Tunnel event and warning illumination linked multi-source data fusion monitoring system

Abstract

The invention relates to a multi-source data fusion monitoring system linked with tunnel events and warning illumination, and belongs to the technical field of intelligent traffic systems, and the system comprises a multi-source data sensing and preprocessing module which is used for obtaining original data inside and outside a tunnel, and carrying out the normalization processing of the obtained original data, so as to generate a structured data stream; the dynamic event feature comprehensive evaluation module is used for receiving the structured data stream and performing quantitative evaluation to generate an event severity score and an event influence range radius; the heterogeneous linkage strategy generation module is used for receiving the event severity score and the event influence range radius, and constructing a space-time warning field model to generate normalized warning intensity; and the gradient control instruction execution module is used for receiving the normalized warning intensity, converting the normalized warning intensity into an equipment control instruction, and issuing the equipment control instruction to the terminal controller.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation systems, specifically a multi-source data fusion monitoring system that links tunnel events with warning lighting. Background Technology

[0002] Highway tunnels, as special sections of the transportation network, are high-risk areas for traffic accidents due to their enclosed environment and drastic changes in lighting. Existing tunnel safety monitoring systems mostly activate pre-set, homogeneous warning plans after detecting abnormal events such as traffic accidents, congestion, or pedestrian intrusion. For example, regardless of the severity or scope of the accident, a uniform warning pattern is used to alert following vehicles. This one-size-fits-all strategy lacks detailed consideration of the specific circumstances of the event, potentially leading to excessive or insufficient warnings. Insufficient warnings fail to effectively alert drivers, while excessive warnings may trigger overreactions, such as emergency braking, thus inducing secondary accidents and reducing the overall traffic efficiency and safety of the tunnel. Therefore, there is an urgent need for an advanced technological solution that can dynamically generate and execute differentiated and tiered warning strategies based on the multi-dimensional characteristics of events.

[0003] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-source data fusion monitoring system that links tunnel events with warning lighting, in order to solve the problems mentioned in the background art.

[0005] The technical solution of the present invention includes: a multi-source data sensing and preprocessing module for acquiring raw data inside and outside the tunnel, and normalizing the acquired raw data to generate a structured data stream;

[0006] The dynamic event feature comprehensive evaluation module is used to receive structured data streams, perform quantitative evaluations, and generate event severity scores and event impact radius.

[0007] The heterogeneous linkage strategy generation module is used to receive the event severity score and the radius of the event's impact range, construct a spatiotemporal warning field model, and generate a normalized warning intensity.

[0008] The gradient control command execution module is used to receive the normalized warning intensity, convert the normalized warning intensity into equipment control commands, and send them to the terminal controller.

[0009] Preferably, the raw data includes the instantaneous position, speed and acceleration of the vehicle, traffic flow and lane occupancy at a specific cross-section, ambient brightness inside and outside the tunnel, and raw event features identified through video analysis.

[0010] Preferably, the process by which the dynamic event feature comprehensive evaluation module generates an event severity score includes:

[0011] The system obtains the number of entities directly associated with the event, the relative rate of change, the Boolean flag of high-risk factors, and the inherent risk coefficient of the event type. It then performs a weighted summation of the obtained data on the number of entities, the relative rate of change, the Boolean flag of high-risk factors, and the inherent risk coefficient of the event type to generate an event severity score.

[0012] Preferably, the process by which the dynamic event feature comprehensive evaluation module generates the radius of influence of an event includes:

[0013] The event's road surface area is calculated using image segmentation technology, and an area correction value is determined based on this area. The dynamic safety distance is determined based on the driver's average reaction time. The determined area correction value and the dynamic safety distance are summed to generate the radius of the event's impact range.

[0014] Preferably, the process by which the heterogeneous linkage strategy generation module generates the normalized warning intensity includes:

[0015] A spatial attenuation factor is generated based on the radius of the event's impact range; a time attenuation factor is generated based on the time attenuation constant; and a traffic density enhancement factor is generated based on the real-time lane occupancy rate. The event severity score is then multiplicatively corrected with the generated spatial attenuation factor, time attenuation factor, and traffic density enhancement factor to generate a normalized warning intensity.

