Tunnel traffic linkage data analysis method based on event detection

Abstract

The invention discloses a tunnel traffic linkage data analysis method based on event detection, and relates to the field of tunnel emergency accident data analysis, and the method comprises the steps: collecting the data of an active sensor in a tunnel, and analyzing the road condition in a blind area of the active sensor; comprising the following steps: inputting active sensor data into a vehicle queuing propagation model in a data assimilation framework to provide a predicted road condition, estimating the predicted road condition based on extended Kalman filtering, and outputting corrected estimation to the vehicle queuing propagation model; and the decision-making and response end executes a traffic guidance strategy until an optimal estimation result is obtained to the decision-making and response end. Through automatic detection and automatic response of road condition events in the blind area of the tunnel sensor, the event response performance is improved, and precious time is won for rescue; the manual monitoring requirement is reduced, and the equipment maintenance cost is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of tunnel emergency incident data analysis, and particularly relates to a tunnel traffic linkage data analysis method based on event detection. BACKGROUND

[0002] Tunnel is a semi-closed traffic space with special environment and high safety risk. Especially in long tunnels, once a fire or traffic accident occurs, smoke quickly accumulates and is difficult to diffuse, forming an environment with rapidly declining visibility, which seriously threatens the safety of drivers and passengers and hinders the development of rescue work. The traditional tunnel monitoring system mainly relies on visible light cameras and simple sensors, which has serious visual shielding effect and sensor performance degradation problem in the smoke environment, resulting in decreased event detection accuracy and increased response delay. Especially in the monitoring blind area section, it is impossible to determine the tunnel emergency event and make timely response. SUMMARY

[0003] The present application provides a tunnel traffic linkage data analysis method based on event detection to solve the problems of the prior art.

[0004] In a first aspect, the present application provides a tunnel traffic linkage data analysis method based on event detection, comprising the following steps: Collecting tunnel active sensor data to analyze the road conditions in the active sensor blind area, comprising: The active sensor data is input into the vehicle queuing propagation model in the data assimilation framework to provide predicted road conditions, and then the predicted road conditions are estimated based on the extended Kalman filter and corrected estimated output to the vehicle queuing propagation model; Until the optimal estimation result is obtained, the decision and response end executes the traffic induction strategy.

[0005] Further, in the data assimilation framework, comprising: Initialization: based on the first received abnormal sensor data, the preliminary accident location estimation is obtained by reverse deduction through the vehicle queuing propagation model; After initialization, the data assimilation cycle strategy is executed as follows: Prediction step: the vehicle queuing propagation model predicts the traffic state of all sections in the tunnel at the current time based on the optimal state estimation output by the extended Kalman filter at the last time, and outputs the predicted data; Update step: according to the observation data of all upper and lower section sensors segmented by active sensors at the current time, the difference between the observation data and the predicted data is analyzed, the Kalman gain is calculated for the fusion weight of the predicted data and the observation data, and the fusion weight is based on the covariance of the predicted data and the observation data to obtain the optimal estimation; When the difference between the predicted data calculated by the optimal estimation of the latest time and the observed data is within the system error range, the iteration is stopped and the optimal estimation is output.

[0006] Further, for the establishment of the vehicle queue propagation model, comprising: State variables: traffic density p ( x,t ) - unit veh / km, average speed v ( x,t ) - unit km / h, traffic flow q ( x,t ) - unit veh / h, position x ; Basic parameters: free flow speed v f - unit km / h, jam density p jam - unit veh / km, speed-density relationship index parameter α , relaxation time parameter t - unit s, traffic flow diffusion coefficient k ; The modified Payne-Whitham motion equation is constructed: Continuity equation: In the modified Payne-Whitham model motion equation: The left side of the equation controls the derivative of the substance, which describes the rate of change of speed; The first term on the right side of the equation is the relaxation term: , Describes the tendency of vehicles to adjust the vehicle speed to an expected equilibrium speed v e ( p ), that is, the equilibrium speed-density relationship describes the equilibrium speed corresponding to the density p In the stable flow state; The second term on the right side of the equation is the pressure term, which describes the response of vehicles to changes in density in front; The modified Payne-Whitham model motion equation and the continuity equation form a closed partial differential equation system. Given the initial conditions and boundary conditions, the traffic state at any time in the future in each section of the tunnel, including density and speed, i.e. predicted data, is obtained by solving.

