical fiber sensing is a common means. Such techniques typically sense vibration or strain signals caused by traffic of the vehicle by embedding sensing fibers in the road structure, thereby enabling detection, counting, or rough speed measurement of the vehicle. However, the existing methods are limited to simple detection and parameter extraction of vibration events, and the processing mode of the existing methods cannot fully mine the abundant space-time information contained in the signals. The common practice is to set a threshold value to judge the occurrence of an event or to perform simple spectrum analysis, wherein the acquired information has a lower dimension, and only discrete and punctiform vehicle passing records can be obtained, so that the motion process of the vehicle in the whole road section and the dynamic evolution of a disturbance field caused by the motion process can not be continuously and finely described. In the aspect of further traffic state analysis and pattern recognition, the existing method often relies on statistical analysis of the basic parameters or comparison with a preset fixed threshold to judge the state of congestion and the like. This approach lacks the ability to mine and learn deep patterns of traffic flow dynamics. The system can only perceive whether real-time data is beyond the normal range, but cannot understand what is the "normal" behavior pattern of the road segment over a particular period of time. Due to the lack of systematic induction and structural expression of historical typical traffic scenes, the reference of state evaluation is fuzzy, automatic knowledge extraction from massive monitoring data is difficult to realize, and accurate quantitative evaluation is carried out on the fine deviation of the current traffic state relative to the historical normalcy. Disclosure of Invention The invention aims to provide a distributed optical fiber-based traffic condition on-line monitoring system so as to solve the problems in the background art. To achieve the above object, the present invention provides a distributed optical fiber-based traffic condition on-line monitoring system, the system comprising: The optical signal sensing unit is used for continuously capturing sensing optical signals generated by the distributed optical fibers laid in the road structure due to disturbance of traffic load; the disturbance track depiction unit is used for carrying out time-frequency joint demodulation on the sensing optical signals to generate dynamic disturbance tracks depicting the motion characteristics of vehicles or traffic flows in two dimensions of time and space, wherein each dynamic disturbance track comprises a starting position, a terminating position, duration and an intensity evolution profile of a vibration event on an optical fiber; The scene slice generation unit is used for cutting the dynamic disturbance track into discrete traffic scene slices with clear space-time boundaries according to preset road section division rules and time window rules, and attaching corresponding road identifiers and time stamps to each traffic scene slice; The behavior pattern distillation unit is used for carrying out clustering and serialization analysis on a plurality of traffic scene slices belonging to the same road identifier in a preset period, and distilling a scene slice template library representing a typical traffic behavior pattern of the road section; And the real-time state mapping unit is used for carrying out multi-granularity matching on the newly generated traffic scene slice and the scene slice template library under the corresponding road identifier, and calculating the state offset of the current traffic scene slice relative to the historical typical mode according to the matching difference degree. Preferably, the disturbance trajectory characterization unit includes: Converting the sensing optical signal captured by the optical signal sensing unit into a digital signal sequence containing phase and amplitude information by adopting a coherent detection technology; Performing space-time two-dimensional transformation on the digital signal sequence to generate a signal intensity distribution diagram taking the position of the optical fiber as a horizontal axis and taking time as a vertical axis; In the signal intensity distribution diagram, a self-adaptive threshold segmentation algorithm is applied to identify a significant vibration region exceeding the ambient noise substrate; carrying out contour tracing and feature point extraction on each remarkable vibration region, and connecti