[0016] Preferably, the process of generating the time decay constant includes:

[0017] Obtain the event severity score and the preset adjustment coefficient, and multiply the two to obtain the first product value; sum the first product value with the preset coefficient representing the shortest baseline duration of the warning information to generate the time decay constant.

[0018] Preferably, the process of generating the traffic density enhancement factor includes:

[0019] The system obtains real-time lane occupancy and a preset critical traffic density; it then compares and analyzes the obtained real-time lane occupancy with the critical traffic density, and generates a traffic density enhancement factor through logistic functions.

[0020] Preferably, the process by which the gradient control instruction execution module converts device control instructions includes:

[0021] By using preset mapping rules, the normalized warning intensity is converted into discrete device control commands;

[0022] Mapping rules include:

[0023] When the normalized warning intensity is greater than the first preset intensity threshold, a first control command is generated;

[0024] When the normalized warning intensity is less than or equal to the first preset intensity threshold and greater than the second preset threshold, a second control command is generated;

[0025] When the normalized warning intensity is less than or equal to the second preset intensity threshold and greater than the third preset threshold, a third control command is generated;

[0026] When the normalized warning intensity is less than or equal to the third preset intensity threshold, a fourth control command is generated.

[0027] This invention improves the multi-source data fusion monitoring system that links tunnel events with warning lighting, and compared with the prior art, it has the following improvements and advantages:

[0028] 1. This solution achieves comprehensive and in-depth acquisition of tunnel event information by setting up a multi-source data perception and preprocessing module. This solution integrates the instantaneous position, speed and acceleration of vehicles, traffic flow and lane occupancy at specific cross-sections, ambient brightness inside and outside the tunnel, and the original event features identified through video analysis. This fusion of multi-source data provides unprecedentedly rich and high-dimensional input for subsequent accurate assessment, ensuring the integrity and reliability of event perception, and is a technical prerequisite for achieving refined control.

[0029] 2. This solution achieves precise quantification of event threat levels through a dynamic event feature comprehensive evaluation module. By converting area dimensions to length dimensions, it solves the fundamental problem that heterogeneous physical quantities cannot be directly superimposed, ensuring that the warning range is set to cover all potentially risky vehicles without being excessively expanded, thus avoiding unnecessary traffic disruptions. The combination of these two models enables the system to fully characterize any abnormal event from both qualitative and quantitative dimensions, which is not available in existing technologies.

[0030] 3. The core advancement of the solution lies in the construction of a dynamic spatiotemporal warning field model through a heterogeneous linkage strategy generation module. This enables the heterogeneity and gradient of warning strategies, minimizing the interference of warning information on normal traffic flow while ensuring safety. It also avoids secondary accidents caused by excessive or insufficient warnings, thereby significantly improving the overall traffic efficiency and safety of the tunnel. Attached Figure Description

[0031] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0032] Figure 1 This is a flowchart of the system of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0034] Example 1:

[0035] Please see Figure 1 This invention provides a technical solution for a multi-source data fusion monitoring system that links tunnel events with warning lighting, including: a multi-source data sensing and preprocessing module for acquiring raw data inside and outside the tunnel, and normalizing the acquired raw data to generate a structured data stream;

[0036] The dynamic event feature comprehensive evaluation module is used to receive structured data streams, perform quantitative evaluations, and generate event severity scores and event impact radius.

[0037] The heterogeneous linkage strategy generation module is used to receive the event severity score and the radius of the event's impact range, construct a spatiotemporal warning field model, and generate a normalized warning intensity.

[0038] The gradient control command execution module is used to receive the normalized warning intensity, convert the normalized warning intensity into equipment control commands, and send them to the terminal controller.