[0007] Further, for the solution of the vehicle queue propagation model, comprising: In the vehicle queuing propagation model, the tunnel is divided into multiple cross sections in space using active sensors as nodes. Each cross section has a length of Δx, and the continuous time is divided into multiple time steps Δt. By labeling the state variables and fundamental parameters with time or space indices, a discretized state equation of the partial differential equation system is constructed and solved using the finite difference method. The derivative terms in the partial differential equations are approximated as differences between discrete variables, and the predicted data is output.

[0008] Furthermore, the implementation process of the extended Kalman filter includes: State vector: , in, N This represents the total number of cross-sections after the tunnel is discretized. k For time indexing; Observation vector: , in ，M =3 * number of active sensors M Value k All sensor observation data at any given time; Process noise covariance matrix: , Observation noise covariance matrix: , Initialize the state vector estimate: , Initialize the state estimation error covariance matrix: , Extended Kalman recursive filtering process, such as Figure 2 As shown: Steps 1-3 are used for time updates in the prediction steps: Step 1, State Observation: , in, f ( ) represents the discretized state equation, u k-1 The system control inputs include: tunnel entrance flow rate; Step 2: Calculate the state transition Jacobian matrix: , Step 3, Error Covariance Prediction: , Steps 4-8: Update measurement and update. Step 4: Calculate the observation residuals: ,in, h ( ) represents the observation function, which maps the state space to the observation space, from the full state vector. Extract the density, velocity, and flow rate values ​​corresponding to the sensor locations; Step 5: Calculate the observation Jacobian matrix: , Step 6: Calculate the Kalman gain: , Step 7: Update the state estimate: , Step 8: Update the error covariance estimate: , where I is the identity matrix; After completing steps 1-8, perform feedback iteration to update the state estimate. and error covariance Feedback is sent to the vehicle queuing propagation model as the initial condition for the next time step. Steps 1-8 are repeated, and the iteration terminates when the following condition is met: , among which, among which This is the threshold for state changes in the system error. The observation error threshold in the system error is a preset threshold based on sensor accuracy; the optimal state estimate is output. To the decision-making and response end.

[0009] Furthermore, the traffic guidance strategy specifically includes a visual warning strategy: Based on the obtained warning location, an active sensor is set up to flash lights upstream of the accident site; Warning location: This indicates the distance of the warning location from the upstream of the accident site; v Indicates the current vehicle speed. t r This indicates the driver's reaction.

[0010] Furthermore, the traffic guidance strategy specifically includes an information dissemination strategy: After analyzing the packaged data, traffic status information is released, including event type, location, scope of impact, and expected duration.

[0011] Furthermore, the traffic guidance strategy specifically includes a rescue dispatch strategy: Based on the event type and corresponding event type coefficient, dispatch rescue resources: L e = w 1. Event type coefficient + w 2. Impact range coefficient + w 3. Personnel entrapment coefficient; in, w 1, w2, w 3 is the weighting coefficient, which is determined based on historical data optimization; This invention provides a tunnel traffic linkage data analysis method based on event detection. By automatically detecting and responding to road condition events in the blind spots of tunnel sensors, it improves event response performance, buys valuable time for rescue, reduces the need for manual monitoring, and lowers equipment maintenance costs. Attached Figure Description

[0012] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and constitute a part of this invention, are not intended to limit the scope of the invention. In the drawings: Figure 1 This is a framework diagram of a tunnel traffic linkage data analysis method based on event detection, provided as an exemplary embodiment of the present invention.