[0039] This embodiment provides a multi-source data fusion monitoring system that links tunnel events with warning lighting. Through the coupling and linkage of four functional modules, it forms a complete technical closed loop from data acquisition, quantitative evaluation, strategy generation to command execution. The multi-source data perception and preprocessing module, as the system's data acquisition unit, is responsible for acquiring various raw information about the tunnel environment. The dynamic event feature comprehensive evaluation module, as the system's analysis unit, performs multi-dimensional quantitative evaluation of detected events. The heterogeneous linkage strategy generation module, as the system's decision-making unit, generates dynamic warning schemes in real time based on the evaluation results. The gradient control command execution module, as the system's execution unit, transforms the decision schemes into specific equipment control commands. This system architecture aims to replace the traditional preset, homogeneous warning mode. Through data-driven dynamic decision-making, it achieves refined and differentiated responses to abnormal tunnel events. The technical effect is to improve the accuracy of information transmission and the suitability for drivers to receive information while ensuring safety, thereby reducing the probability of secondary accidents and optimizing the overall traffic efficiency of the tunnel.

[0040] This solution achieves comprehensive and in-depth acquisition of tunnel event information by setting up a multi-source data perception and preprocessing module. Existing technologies mostly rely on single-type sensors, which have limited information dimensions and are prone to misjudgment or omission. This solution integrates the instantaneous position, speed and acceleration of vehicles, traffic flow and lane occupancy at specific cross-sections, ambient brightness inside and outside the tunnel, and the original event features identified through video analysis. This fusion of multi-source data provides unprecedentedly rich and high-dimensional input for subsequent accurate assessment, ensuring the integrity and reliability of event perception, and is a technical prerequisite for achieving refined control.

[0041] Example 2:

[0042] The raw data includes the instantaneous position, speed and acceleration of vehicles, traffic flow and lane occupancy at specific cross-sections, ambient brightness inside and outside the tunnel, and raw event features identified through video analysis.

[0043] The process by which the dynamic event feature comprehensive assessment module generates an event severity score includes:

[0044] Obtain the number of entities directly related to the event, the relative rate of change, the Boolean flag of high-risk factors, and the inherent risk coefficient of the event type; perform a weighted summation on the obtained number of entities, relative rate of change, Boolean flag of high-risk factors, and inherent risk coefficient of the event type to generate an event severity score;

[0045] In this embodiment, the multi-source data perception and preprocessing module continuously acquires vehicle dynamic parameters, cross-sectional traffic flow data, ambient brightness, and event features such as smoke, flames, or debris directly identified by video AI algorithms from devices such as integrated radar-visual machines, brightness sensors, and high-definition cameras deployed inside and outside the tunnel. After the collected heterogeneous data undergoes standardization processing such as timestamp alignment and coordinate system registration, it forms a structured data stream, providing input for the subsequent evaluation module.

[0046] The dynamic event feature comprehensive assessment module receives this structured data stream and initiates a quantitative assessment of the event severity. The core of this assessment process lies in applying a linear weighted model derived from multi-attribute decision theory. The technical objective is to integrate multiple heterogeneous factors influencing event severity into a single, comparable value, thereby establishing a clear benchmark strength for subsequent warning strategies. This module calculates the dimensionless event severity score using the following formula: ;

[0047] in, This represents the dimensionless severity score of the event. The number of entities directly associated with the event, which is a dimensionless pure number; : Weighting coefficient corresponding to the number of entities, dimensionless; : The weighting coefficient corresponding to the rate of change of relative velocity, dimensionless; : The weighting coefficient corresponding to the Boolean sign of high-risk factors, dimensionless; : The weighting coefficient of the inherent risk coefficient corresponding to the event type, dimensionless; The inherent risk coefficient is predetermined based on the event type and is dimensionless.

[0048] Furthermore, "entity" specifically refers to motor vehicles, and "direct connection" refers to physical contact occurring during an accident, or emergency braking taken to avoid an accident ahead, such as deceleration greater than [a certain value]. The number of vehicles that make emergency lane changes, or those that do so. Vehicle trajectories and status can be identified and counted through real-time analysis of radar-visual fusion sensing devices; The reference speed for the section of road where the accident occurred. The average vehicle speed measured after the event, and the ratio of the difference between the two speeds to the reference speed, constitute the dimensionless rate of change of relative speed, with its subscript... Characterization reference, Characterizing events;

[0049] Furthermore, reference speed It is not a fixed road speed limit, but rather a calculated speed based on historical data, representing the expected traffic speed or free-flow speed on the affected road segment during the same time period, such as a weekday morning rush hour, and under similar weather conditions, without the impact of the incident. This dynamic reference value makes the assessment of the rate of change of speed more accurate;

[0050] Furthermore, the average vehicle speed measured after the incident... This refers to the spatially and temporally weighted average of the speeds of all monitored vehicles within a specific time window downstream of the event point, such as 200 meters, within a specific length downstream of the event point, within a range of 5 to 25 seconds after the event is confirmed by the system. This ensures... It can reliably reflect the immediate impact of events on local traffic flow; A Boolean flag indicating whether there are high-risk factors such as pedestrians or vehicles carrying dangerous goods; 1 for yes, 0 for no, and dimensionless. The inherent risk coefficient is preset based on the event type and is dimensionless;

[0051] For example, the following reference values ​​can be set based on the potential severity of the event: for general obstacle events such as vehicle debris, The possible value is 0.3; for serious traffic violations such as vehicles driving in the wrong direction or pedestrians running across the road, A possible value is 0.7; for accidents involving fires, dense smoke, or vehicles carrying hazardous materials inside the tunnel, The possible value is 1.0; these values ​​can be adjusted based on the specific risk assessment of different tunnels. These are the weight coefficients for the corresponding factors, all dimensionless and normalized to 1; weight coefficients The initial value calibration can be performed using the analytic hierarchy process combined with an expert knowledge base in the field of traffic safety. After the system is actually deployed, it can utilize historical event data and corresponding handling results to perform offline iterative optimization of the weights through supervised learning or reinforcement learning algorithms to enhance its adaptability to specific tunnel traffic patterns. This method can accurately quantify the degree of danger of events and provide an objective data basis for generating warning strategies that match the risk level.

[0052] This solution achieves precise quantification of event threat levels through a dynamic event feature comprehensive evaluation module. Existing technologies typically respond to events based on preset categories, failing to differentiate between different severity levels of similar events, leading to a one-size-fits-all approach to alerts. This solution creatively constructs two core quantification models:

[0053] Severity score of the event The computational model, The practical significance of this formula lies in its ability to account for multiple heterogeneous factors influencing the hazard of an event, including the number of entities directly related to the event. Relative velocity change rate High-risk factors Boolean markers and the inherent risk coefficient of the event type Through a set of weight coefficients that can be optimized using the analytic hierarchy process and machine learning. It performs a linear combination; it is not derived from other formulas, but is a comprehensive evaluation function built on multi-attribute decision theory. Its purpose is to transform the complex on-site situation into a single, dimensionless severity score. This provides an objective and measurable benchmark for the strength of warning strategies;

[0054] Radius of event impact The computational model, The practical significance of this formula lies in its scientific definition of the physical impact area of ​​an event; it innovatively incorporates the static road surface area occupied by the event. By taking the square root, it can be converted into a characteristic length. Then, the dynamic safety distance is determined by the driver's average reaction time. Linear superposition is performed; the formula is based on the physical principles of traffic safety engineering rather than pure mathematical derivation. By converting the area dimension to the length dimension, it solves the problem that heterogeneous physical quantities cannot be directly superimposed, ensuring that the warning range can cover all potentially risky vehicles without being excessively expanded, thus avoiding unnecessary traffic disturbances.

[0055] The combination of these two models enables the system to fully characterize any abnormal event from both qualitative and quantitative dimensions, which is something that existing technologies do not possess.