[0013] Figure 2 The flowchart of the extended Kalman recursive filter algorithm in a tunnel traffic linkage data analysis method based on event detection, provided as an exemplary embodiment of the present invention, is shown below. Detailed Implementation

[0014] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention.

[0015] This invention provides a tunnel traffic linkage data analysis method based on event detection, which aims to solve the above-mentioned technical problems in the prior art.

[0016] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0017] Example 1: The configuration of multiple active sensor data sources in Example 1 is as follows: Visible light video (25fps, 1920×1080 resolution), infrared thermal imaging video (15fps, 640×512 resolution), environmental sensor data (CO concentration, temperature, visibility, sampling frequency 1Hz), and traffic flow data (microwave vehicle detector, sampling frequency 0.5Hz). Data preprocessing: For image data, a smoke removal algorithm based on Dark Channel Prior is used to preprocess visible light video to enhance image quality in smoky environments; non-uniformity correction and temperature calibration are performed on infrared images; for sensor data, median filtering is used to remove outliers, and time series data is smoothed using Kalman filtering; for data synchronization, all sensor data are synchronized via the NTP protocol with timestamp accuracy <10ms. Figure 1 A framework diagram of a tunnel traffic linkage data analysis method based on event detection is provided as an exemplary embodiment of the present invention, as shown below. Figure 1 As shown; Event detection and data assimilation process within the data assimilation framework: Initial event detection: Video analysis: Detection of vehicle stagnation using an improved YOLOv5 algorithm (confidence threshold 0.7, duration >10s); Environmental analysis: Detection of sudden changes in CO concentration (change rate >5ppm / s) combined with temperature anomalies (>60℃); Traffic flow analysis: Detection of sudden changes in speed (change rate >20km / h / 10s). There are two initialization states: one is to initialize the system after it is initialized and then to initialize the encounter event, and the other is to directly use the initialization of the encounter event as the initialization process of the data assimilation framework. System initialization of the data assimilation framework: Through reverse engineering using a vehicle queuing propagation model, the initial parameter settings are: free-flow velocity. v f =80km / h, blockage density p jam =180veh / km, relaxation time t =30s, preliminary accident location estimation: x initial = x sensor + w × Δt, where x sensor For sensor position, w The initial estimate of the shock wave propagation velocity is: , which is usually negative, indicating that the wave propagates upstream. The numerator is the flow rate (veh / h) and density (veh / km) of the free flow state upstream of the shock wave, and the denominator is the flow rate (veh / h) and density (veh / km) of the congested state downstream of the shock wave. Initialization of encounter events in the data assimilation framework: Real-time monitoring of shock waves wData from upstream and downstream sensors is used for initial event detection. When a sudden change in data at a certain cross-section is detected, it is marked as the time when the shock wave arrives at that cross-section. The shock wave velocity is calculated, and data from a period before and after the shock wave's arrival are taken to calculate the average flow rate and density upstream and downstream of the shock wave, respectively. These values ​​are then substituted into the formula to obtain: , Reverse deduction of event location: x evecnt = x down + w ×( t down - t 0 ),in, t 0 Since the time of the event is unknown, the time difference between upstream and downstream sensors is used as a substitute. t down - t 0 = t down - t up ,at this time, x down Using upstream sensors as a reference x up Since the shock wave is a negative value, a minus sign should be used when substituting it into the calculation; Extended Kalman Filter Implementation: Based on multi-source sensors, the tunnel is segmented to obtain state and observation vectors. Parameter settings: In the process noise covariance matrix, the diagonal elements are set to [0.1veh / km, 0.5km / h, 1veh / h]^2. The observation noise covariance is determined based on sensor accuracy. In the microwave vehicle detector, the diagonal elements are set to [1veh / km, 1km / h, 2veh / h]^2. The state transition Jacobian matrix F... k The calculations performed using numerical differentiation involve steps 1-8 as follows: Step 1, State Observation: , in, f ( ) represents the discretized state equation, u k-1 The system control inputs include: tunnel entrance flow rate; Step 2: Calculate the state transition Jacobian matrix: , Step 3, Error Covariance Prediction: , Step 4: Calculate the observation residuals: ,in, h ( ) represents the observation function, which maps the state space to the observation space, from the full state vector. Extract the density, velocity, and flow rate values ​​corresponding to the sensor locations; Step 5: Calculate the observation Jacobian matrix: , Step 6: Calculate the Kalman gain: , Step 7: Update the state estimate: , Step 8: Update the error covariance estimate: , where I is the identity matrix; set up It is 0.1. The value is 1.0, which satisfies the iteration stopping condition; Continue executing event confirmation: When the confidence level of the assimilated output data is >0.9, the event is confirmed to have occurred; Implement a coordinated lighting strategy, controlling LED lights to gradually increase to 100% brightness within 2 seconds via a dimming protocol; and coordinate with other lights to implement a visual warning strategy: activate intelligent flashing warnings 300-500m upstream, and disseminate event information via V2X communication: {Event Type: Fire, Location: K25+300, Affected Area: 1.2km, Estimated Duration: 45min}, and dispatch rescue efforts based on the event level. L e =0.92 (Level 1 event), automatically dispatching fire, medical, and traffic police resources.

[0018] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0019] Furthermore, in the embodiments of the present invention, the functional modules can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in the form of hardware plus software functional modules.

[0020] Those skilled in the art will understand that embodiments of the present invention can be provided as methods or systems. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects.

[0021] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0022] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

[0023] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the invention disclosed herein in the specification and examples. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the foregoing claims.

[0024] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for analyzing tunnel traffic linkage data based on event detection, characterized in that, Includes the following steps: Data from active sensors inside the tunnel is collected and analyzed to determine road conditions within the blind spots of the active sensors, including: The vehicle queue propagation model in the active sensor data input assimilation framework provides predicted road conditions, and then the predicted road conditions are estimated based on extended Kalman filtering and the corrected estimate is output to the vehicle queue propagation model. The process continues until the optimal estimation result is obtained and sent to the decision-making and response end, where traffic guidance strategies are implemented.

2. The method for analyzing tunnel traffic linkage data based on event detection according to claim 1, characterized in that, The data assimilation framework includes: Initialization: Based on the first received abnormal sensor data, a preliminary accident location estimate is obtained by reverse engineering through a vehicle queuing propagation model; After initialization, the data assimilation loop strategy is executed as follows: Prediction Step: The vehicle queue propagation model uses the optimal state estimate from the extended Kalman filter output of the previous time step to predict the traffic state of all sections in the tunnel at the current time step and outputs the prediction data. Update step: Based on the current observation data of all upper and lower section sensors segmented by active sensors, analyze the difference between the observation data and the predicted data, calculate the Kalman gain to match the fusion weights of the predicted data and the observation data, and obtain the optimal estimate based on the covariance of the predicted data and the observation data. When the difference between the predicted data calculated by the optimal estimate at the latest moment and the observed data is within the range of systematic error, the iteration stops and the optimal estimate is output.

3. The method for analyzing tunnel traffic linkage data based on event detection according to claim 2, characterized in that, The establishment of the vehicle queuing propagation model includes: State variable: Traffic density ρ ( x,t - Unit: veh / km, average speed v ( x,t - Unit: km / h, traffic flow q ( x, t - Unit: veh / h, Location x ; Basic parameters: Free flow velocity v f -Unit: km / h, Blockage density ρ jam -Unit: veh / km, velocity-density index parameter α Relaxation time parameters τ -Unit: seconds, traffic flow diffusion coefficient κ ; Constructing the modified Payne-Whitham equations of motion: ； Continuity equation: ； In the modified Payne-Whitham model equations of motion: The left side of the equal sign controls the mass derivative, which describes the rate of change of velocity. The first term on the right side of the equation is the relaxation term: ， The description refers to the vehicle's tendency to adjust its speed to a desired equilibrium speed. v e ( ρ The equilibrium velocity-density relationship describes the density under steady-state flow conditions as follows: ρ The corresponding equilibrium speed at that time; The second term on the right side of the equation is the pressure term, which describes the vehicle's response to changes in density ahead. The modified Payne-Whitham model's equations of motion and continuity equations form a closed set of partial differential equations. Given initial and boundary conditions, solving these equations yields the traffic state at every point in every cross-section of the tunnel at any future time, including density and speed, i.e., the predicted data.