[0056] Example 3:

[0057] The process by which the dynamic event feature comprehensive evaluation module generates the radius of influence of an event includes:

[0058] The event-occupied road area is calculated using image segmentation technology, and an area correction value is determined based on this area. The dynamic safety distance is determined based on the average driver reaction time. The determined area correction value and the dynamic safety distance are summed to generate the radius of the event's impact range.

[0059] The process of generating normalized warning intensity by the heterogeneous linkage strategy generation module includes:

[0060] A spatial attenuation factor is generated based on the radius of the event's impact range; a time attenuation factor is generated based on the time attenuation constant; a traffic density enhancement factor is generated based on the real-time lane occupancy rate; the event severity score is multiplicatively corrected with the generated spatial attenuation factor, time attenuation factor, and traffic density enhancement factor to generate a normalized warning intensity.

[0061] In this embodiment, the dynamic event feature comprehensive assessment module completes the severity assessment. After calculation, the physical impact range is further assessed. The underlying logic of this assessment is that the final impact range of an event consists of two superimposed parts: firstly, the characteristic length derived from the static physical area occupied by the accident vehicle or spilled material; and secondly, the dynamic buffer zone required to ensure the safe reaction of vehicles behind. To ensure dimensional consistency and the rationality of the physical meaning, this module calculates the radius of impact range using the following formula. : ;

[0062] in, The radius of the spatial extent affected by the event, measured in length, meters; The area of ​​the road surface occupied by the event is calculated using image segmentation technology, with the dimension being meters². The reference speed for the affected road section is expressed in meters per second. The core correction to the formula lies in adjusting the area... Taking the square root yields an equivalent radius with the dimension of length, thus scientifically characterizing the spatial features of static obstacles.

[0063] The physical meaning and dimensional analysis of each term on the right side of the formula are as follows:

[0064] First item Represents the range of static physical influence, among which Convert the area unit, meters², to the length unit, meters. This is a dimensionless area correction factor used to calibrate the conversion relationship from the area occupied by irregularities to the equivalent radius of influence. This factor is typically slightly greater than 1, used to characterize the actual impact of irregularly shaped obstacles that exceeds their equivalent circular area. In a typical highway tunnel scenario, The value of is usually between 1.1 and 1.5. In this embodiment, the preferred value is 1.2. This coefficient can be obtained by regression analysis fitting of massive traffic accident statistics.

[0065] Second item Represents the dynamic safety buffer distance, where It is an equivalent time constant representing the driver's average reaction time, measured in seconds; The reference speed is used; the product of the two is in the dimension of length, in meters. The value can be taken from recognized standards in traffic engineering or precisely calibrated through driving simulation experiments;

[0066] This formula sums two terms that are both in the dimension of length, ensuring that it is consistent with the radius on the left. The dimensions are completely consistent, giving the model a solid physical foundation.

[0067] Severity score calculated by the module and the radius of influence These two core parameters are passed together to the heterogeneous linkage strategy generation module; this module receives... and Subsequently, a spatiotemporal alert model is constructed, aiming to describe the alert intensity as a function that varies continuously in time and space; this model calculates the alert intensity at a distance from the event center using the following formula. After the incident Normalized alert intensity at any given time: ;

[0068] The theoretical basis of this formula stems from the diffusion and attenuation models in physics. Its technological motivation lies in creating an alert model that can simulate the smooth and continuous changes in information and impact over time and space in the real world, replacing the rigid, black-and-white, on-or-off logic of traditional alerts. Normalized dimensionless warning intensity at a distance l from the event center and at time t after the event occurs; The normalized dimensionless warning intensity; The distance from the warning device to the incident center is measured in meters. The time difference, in seconds, between the occurrence of the event; This refers to the dimensionless event severity score calculated above; The radius of the event's impact area, calculated above, is [radius value] in meters. The time decay constant for the warning message, in seconds; The lane occupancy rate or traffic density is obtained in real time and is dimensionless. The preset critical traffic density is dimensionless, and its subscript is... The critical value represents the level of service (Level III or IV) in traffic engineering, indicating a state where traffic flow begins to become unstable or congested. For example, for a two-lane tunnel designed for a speed of 80 km / h, the critical lane occupancy rate can be determined through historical data analysis. The standard is set at 35%; is the gain coefficient of the logistic function, which is dimensionless; It is an abbreviation for exponential function;

[0069] This coefficient determines the degree to which the traffic density enhancement factor changes with traffic density; the higher the value, the more drastic the change in real-time density. Approaching critical density Warning intensity The faster the growth, the more powerful the system becomes; this gives it the ability to provide early and rapid warnings when traffic congestion is about to occur. In practical applications, The value range is usually between 5 and 20. In this embodiment, it can be 10 to achieve a sensitive and smooth transition.

[0070] This model can generate a warning distribution centered on the accident point, with the intensity smoothly decreasing from near to far, and dynamically evolving over time and according to changes in traffic conditions, greatly improving the accuracy of information transmission.

[0071] The core advancement of this solution lies in constructing a dynamic spatiotemporal warning field model through a heterogeneous linkage strategy generation module, achieving heterogeneity and gradient in warning strategies. Existing warning solutions are homogeneous, meaning they activate a uniform warning mode within a preset range. This solution generates normalized warning intensity. From the formula The decision; the logical construction process of this formula demonstrates a high degree of innovation: it uses event severity scores. The base strength is adjusted multiplicatively using three independent factors derived from physical and statistical models; spatial attenuation factor. The influence intensity as a function of distance was simulated using a Gaussian function. The natural law of smooth decay; time decay factor An exponential function is used to describe the warning information over time. The natural decay process, and its decay rate depends on the severity. Positive correlation constant Control; Traffic density enhancement factor Then, using the logistic function, based on the real-time lane occupancy rate... With critical traffic density Based on the comparison results, the warning intensity is nonlinearly enhanced;

[0072] The normalized warning intensity generated by this model It is no longer a static, uniform command, but a continuous function that dynamically evolves with space, time, and traffic density; for example, for a serious multi-vehicle rear-end collision, the system calculates... Value and The values ​​are all relatively large, thus creating a high-intensity and wide-ranging warning field near the accident site; while for a minor spill incident, Value and When the value is small, the resulting warning field is weak and small in range, and it decays rapidly over time; this differentiated processing capability is what distinguishes this solution from existing technologies.

[0073] Example 4:

[0074] The process of generating the time decay constant includes:

[0075] Obtain the event severity score and the preset adjustment coefficient, and multiply the two to obtain the first product value; sum the first product value with the preset coefficient representing the shortest baseline duration of the warning information to generate the time decay constant;

[0076] The generation process of the traffic density enhancement factor includes:

[0077] The system obtains real-time lane occupancy and a preset critical traffic density; it then compares and analyzes the obtained real-time lane occupancy with the critical traffic density, and generates a traffic density enhancement factor through logistic functions.

[0078] In this embodiment, to ensure the dynamism and adaptability of the warning model, its key parameters are also designed to be dynamically calculated; the time decay constant... The underlying logic of this calculation is that the higher the severity of the event, the longer the warning message should last. The calculation follows the following linear relationship: ;

[0079] The technical motivation behind this relationship is to directly link the duration of the alert with the severity of the event, thereby enabling on-demand allocation of resources; among which, The unit of measurement is time, seconds; It is a dimensionless severity score; coefficient Represents the minimum baseline duration of a warning message after any event occurs, measured in time and seconds; coefficient It is an adjustment coefficient, also measured in seconds, which determines the rate at which the warning duration increases with the severity of the event; to ensure consistency of measurement, The unit of measurement for the item is time; in this embodiment, to ensure that any event receives a minimum alert duration, the minimum baseline duration is set. It can be set to 120 seconds; adjustment coefficient This is used to linearly map the severity score of an event to an additional alert duration, and can be set to 300 seconds; therefore, a severity score... The event, the time decay constant of its warning information Will be Seconds; coefficient and The value can be determined by statistical analysis of historical accident data, from the average effective handling time and the shortest response time of events of different severity levels;