4. The tunnel traffic linkage data analysis method based on event detection according to claim 3, characterized in that, Solving the vehicle queuing propagation model includes: In the vehicle queuing propagation model, the tunnel is divided into multiple cross-sections in space using active sensors as nodes. Each cross-section has a length of Δx, and the continuous time is divided into multiple time steps Δt. By labeling the state variables and fundamental parameters with time or space indices, a discretized state equation of the partial differential equation system is constructed and solved using the finite difference method. The derivative terms in the partial differential equations are approximated as differences between discrete variables, and the predicted data is output.

5. The tunnel traffic linkage data analysis method based on event detection according to claim 4, characterized in that, The implementation process of the extended Kalman filter includes: State vector: ， in, N This represents the total number of cross-sections after the tunnel is discretized. k For time indexing; Observation vector: ， in ，M =3 * number of active sensors M Value k All sensor observation data at any given time; Process noise covariance matrix: , Observation noise covariance matrix: , Initialize the state vector estimate: , Initialize the state estimation error covariance matrix: , Extended Kalman recursive filtering process: Steps 1-3 are used for time updates in the prediction steps: Step 1, State Observation: ， in, f ( ) represents the discretized state equation, u k-1 The system control inputs include: tunnel entrance flow rate; Step 2: Calculate the state transition Jacobian matrix: , Step 3, Error Covariance Prediction: , Steps 4-8: Update measurement and update. Step 4: Calculate the observation residuals: ,in, h ( ) represents the observation function, which maps the state space to the observation space, from the full state vector. Extract the density, velocity, and flow rate values ​​corresponding to the sensor locations; Step 5: Calculate the observation Jacobian matrix: ， Step 6: Calculate the Kalman gain: , Step 7: Update the state estimate: , Step 8: Update the error covariance estimate: , where I is the identity matrix; After completing steps 1-8, perform feedback iteration to update the state estimate. and error covariance Feedback is sent to the vehicle queuing propagation model as the initial condition for the next time step. Steps 1-8 are repeated, and the iteration terminates when the following condition is met: , among which, among which This is the threshold for state changes in the system error. The observation error threshold in the system error is a preset threshold based on sensor accuracy; the optimal state estimate is output. To the decision-making and response end.

6. The tunnel traffic linkage data analysis method based on event detection according to claim 5, characterized in that, The traffic guidance strategy specifically includes a visual warning strategy: Based on the obtained warning location, an active sensor is set up to flash lights upstream of the accident site; Warning location: This indicates the distance of the warning location from the upstream of the accident site; v Indicates the current vehicle speed. t r This indicates the driver's reaction.

7. The method for analyzing tunnel traffic linkage data based on event detection according to claim 5, characterized in that, The traffic guidance strategy specifically includes an information dissemination strategy: After analyzing the packaged data, traffic status information is released, including event type, location, scope of impact, and expected duration.

8. The method for analyzing tunnel traffic linkage data based on event detection according to claim 5, characterized in that, The traffic guidance strategy specifically includes a rescue dispatch strategy: Based on the event type and corresponding event type coefficient, dispatch rescue resources: L e = w 1. Event type coefficient + w 2. Impact range coefficient + w 3. Personnel entrapment coefficient; in, w 1, w 2, w 3 is the weighting coefficient, which is determined by optimization based on historical data.

Images