[0080] Meanwhile, the traffic density enhancement factor is generated through the logistic function, the core of which is to compare the real-time lane occupancy rate. and the preset critical traffic density ; Based on historical traffic flow data and design capacity of a specific tunnel, the critical inflection point of its congestion state is determined through statistical analysis; when the real-time density... far below When the factor has a small impact on the warning intensity, it does not significantly affect the warning strength. Approaching and exceeding At this time, the factor will increase rapidly, thus significantly amplifying the warning intensity. This mechanism gives the system foresight, providing relatively mild warnings when traffic is sparse, while automatically increasing the warnings when traffic is heavy, using stronger stimulating signals to penetrate complex traffic environments and effectively prevent chain rear-end collisions caused by the braking of the vehicle in front.

[0081] Example 5:

[0082] The process of the gradient control instruction execution module converting device control instructions includes:

[0083] By using preset mapping rules, the normalized warning intensity is converted into discrete device control commands;

[0084] Mapping rules include:

[0085] When the normalized warning intensity is greater than the first preset intensity threshold, a first control command is generated;

[0086] When the normalized warning intensity is less than or equal to the first preset intensity threshold and greater than the second preset threshold, a second control command is generated;

[0087] When the normalized warning intensity is less than or equal to the second preset intensity threshold and greater than the third preset threshold, a third control command is generated;

[0088] When the normalized warning intensity is less than or equal to the third preset intensity threshold, a fourth control command is generated;

[0089] In this embodiment, the gradient control command execution module is responsible for translating abstract decision-making schemes into concrete actions in the physical world; this module receives continuous dimensionless warning intensity values ​​calculated in real time for each control unit. And through a set of preset mapping rules, this continuous value is converted into discrete device control commands;

[0090] This preset mapping rule is based on traffic safety engineering, ergonomics, and the performance specifications of warning devices, and defines... The correspondence between value ranges and equipment status; taking the segmented contour light controller in the tunnel as an example, the mapping rule is set as follows: the first preset intensity threshold is set to 0.8, the second preset intensity threshold is set to 0.5, and the third preset intensity threshold is set to 0.2; when the calculated When the module generates the first control command, a red flashing indicator; when When, a second control command is generated, namely, a yellow flashing signal; when When this occurs, a third control command is generated, namely, a solid yellow light; when When this happens, a fourth control command is generated, namely, to shut down;

[0091] This module sends these commands to each terminal controller via industrial Ethernet or power line carrier communication; because each device operates at different times and locations... The values ​​vary, and the final effect is that the warning system of the entire tunnel presents a refined control mode that gradually decreases in space from the center of the event to the periphery and dynamically evolves in time. This gradient presentation method is more in line with the driver's cognitive habits than the traditional homogeneous alarm, and can significantly reduce the overreaction caused by abrupt alarms, thereby ensuring the driving safety of the tunnel at the operational level.

[0092] The gradient control command execution module normalizes the intensity of continuous warnings. By using pre-defined mapping rules based on traffic safety standards, discrete equipment control commands are generated; this precisely translates complex dynamic strategies into specific actions of physical equipment, such as controlling different sections. The value is mapped to the outline light's red flashing, yellow flashing, or solid yellow light;

[0093] In summary, the beneficial effects of this solution lie in its abandonment of the passive, extensive, and homogeneous warning methods of existing technologies, and the establishment of a proactive, precise, and heterogeneous closed-loop control system. Technological advancements are reflected in: improved perception comprehensiveness through multi-source data fusion; enhanced assessment accuracy through multi-dimensional quantification models; increased strategy flexibility through spatiotemporal warning field models; and improved control precision through gradient execution. These advancements work together to minimize the interference of warning information on normal traffic flow while ensuring safety, avoiding secondary accidents caused by excessive or insufficient warnings, thereby significantly improving the overall traffic efficiency and safety of the tunnel.

[0094] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multi-source data fusion monitoring system linking tunnel incidents and warning lighting, characterized in that, include: The multi-source data sensing and preprocessing module is used to acquire raw data from inside and outside the tunnel, and to normalize the acquired raw data to generate a structured data stream. The dynamic event feature comprehensive evaluation module is used to receive structured data streams, perform quantitative evaluations, and generate event severity scores and event impact radius. The heterogeneous linkage strategy generation module is used to receive the event severity score and the radius of the event's impact range, construct a spatiotemporal warning field model, and generate a normalized warning intensity. The gradient control command execution module is used to receive the normalized warning intensity, convert the normalized warning intensity into equipment control commands, and send them to the terminal controller.

2. The multi-source data fusion monitoring system for linking tunnel events and warning lighting according to claim 1, characterized in that, The raw data includes the instantaneous position, speed, and acceleration of vehicles, traffic flow and lane occupancy at specific cross-sections, ambient brightness inside and outside the tunnel, and raw event features identified through video analysis.

3. The multi-source data fusion monitoring system for linking tunnel events and warning lighting according to claim 1, characterized in that, The process by which the dynamic event feature comprehensive evaluation module generates an event severity score includes: The system obtains the number of entities directly associated with the event, the relative rate of change, the Boolean flag of high-risk factors, and the inherent risk coefficient of the event type. It then performs a weighted summation of the obtained data on the number of entities, the relative rate of change, the Boolean flag of high-risk factors, and the inherent risk coefficient of the event type to generate an event severity score.

4. The multi-source data fusion monitoring system for linking tunnel events and warning lighting according to claim 1, characterized in that, The process by which the dynamic event feature comprehensive evaluation module generates the radius of influence of an event includes: The event's road surface area is calculated using image segmentation technology, and an area correction value is determined based on this area. The dynamic safety distance is determined based on the driver's average reaction time. The determined area correction value and the dynamic safety distance are summed to generate the radius of the event's impact range.

5. The multi-source data fusion monitoring system for linking tunnel events and warning lighting according to claim 1, characterized in that, The process by which the heterogeneous linkage strategy generation module generates the normalized warning intensity includes: A spatial attenuation factor is generated based on the radius of the event's impact range; a time attenuation factor is generated based on the time attenuation constant; and a traffic density enhancement factor is generated based on the real-time lane occupancy rate. The event severity score is then multiplicatively corrected with the generated spatial attenuation factor, time attenuation factor, and traffic density enhancement factor to generate a normalized warning intensity.

6. The multi-source data fusion monitoring system for linking tunnel events and warning lighting according to claim 5, characterized in that, The process of generating the time decay constant includes: Obtain the event severity score and the preset adjustment coefficient, and multiply the two to obtain the first product value; sum the first product value with the preset coefficient representing the shortest baseline duration of the warning information to generate the time decay constant.

7. The multi-source data fusion monitoring system for linking tunnel events and warning lighting according to claim 5, characterized in that, The process of generating the traffic density enhancement factor includes: The system obtains real-time lane occupancy and a preset critical traffic density; it then compares and analyzes the obtained real-time lane occupancy with the critical traffic density, and generates a traffic density enhancement factor through logistic functions.

8. The multi-source data fusion monitoring system for linking tunnel events and warning lighting according to claim 1, characterized in that, The process by which the gradient control command execution module converts device control commands includes: By using preset mapping rules, the normalized warning intensity is converted into discrete device control commands; Mapping rules include: When the normalized warning intensity is greater than the first preset intensity threshold, a first control command is generated; When the normalized warning intensity is less than or equal to the first preset intensity threshold and greater than the second preset threshold, a second control command is generated; When the normalized warning intensity is less than or equal to the second preset intensity threshold and greater than the third preset threshold, a third control command is generated; When the normalized warning intensity is less than or equal to the third preset intensity threshold, a fourth control command is generated.

Images